diff --git a/.Rprofile b/.Rprofile
new file mode 100644
index 0000000..81b960f
--- /dev/null
+++ b/.Rprofile
@@ -0,0 +1 @@
+source("renv/activate.R")
diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..bc509d0
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,51 @@
+# repo specific files
+_notrack/
+*.ai
+
+# quarto freeze
+_freeze/
+
+# History files
+.Rhistory
+.Rapp.history
+
+# Session Data files
+.RData
+.RDataTmp
+
+# User-specific files
+.Ruserdata
+
+# Example code in package build process
+*-Ex.R
+
+# Output files from R CMD check
+/*.Rcheck/
+
+# RStudio files
+.Rproj.user/
+
+# produced vignettes
+vignettes/*.html
+vignettes/*.pdf
+
+# OAuth2 token, see https://github.com/hadley/httr/releases/tag/v0.3
+.httr-oauth
+
+# knitr and R markdown default cache directories
+*_cache/
+/cache/
+
+# Temporary files created by R markdown
+*.utf8.md
+*.knit.md
+
+# R Environment Variables
+.Renviron
+
+# translation temp files
+po/*~
+
+# RStudio Connect folder
+rsconnect/
+/.quarto/
diff --git a/LICENSE b/LICENSE
new file mode 100644
index 0000000..a46b531
--- /dev/null
+++ b/LICENSE
@@ -0,0 +1,661 @@
+ GNU AFFERO GENERAL PUBLIC LICENSE
+ Version 3, 19 November 2007
+
+ Copyright (C) 2007 Free Software Foundation, Inc.
+ Everyone is permitted to copy and distribute verbatim copies
+ of this license document, but changing it is not allowed.
+
+ Preamble
+
+ The GNU Affero General Public License is a free, copyleft license for
+software and other kinds of works, specifically designed to ensure
+cooperation with the community in the case of network server software.
+
+ The licenses for most software and other practical works are designed
+to take away your freedom to share and change the works. By contrast,
+our General Public Licenses are intended to guarantee your freedom to
+share and change all versions of a program--to make sure it remains free
+software for all its users.
+
+ When we speak of free software, we are referring to freedom, not
+price. Our General Public Licenses are designed to make sure that you
+have the freedom to distribute copies of free software (and charge for
+them if you wish), that you receive source code or can get it if you
+want it, that you can change the software or use pieces of it in new
+free programs, and that you know you can do these things.
+
+ Developers that use our General Public Licenses protect your rights
+with two steps: (1) assert copyright on the software, and (2) offer
+you this License which gives you legal permission to copy, distribute
+and/or modify the software.
+
+ A secondary benefit of defending all users' freedom is that
+improvements made in alternate versions of the program, if they
+receive widespread use, become available for other developers to
+incorporate. Many developers of free software are heartened and
+encouraged by the resulting cooperation. However, in the case of
+software used on network servers, this result may fail to come about.
+The GNU General Public License permits making a modified version and
+letting the public access it on a server without ever releasing its
+source code to the public.
+
+ The GNU Affero General Public License is designed specifically to
+ensure that, in such cases, the modified source code becomes available
+to the community. It requires the operator of a network server to
+provide the source code of the modified version running there to the
+users of that server. Therefore, public use of a modified version, on
+a publicly accessible server, gives the public access to the source
+code of the modified version.
+
+ An older license, called the Affero General Public License and
+published by Affero, was designed to accomplish similar goals. This is
+a different license, not a version of the Affero GPL, but Affero has
+released a new version of the Affero GPL which permits relicensing under
+this license.
+
+ The precise terms and conditions for copying, distribution and
+modification follow.
+
+ TERMS AND CONDITIONS
+
+ 0. Definitions.
+
+ "This License" refers to version 3 of the GNU Affero General Public License.
+
+ "Copyright" also means copyright-like laws that apply to other kinds of
+works, such as semiconductor masks.
+
+ "The Program" refers to any copyrightable work licensed under this
+License. Each licensee is addressed as "you". "Licensees" and
+"recipients" may be individuals or organizations.
+
+ To "modify" a work means to copy from or adapt all or part of the work
+in a fashion requiring copyright permission, other than the making of an
+exact copy. The resulting work is called a "modified version" of the
+earlier work or a work "based on" the earlier work.
+
+ A "covered work" means either the unmodified Program or a work based
+on the Program.
+
+ To "propagate" a work means to do anything with it that, without
+permission, would make you directly or secondarily liable for
+infringement under applicable copyright law, except executing it on a
+computer or modifying a private copy. Propagation includes copying,
+distribution (with or without modification), making available to the
+public, and in some countries other activities as well.
+
+ To "convey" a work means any kind of propagation that enables other
+parties to make or receive copies. Mere interaction with a user through
+a computer network, with no transfer of a copy, is not conveying.
+
+ An interactive user interface displays "Appropriate Legal Notices"
+to the extent that it includes a convenient and prominently visible
+feature that (1) displays an appropriate copyright notice, and (2)
+tells the user that there is no warranty for the work (except to the
+extent that warranties are provided), that licensees may convey the
+work under this License, and how to view a copy of this License. If
+the interface presents a list of user commands or options, such as a
+menu, a prominent item in the list meets this criterion.
+
+ 1. Source Code.
+
+ The "source code" for a work means the preferred form of the work
+for making modifications to it. "Object code" means any non-source
+form of a work.
+
+ A "Standard Interface" means an interface that either is an official
+standard defined by a recognized standards body, or, in the case of
+interfaces specified for a particular programming language, one that
+is widely used among developers working in that language.
+
+ The "System Libraries" of an executable work include anything, other
+than the work as a whole, that (a) is included in the normal form of
+packaging a Major Component, but which is not part of that Major
+Component, and (b) serves only to enable use of the work with that
+Major Component, or to implement a Standard Interface for which an
+implementation is available to the public in source code form. A
+"Major Component", in this context, means a major essential component
+(kernel, window system, and so on) of the specific operating system
+(if any) on which the executable work runs, or a compiler used to
+produce the work, or an object code interpreter used to run it.
+
+ The "Corresponding Source" for a work in object code form means all
+the source code needed to generate, install, and (for an executable
+work) run the object code and to modify the work, including scripts to
+control those activities. However, it does not include the work's
+System Libraries, or general-purpose tools or generally available free
+programs which are used unmodified in performing those activities but
+which are not part of the work. For example, Corresponding Source
+includes interface definition files associated with source files for
+the work, and the source code for shared libraries and dynamically
+linked subprograms that the work is specifically designed to require,
+such as by intimate data communication or control flow between those
+subprograms and other parts of the work.
+
+ The Corresponding Source need not include anything that users
+can regenerate automatically from other parts of the Corresponding
+Source.
+
+ The Corresponding Source for a work in source code form is that
+same work.
+
+ 2. Basic Permissions.
+
+ All rights granted under this License are granted for the term of
+copyright on the Program, and are irrevocable provided the stated
+conditions are met. This License explicitly affirms your unlimited
+permission to run the unmodified Program. The output from running a
+covered work is covered by this License only if the output, given its
+content, constitutes a covered work. This License acknowledges your
+rights of fair use or other equivalent, as provided by copyright law.
+
+ You may make, run and propagate covered works that you do not
+convey, without conditions so long as your license otherwise remains
+in force. You may convey covered works to others for the sole purpose
+of having them make modifications exclusively for you, or provide you
+with facilities for running those works, provided that you comply with
+the terms of this License in conveying all material for which you do
+not control copyright. Those thus making or running the covered works
+for you must do so exclusively on your behalf, under your direction
+and control, on terms that prohibit them from making any copies of
+your copyrighted material outside their relationship with you.
+
+ Conveying under any other circumstances is permitted solely under
+the conditions stated below. Sublicensing is not allowed; section 10
+makes it unnecessary.
+
+ 3. Protecting Users' Legal Rights From Anti-Circumvention Law.
+
+ No covered work shall be deemed part of an effective technological
+measure under any applicable law fulfilling obligations under article
+11 of the WIPO copyright treaty adopted on 20 December 1996, or
+similar laws prohibiting or restricting circumvention of such
+measures.
+
+ When you convey a covered work, you waive any legal power to forbid
+circumvention of technological measures to the extent such circumvention
+is effected by exercising rights under this License with respect to
+the covered work, and you disclaim any intention to limit operation or
+modification of the work as a means of enforcing, against the work's
+users, your or third parties' legal rights to forbid circumvention of
+technological measures.
+
+ 4. Conveying Verbatim Copies.
+
+ You may convey verbatim copies of the Program's source code as you
+receive it, in any medium, provided that you conspicuously and
+appropriately publish on each copy an appropriate copyright notice;
+keep intact all notices stating that this License and any
+non-permissive terms added in accord with section 7 apply to the code;
+keep intact all notices of the absence of any warranty; and give all
+recipients a copy of this License along with the Program.
+
+ You may charge any price or no price for each copy that you convey,
+and you may offer support or warranty protection for a fee.
+
+ 5. Conveying Modified Source Versions.
+
+ You may convey a work based on the Program, or the modifications to
+produce it from the Program, in the form of source code under the
+terms of section 4, provided that you also meet all of these conditions:
+
+ a) The work must carry prominent notices stating that you modified
+ it, and giving a relevant date.
+
+ b) The work must carry prominent notices stating that it is
+ released under this License and any conditions added under section
+ 7. This requirement modifies the requirement in section 4 to
+ "keep intact all notices".
+
+ c) You must license the entire work, as a whole, under this
+ License to anyone who comes into possession of a copy. This
+ License will therefore apply, along with any applicable section 7
+ additional terms, to the whole of the work, and all its parts,
+ regardless of how they are packaged. This License gives no
+ permission to license the work in any other way, but it does not
+ invalidate such permission if you have separately received it.
+
+ d) If the work has interactive user interfaces, each must display
+ Appropriate Legal Notices; however, if the Program has interactive
+ interfaces that do not display Appropriate Legal Notices, your
+ work need not make them do so.
+
+ A compilation of a covered work with other separate and independent
+works, which are not by their nature extensions of the covered work,
+and which are not combined with it such as to form a larger program,
+in or on a volume of a storage or distribution medium, is called an
+"aggregate" if the compilation and its resulting copyright are not
+used to limit the access or legal rights of the compilation's users
+beyond what the individual works permit. Inclusion of a covered work
+in an aggregate does not cause this License to apply to the other
+parts of the aggregate.
+
+ 6. Conveying Non-Source Forms.
+
+ You may convey a covered work in object code form under the terms
+of sections 4 and 5, provided that you also convey the
+machine-readable Corresponding Source under the terms of this License,
+in one of these ways:
+
+ a) Convey the object code in, or embodied in, a physical product
+ (including a physical distribution medium), accompanied by the
+ Corresponding Source fixed on a durable physical medium
+ customarily used for software interchange.
+
+ b) Convey the object code in, or embodied in, a physical product
+ (including a physical distribution medium), accompanied by a
+ written offer, valid for at least three years and valid for as
+ long as you offer spare parts or customer support for that product
+ model, to give anyone who possesses the object code either (1) a
+ copy of the Corresponding Source for all the software in the
+ product that is covered by this License, on a durable physical
+ medium customarily used for software interchange, for a price no
+ more than your reasonable cost of physically performing this
+ conveying of source, or (2) access to copy the
+ Corresponding Source from a network server at no charge.
+
+ c) Convey individual copies of the object code with a copy of the
+ written offer to provide the Corresponding Source. This
+ alternative is allowed only occasionally and noncommercially, and
+ only if you received the object code with such an offer, in accord
+ with subsection 6b.
+
+ d) Convey the object code by offering access from a designated
+ place (gratis or for a charge), and offer equivalent access to the
+ Corresponding Source in the same way through the same place at no
+ further charge. You need not require recipients to copy the
+ Corresponding Source along with the object code. If the place to
+ copy the object code is a network server, the Corresponding Source
+ may be on a different server (operated by you or a third party)
+ that supports equivalent copying facilities, provided you maintain
+ clear directions next to the object code saying where to find the
+ Corresponding Source. Regardless of what server hosts the
+ Corresponding Source, you remain obligated to ensure that it is
+ available for as long as needed to satisfy these requirements.
+
+ e) Convey the object code using peer-to-peer transmission, provided
+ you inform other peers where the object code and Corresponding
+ Source of the work are being offered to the general public at no
+ charge under subsection 6d.
+
+ A separable portion of the object code, whose source code is excluded
+from the Corresponding Source as a System Library, need not be
+included in conveying the object code work.
+
+ A "User Product" is either (1) a "consumer product", which means any
+tangible personal property which is normally used for personal, family,
+or household purposes, or (2) anything designed or sold for incorporation
+into a dwelling. In determining whether a product is a consumer product,
+doubtful cases shall be resolved in favor of coverage. For a particular
+product received by a particular user, "normally used" refers to a
+typical or common use of that class of product, regardless of the status
+of the particular user or of the way in which the particular user
+actually uses, or expects or is expected to use, the product. A product
+is a consumer product regardless of whether the product has substantial
+commercial, industrial or non-consumer uses, unless such uses represent
+the only significant mode of use of the product.
+
+ "Installation Information" for a User Product means any methods,
+procedures, authorization keys, or other information required to install
+and execute modified versions of a covered work in that User Product from
+a modified version of its Corresponding Source. The information must
+suffice to ensure that the continued functioning of the modified object
+code is in no case prevented or interfered with solely because
+modification has been made.
+
+ If you convey an object code work under this section in, or with, or
+specifically for use in, a User Product, and the conveying occurs as
+part of a transaction in which the right of possession and use of the
+User Product is transferred to the recipient in perpetuity or for a
+fixed term (regardless of how the transaction is characterized), the
+Corresponding Source conveyed under this section must be accompanied
+by the Installation Information. But this requirement does not apply
+if neither you nor any third party retains the ability to install
+modified object code on the User Product (for example, the work has
+been installed in ROM).
+
+ The requirement to provide Installation Information does not include a
+requirement to continue to provide support service, warranty, or updates
+for a work that has been modified or installed by the recipient, or for
+the User Product in which it has been modified or installed. Access to a
+network may be denied when the modification itself materially and
+adversely affects the operation of the network or violates the rules and
+protocols for communication across the network.
+
+ Corresponding Source conveyed, and Installation Information provided,
+in accord with this section must be in a format that is publicly
+documented (and with an implementation available to the public in
+source code form), and must require no special password or key for
+unpacking, reading or copying.
+
+ 7. Additional Terms.
+
+ "Additional permissions" are terms that supplement the terms of this
+License by making exceptions from one or more of its conditions.
+Additional permissions that are applicable to the entire Program shall
+be treated as though they were included in this License, to the extent
+that they are valid under applicable law. If additional permissions
+apply only to part of the Program, that part may be used separately
+under those permissions, but the entire Program remains governed by
+this License without regard to the additional permissions.
+
+ When you convey a copy of a covered work, you may at your option
+remove any additional permissions from that copy, or from any part of
+it. (Additional permissions may be written to require their own
+removal in certain cases when you modify the work.) You may place
+additional permissions on material, added by you to a covered work,
+for which you have or can give appropriate copyright permission.
+
+ Notwithstanding any other provision of this License, for material you
+add to a covered work, you may (if authorized by the copyright holders of
+that material) supplement the terms of this License with terms:
+
+ a) Disclaiming warranty or limiting liability differently from the
+ terms of sections 15 and 16 of this License; or
+
+ b) Requiring preservation of specified reasonable legal notices or
+ author attributions in that material or in the Appropriate Legal
+ Notices displayed by works containing it; or
+
+ c) Prohibiting misrepresentation of the origin of that material, or
+ requiring that modified versions of such material be marked in
+ reasonable ways as different from the original version; or
+
+ d) Limiting the use for publicity purposes of names of licensors or
+ authors of the material; or
+
+ e) Declining to grant rights under trademark law for use of some
+ trade names, trademarks, or service marks; or
+
+ f) Requiring indemnification of licensors and authors of that
+ material by anyone who conveys the material (or modified versions of
+ it) with contractual assumptions of liability to the recipient, for
+ any liability that these contractual assumptions directly impose on
+ those licensors and authors.
+
+ All other non-permissive additional terms are considered "further
+restrictions" within the meaning of section 10. If the Program as you
+received it, or any part of it, contains a notice stating that it is
+governed by this License along with a term that is a further
+restriction, you may remove that term. If a license document contains
+a further restriction but permits relicensing or conveying under this
+License, you may add to a covered work material governed by the terms
+of that license document, provided that the further restriction does
+not survive such relicensing or conveying.
+
+ If you add terms to a covered work in accord with this section, you
+must place, in the relevant source files, a statement of the
+additional terms that apply to those files, or a notice indicating
+where to find the applicable terms.
+
+ Additional terms, permissive or non-permissive, may be stated in the
+form of a separately written license, or stated as exceptions;
+the above requirements apply either way.
+
+ 8. Termination.
+
+ You may not propagate or modify a covered work except as expressly
+provided under this License. Any attempt otherwise to propagate or
+modify it is void, and will automatically terminate your rights under
+this License (including any patent licenses granted under the third
+paragraph of section 11).
+
+ However, if you cease all violation of this License, then your
+license from a particular copyright holder is reinstated (a)
+provisionally, unless and until the copyright holder explicitly and
+finally terminates your license, and (b) permanently, if the copyright
+holder fails to notify you of the violation by some reasonable means
+prior to 60 days after the cessation.
+
+ Moreover, your license from a particular copyright holder is
+reinstated permanently if the copyright holder notifies you of the
+violation by some reasonable means, this is the first time you have
+received notice of violation of this License (for any work) from that
+copyright holder, and you cure the violation prior to 30 days after
+your receipt of the notice.
+
+ Termination of your rights under this section does not terminate the
+licenses of parties who have received copies or rights from you under
+this License. If your rights have been terminated and not permanently
+reinstated, you do not qualify to receive new licenses for the same
+material under section 10.
+
+ 9. Acceptance Not Required for Having Copies.
+
+ You are not required to accept this License in order to receive or
+run a copy of the Program. Ancillary propagation of a covered work
+occurring solely as a consequence of using peer-to-peer transmission
+to receive a copy likewise does not require acceptance. However,
+nothing other than this License grants you permission to propagate or
+modify any covered work. These actions infringe copyright if you do
+not accept this License. Therefore, by modifying or propagating a
+covered work, you indicate your acceptance of this License to do so.
+
+ 10. Automatic Licensing of Downstream Recipients.
+
+ Each time you convey a covered work, the recipient automatically
+receives a license from the original licensors, to run, modify and
+propagate that work, subject to this License. You are not responsible
+for enforcing compliance by third parties with this License.
+
+ An "entity transaction" is a transaction transferring control of an
+organization, or substantially all assets of one, or subdividing an
+organization, or merging organizations. If propagation of a covered
+work results from an entity transaction, each party to that
+transaction who receives a copy of the work also receives whatever
+licenses to the work the party's predecessor in interest had or could
+give under the previous paragraph, plus a right to possession of the
+Corresponding Source of the work from the predecessor in interest, if
+the predecessor has it or can get it with reasonable efforts.
+
+ You may not impose any further restrictions on the exercise of the
+rights granted or affirmed under this License. For example, you may
+not impose a license fee, royalty, or other charge for exercise of
+rights granted under this License, and you may not initiate litigation
+(including a cross-claim or counterclaim in a lawsuit) alleging that
+any patent claim is infringed by making, using, selling, offering for
+sale, or importing the Program or any portion of it.
+
+ 11. Patents.
+
+ A "contributor" is a copyright holder who authorizes use under this
+License of the Program or a work on which the Program is based. The
+work thus licensed is called the contributor's "contributor version".
+
+ A contributor's "essential patent claims" are all patent claims
+owned or controlled by the contributor, whether already acquired or
+hereafter acquired, that would be infringed by some manner, permitted
+by this License, of making, using, or selling its contributor version,
+but do not include claims that would be infringed only as a
+consequence of further modification of the contributor version. For
+purposes of this definition, "control" includes the right to grant
+patent sublicenses in a manner consistent with the requirements of
+this License.
+
+ Each contributor grants you a non-exclusive, worldwide, royalty-free
+patent license under the contributor's essential patent claims, to
+make, use, sell, offer for sale, import and otherwise run, modify and
+propagate the contents of its contributor version.
+
+ In the following three paragraphs, a "patent license" is any express
+agreement or commitment, however denominated, not to enforce a patent
+(such as an express permission to practice a patent or covenant not to
+sue for patent infringement). To "grant" such a patent license to a
+party means to make such an agreement or commitment not to enforce a
+patent against the party.
+
+ If you convey a covered work, knowingly relying on a patent license,
+and the Corresponding Source of the work is not available for anyone
+to copy, free of charge and under the terms of this License, through a
+publicly available network server or other readily accessible means,
+then you must either (1) cause the Corresponding Source to be so
+available, or (2) arrange to deprive yourself of the benefit of the
+patent license for this particular work, or (3) arrange, in a manner
+consistent with the requirements of this License, to extend the patent
+license to downstream recipients. "Knowingly relying" means you have
+actual knowledge that, but for the patent license, your conveying the
+covered work in a country, or your recipient's use of the covered work
+in a country, would infringe one or more identifiable patents in that
+country that you have reason to believe are valid.
+
+ If, pursuant to or in connection with a single transaction or
+arrangement, you convey, or propagate by procuring conveyance of, a
+covered work, and grant a patent license to some of the parties
+receiving the covered work authorizing them to use, propagate, modify
+or convey a specific copy of the covered work, then the patent license
+you grant is automatically extended to all recipients of the covered
+work and works based on it.
+
+ A patent license is "discriminatory" if it does not include within
+the scope of its coverage, prohibits the exercise of, or is
+conditioned on the non-exercise of one or more of the rights that are
+specifically granted under this License. You may not convey a covered
+work if you are a party to an arrangement with a third party that is
+in the business of distributing software, under which you make payment
+to the third party based on the extent of your activity of conveying
+the work, and under which the third party grants, to any of the
+parties who would receive the covered work from you, a discriminatory
+patent license (a) in connection with copies of the covered work
+conveyed by you (or copies made from those copies), or (b) primarily
+for and in connection with specific products or compilations that
+contain the covered work, unless you entered into that arrangement,
+or that patent license was granted, prior to 28 March 2007.
+
+ Nothing in this License shall be construed as excluding or limiting
+any implied license or other defenses to infringement that may
+otherwise be available to you under applicable patent law.
+
+ 12. No Surrender of Others' Freedom.
+
+ If conditions are imposed on you (whether by court order, agreement or
+otherwise) that contradict the conditions of this License, they do not
+excuse you from the conditions of this License. If you cannot convey a
+covered work so as to satisfy simultaneously your obligations under this
+License and any other pertinent obligations, then as a consequence you may
+not convey it at all. For example, if you agree to terms that obligate you
+to collect a royalty for further conveying from those to whom you convey
+the Program, the only way you could satisfy both those terms and this
+License would be to refrain entirely from conveying the Program.
+
+ 13. Remote Network Interaction; Use with the GNU General Public License.
+
+ Notwithstanding any other provision of this License, if you modify the
+Program, your modified version must prominently offer all users
+interacting with it remotely through a computer network (if your version
+supports such interaction) an opportunity to receive the Corresponding
+Source of your version by providing access to the Corresponding Source
+from a network server at no charge, through some standard or customary
+means of facilitating copying of software. This Corresponding Source
+shall include the Corresponding Source for any work covered by version 3
+of the GNU General Public License that is incorporated pursuant to the
+following paragraph.
+
+ Notwithstanding any other provision of this License, you have
+permission to link or combine any covered work with a work licensed
+under version 3 of the GNU General Public License into a single
+combined work, and to convey the resulting work. The terms of this
+License will continue to apply to the part which is the covered work,
+but the work with which it is combined will remain governed by version
+3 of the GNU General Public License.
+
+ 14. Revised Versions of this License.
+
+ The Free Software Foundation may publish revised and/or new versions of
+the GNU Affero General Public License from time to time. Such new versions
+will be similar in spirit to the present version, but may differ in detail to
+address new problems or concerns.
+
+ Each version is given a distinguishing version number. If the
+Program specifies that a certain numbered version of the GNU Affero General
+Public License "or any later version" applies to it, you have the
+option of following the terms and conditions either of that numbered
+version or of any later version published by the Free Software
+Foundation. If the Program does not specify a version number of the
+GNU Affero General Public License, you may choose any version ever published
+by the Free Software Foundation.
+
+ If the Program specifies that a proxy can decide which future
+versions of the GNU Affero General Public License can be used, that proxy's
+public statement of acceptance of a version permanently authorizes you
+to choose that version for the Program.
+
+ Later license versions may give you additional or different
+permissions. However, no additional obligations are imposed on any
+author or copyright holder as a result of your choosing to follow a
+later version.
+
+ 15. Disclaimer of Warranty.
+
+ THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY
+APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT
+HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY
+OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO,
+THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM
+IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF
+ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
+
+ 16. Limitation of Liability.
+
+ IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
+WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS
+THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY
+GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE
+USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF
+DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD
+PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS),
+EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF
+SUCH DAMAGES.
+
+ 17. Interpretation of Sections 15 and 16.
+
+ If the disclaimer of warranty and limitation of liability provided
+above cannot be given local legal effect according to their terms,
+reviewing courts shall apply local law that most closely approximates
+an absolute waiver of all civil liability in connection with the
+Program, unless a warranty or assumption of liability accompanies a
+copy of the Program in return for a fee.
+
+ END OF TERMS AND CONDITIONS
+
+ How to Apply These Terms to Your New Programs
+
+ If you develop a new program, and you want it to be of the greatest
+possible use to the public, the best way to achieve this is to make it
+free software which everyone can redistribute and change under these terms.
+
+ To do so, attach the following notices to the program. It is safest
+to attach them to the start of each source file to most effectively
+state the exclusion of warranty; and each file should have at least
+the "copyright" line and a pointer to where the full notice is found.
+
+
+ Copyright (C) 2023 Shane Hogle
+
+ This program is free software: you can redistribute it and/or modify
+ it under the terms of the GNU Affero General Public License as published
+ by the Free Software Foundation, either version 3 of the License, or
+ (at your option) any later version.
+
+ This program is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU Affero General Public License for more details.
+
+ You should have received a copy of the GNU Affero General Public License
+ along with this program. If not, see .
+
+Also add information on how to contact you by electronic and paper mail.
+
+ If your software can interact with users remotely through a computer
+network, you should also make sure that it provides a way for users to
+get its source. For example, if your program is a web application, its
+interface could display a "Source" link that leads users to an archive
+of the code. There are many ways you could offer source, and different
+solutions will be better for different programs; see section 13 for the
+specific requirements.
+
+ You should also get your employer (if you work as a programmer) or school,
+if any, to sign a "copyright disclaimer" for the program, if necessary.
+For more information on this, and how to apply and follow the GNU AGPL, see
+.
diff --git a/R/illumina_v3v4/01_rpkm2tab.qmd b/R/illumina_v3v4/01_rpkm2tab.qmd
new file mode 100644
index 0000000..9e13391
--- /dev/null
+++ b/R/illumina_v3v4/01_rpkm2tab.qmd
@@ -0,0 +1,143 @@
+---
+title: "Process and format amplicon count data"
+author: "Shane Hogle"
+date: today
+abstract: "This notebook formats 16S rRNA amplicon counts that have been mapped to HAMBI 16S genes."
+---
+
+# Setup
+
+Libraries and global variables
+
+```{r}
+#| output: false
+library(here)
+library(tidyverse)
+library(fs)
+library(archive)
+source(here::here("R", "utils_generic.R"))
+```
+
+Set up some directories
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+data_raw <- here::here("_data_raw", "illumina_v3v4")
+amptar <- here::here(data_raw, "bbmap_rpkm.tar.gz")
+
+# create temporary location to decompress
+tmpdir <- fs::file_temp()
+
+# make processed data directory if it doesn't exist
+data <- here::here("data", "illumina_v3v4")
+fs::dir_create(data)
+```
+
+Untar and decompress
+
+```{r}
+# untar directory containing variant tables
+archive::archive_extract(
+ amptar,
+ dir = tmpdir,
+ files = NULL,
+ options = character(),
+ strip_components = 0L
+)
+
+ampdir <- here::here(tmpdir, "bbmap_rpkm")
+```
+
+# Reading and small formatting of data
+
+Coverage data
+```{r}
+ampfiles <- fs::dir_ls(
+ path = ampdir,
+ all = FALSE,
+ recurse = TRUE,
+ type = "file",
+ glob = "*.rpkm",
+ regexp = NULL,
+ invert = FALSE,
+ fail = TRUE
+)
+
+ampslurped <- readr::read_tsv(
+ ampfiles,
+ comment = "#",
+ col_names = c(
+ "strainID",
+ "Length",
+ "Bases",
+ "Coverage",
+ "count",
+ "RPKM",
+ "Frags",
+ "FPKM"
+ ),
+ col_types = "cddddddd",
+ id = "file_name"
+)
+```
+
+format the data nicely
+
+```{r}
+ampslurpedfmt <- ampslurped %>%
+ mutate(sample = str_extract(file_name, "HAMBI[:digit:]{4}")) %>%
+ left_join(tax, by = join_by(strainID)) %>%
+ dplyr::select(sample, strainID, genus, species, count) %>%
+ mutate(strainID = str_replace(strainID, "-", "_"))
+```
+
+cleanup temporary directory
+
+```{r}
+fs::dir_delete(tmpdir)
+```
+
+```{r}
+ampslurpedfmtnrm <- ampslurpedfmt %>%
+ mutate(
+ count = case_when(
+ # trunc ensures we get integers. important for some alpha div measures
+ .$strainID == "HAMBI_1842" ~ trunc(.$count/4),
+ .$strainID == "HAMBI_2659" ~ trunc(.$count/2),
+ TRUE ~ as.numeric(.$count)
+ ))
+```
+
+```{r}
+#| include: false
+ampslurpedfmtnrm <- ampslurpedfmt %>%
+ left_join(rRNA_counts, by = join_by(strainID)) %>%
+ group_by(sample) %>%
+ mutate(tot = sum(count)) %>%
+ mutate(count_norm = count/copies) %>%
+ mutate(f_norm = count_norm/sum(count_norm)) %>%
+ mutate(count_norm = trunc(f_norm*tot)) %>%
+ ungroup() %>%
+ dplyr::select(sample:count, count_norm)
+```
+
+```{r}
+#| include: false
+# Check for low total read count samples
+ids_good <- ampslurpedfmtnrm |>
+ group_by(sample) |>
+ summarize(sum=sum(count)) |>
+ filter(sum >= 10000) |>
+ pull(sample)
+# exclude low count samples
+ampslurpedfmtnrmflt <-ampslurpedfmtnrm |>
+ filter(sample %in% ids_good)
+```
+
+# Export
+
+```{r}
+write_tsv(ampslurpedfmtnrm, here(data, "species_counts.tsv"))
+```
diff --git a/R/illumina_v3v4/02_analysis.qmd b/R/illumina_v3v4/02_analysis.qmd
new file mode 100644
index 0000000..633809a
--- /dev/null
+++ b/R/illumina_v3v4/02_analysis.qmd
@@ -0,0 +1,570 @@
+---
+title: "Analysis of community composition"
+author: "Shane Hogle"
+date: today
+abstract: "This notebook takes formatted 16S rRNA amplicon counts and plots and analyzes them in different ways to look at community composition"
+---
+
+# Setup
+
+Libraries and global variables
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+library(tidyverse)
+library(here)
+library(fs)
+library(fishualize)
+library(Polychrome)
+library(withr)
+library(scales)
+library(patchwork)
+
+source(here::here("R", "utils_generic.R"))
+```
+
+Set up some directories
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+#|
+data_raw <- here::here("_data_raw", "illumina_v3v4")
+
+# make processed data directory if it doesn't exist
+data <- here::here("data", "illumina_v3v4")
+figs <- here::here("figs", "illumina_v3v4")
+
+fs::dir_create(data)
+fs::dir_create(figs)
+```
+
+## Some functions
+
+```{r}
+myarplot <- function(.data){
+ cil <- unique(.data$predator_history)
+ bac <- unique(.data$prey_history)
+
+ mytitle <- case_when(cil == "anc" & bac == "anc" ~ "anc bac + anc cil",
+ cil == "anc" & bac == "evo" ~ "evo bac + anc cil",
+ cil == "evo" & bac == "anc" ~ "anc bac + evo cil",
+ cil == "evo" & bac == "evo" ~ "evo bac + evo cil",
+ cil == "nopredator" & bac == "anc" ~ "anc bac",
+ cil == "nopredator" & bac == "evo" ~ "evo bac",
+ #cil == "nopredator" & bac == "evo" ~ "starting anc bac",
+ #cil == "nopredator" & bac == "evo" ~ "starting evo bac"
+ )
+
+ ggplot(.data) +
+ geom_area(aes(x=time_days, y=f, fill=strainID),
+ color="black", size=0.1) +
+ facet_wrap( ~ replicate, strip.position = "bottom") +
+ scale_fill_manual(values = hambi_colors) +
+ scale_y_continuous(limits = c(0,1), expand = c(0, 0), labels = scales::percent) +
+ scale_x_continuous(limits = c(4,60), breaks = c(20, 40, 60)) +
+ labs(x="", y="", fill="", title=mytitle) +
+ theme_bw() +
+ myartheme()
+
+}
+
+myartheme <- function(...){
+ theme(
+ panel.spacing.x = unit(0.05,"line"),
+ strip.placement = 'outside',
+ strip.background.x = element_blank(),
+ panel.grid = element_blank(),
+ panel.border = element_blank(),
+ panel.background = element_blank(),
+ #axis.text.x = element_blank(),
+ axis.line.x = element_line(color = "black"),
+ axis.line.y = element_line(color = "black"),
+ legend.title = element_blank(),
+ legend.background = element_blank(),
+ legend.key = element_blank(),
+ ...)
+}
+
+strain.order <- c("HAMBI_2443", "HAMBI_3031", "HAMBI_1988", "HAMBI_3237", "HAMBI_0262",
+ "HAMBI_2159", "HAMBI_1842", "HAMBI_2164", "HAMBI_0006", "HAMBI_0097",
+ "HAMBI_2494", "HAMBI_1299", "HAMBI_2160", "HAMBI_1279", "HAMBI_2792",
+ "HAMBI_1896", "HAMBI_0105", "HAMBI_0403", "HAMBI_1977", "HAMBI_1923",
+ "HAMBI_2659", "HAMBI_1292", "HAMBI_1972", "HAMBI_1287")
+```
+
+Read data
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+counts <- read_tsv(here::here(data, "species_counts.tsv"))
+metadf <- read_rds(here::here("data", "shared", "metadata_formatted.rds"))
+
+counts_f <- left_join(metadf, counts) %>%
+ group_by(sample) %>%
+ mutate(f=count/sum(count)) %>%
+ ungroup() %>%
+ filter(!is.na(strainID))
+```
+
+# Plot Community Composition
+
+community composition of the initial conditions
+
+```{r}
+pinit <- counts_f %>%
+ mutate(strainID=factor(strainID, levels=strain.order)) %>%
+ filter(str_detect(condition_prey_pred, "inoculum")) %>%
+ ggplot() +
+ geom_bar(aes(y = f, x=replicate, fill = strainID),
+ color="black", size=0.25, stat="identity") +
+ facet_grid(~ prey_history, scales="free_x") +
+ scale_fill_manual(values = hambi_colors) +
+ scale_y_continuous(limits = c(0,1), expand = c(0, 0), labels = scales::percent) +
+ labs(x="", y="% abundance", fill="", title="Community composition in starting inocula") +
+ theme_bw() +
+ myartheme()
+```
+
+::: {#fig-01}
+```{r}
+#| warning: false
+#| error: false
+#| fig.width: 8
+#| fig.height: 7
+pinit
+```
+Community composition of the inoculae used to start the experiment
+:::
+
+```{r}
+#| echo: false
+ggsave(
+ here::here(figs, "species_comp_init.png"),
+ pinit,
+ width = 7,
+ height = 5,
+ units = "in",
+ device = "png",
+ dpi = 320
+)
+
+ggsave(
+ here::here(figs, "species_comp_init.pdf"),
+ pinit,
+ width = 7,
+ height = 5,
+ units = "in",
+ device = "pdf"
+)
+```
+
+Plot community compositions during the experiment
+
+```{r}
+phi <- counts_f %>%
+ mutate(strainID=factor(strainID, levels=strain.order),
+ time_days = as.numeric(as.character(time_days))) %>%
+ filter(!is.na(time_days)) %>%
+ group_by(prey_history, predator_history) %>%
+ group_split() %>%
+ map(myarplot) %>%
+ wrap_plots(., ncol = 3) +
+ plot_layout(guides = 'collect') +
+ plot_annotation(tag_levels = 'A') &
+ theme(legend.position = 'bottom')
+```
+
+::: {#fig-02}
+```{r}
+#| warning: false
+#| error: false
+#| fig.width: 14
+#| fig.height: 8
+phi
+```
+Community composition temporal trajectories during the experiment
+:::
+
+```{r}
+#| echo: false
+ggsave(
+ here::here(figs, "species_comp_full.png"),
+ phi,
+ width = 14,
+ height = 8,
+ units = "in",
+ device = "png",
+ dpi = 320
+)
+
+ggsave(
+ here::here(figs, "species_comp_full.pdf"),
+ phi,
+ width = 14,
+ height = 8,
+ units = "in",
+ device = "pdf"
+)
+```
+
+# Ordination of community composition
+
+Need some additional libraries here
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+library(compositions)
+library(zCompositions)
+library(vegan)
+library(ape)
+library(corrr)
+library(ggrepel)
+```
+
+## Transform data
+
+zCompositions has problems with species with \< 2 observations so we need to filter these out
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+lowstrains <- counts_f %>%
+ filter(!str_detect(condition_prey_pred, "inoculum")) %>%
+ filter(!is.na(count)) %>%
+ group_by(strainID) %>%
+ summarize(n_samples = n(),
+ n_gt0 = sum(count > 0),
+ p_gt0 = n_gt0 / n_samples) %>%
+ ungroup()
+```
+
+Here we remove strains present in \< 50 samples across transfer categories and present in \< 20 samples in at least 2/3 transfer categories
+
+```{r}
+lowstrainsv <- c(
+ "HAMBI_0097",
+ "HAMBI_0262",
+ "HAMBI_1842",
+ "HAMBI_1988",
+ "HAMBI_2443",
+ "HAMBI_2792",
+ "HAMBI_3031",
+ "HAMBI_3237"
+)
+```
+
+transform to matrix
+
+```{r}
+mymat <- counts_f %>%
+ filter(!is.na(count)) %>%
+ filter(strainID %nin% lowstrainsv) %>%
+ filter(!str_detect(condition_prey_pred, "inoculum")) %>%
+ dplyr::select(sample, strainID, count) %>%
+ # important to arrange by sample as this makes some later joins easier
+ arrange(sample) %>%
+ pivot_wider(names_from = "strainID", values_from = "count") %>%
+ column_to_rownames(var = "sample") %>%
+ data.frame()
+```
+
+## Replace zeros
+
+Compositional analysis with the centered log-ratio can't handle zero values. Some people just replace them with a pseudocount. Another way is to impute them based on various different strategies.
+
+Literature:
+
+- [A field guide for the compositional analysis of any-omics data](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6755255/)
+ - [Supplemental material](https://zenodo.org/record/3270954#.Y1KastJBxhE)
+- [zCompositions --- R package for multivariate imputation of left-censored data under a compositional approach](https://doi.org/10.1016/j.chemolab.2015.02.019)
+
+Here we will uses a Geometric Bayesian-multiplicative replacement strategy that preserves the ratios between the non-zero components. The "prop" option returns relative abundances.
+
+```{r}
+set.seed(12378)
+comp <- zCompositions::cmultRepl(mymat, method = "GBM", output = "prop")
+```
+
+## Calculate Bray-curtis dissimilarity
+
+```{r}
+set.seed(12353)
+bray_dist <- vegan::vegdist(comp, method = "bray")
+pcoa_ord_bray <- ape::pcoa(bray_dist)
+```
+
+## Calculate with Aitchison distance
+
+Aitchison distance is the Euclidean distance of the centered log-ratio transform (clr). This distance (unlike Euclidean distance on read counts) has scale invariance, perturbation invariance, permutation invariance and sub-compositional dominance.
+
+```{r}
+set.seed(12354)
+balclr <- compositions::clr(comp)
+aitc_dist <- vegan::vegdist(balclr, method = "euclidean")
+```
+
+## Compare Aitchison distance with CLR
+
+When the Aitchison distance is used in Principle co-ordinate Analysis (PCoA) it is equivalent to standard Principle Component Analyis (PCA) on the clr transformed data
+
+```{r}
+set.seed(12355)
+pcoa_ord_aitc <- ape::pcoa(aitc_dist)
+pca_ord_aitc <- prcomp(balclr)
+```
+
+For example, these ordinations are the same, just that Axis2 is the mirrorimage between. Since the rotation is arbitrary this does not matter.
+
+::: {#fig-03}
+```{r}
+#| warning: false
+#| error: false
+#| fig.width: 6
+#| fig.height: 6
+plot(pcoa_ord_aitc$vectors)
+```
+
+Plot of Principal Coordinate Analysis (PCoA) done in the Aitchison geometry of the simplex (i.e. PCoA on the euclidean distance of the centered log-ratio transform species compositions).
+:::
+
+::: {#fig-04}
+```{r}
+#| warning: false
+#| error: false
+#| fig.width: 6
+#| fig.height: 6
+plot(pca_ord_aitc$x)
+```
+Plot of Principal Component Analysis (PCA) done in the Aitchison geometry of the simplex (i.e. PCA on the centered log-ratio transform species compositions). Notice that the PCoA using the euclidean distance of centered log-ratio transformed species frequencies is equivalent to the PCA directly using clr-transformed values in @fig-03.
+:::
+
+## Environment vectors
+
+left_join with metadata
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+pca2plot <- data.frame(pca_ord_aitc$x) %>%
+ rownames_to_column(var = "sample") %>%
+ left_join(metadf) %>%
+ arrange(sample) %>%
+ mutate(expcombo=interaction(condition_prey_pred, time_days, sep = "_"))
+```
+
+About 85% of variance explained in first 5 PCs
+
+```{r}
+pca_ord_aitc_importance <- as.data.frame(summary(pca_ord_aitc)$importance) %>%
+ rownames_to_column(var = "parameter")
+
+pca_ord_aitc_importance
+```
+
+Environmental/experimental variables associated with ordinatoion
+
+```{r}
+#| warning: false
+#| error: false
+ef <-
+ envfit(
+ pca_ord_aitc ~ prey_history + predator_history + ciliates_ml + time_days,
+ data = semi_join(metadf, dplyr::select(pca2plot, sample)),
+ na.rm = T,
+ choices = c(1, 2)
+ )
+```
+
+::: {#fig-05}
+```{r}
+#| warning: false
+#| error: false
+#| fig.width: 6
+#| fig.height: 6
+biplot(pca_ord_aitc, choices=c(1,2), labSize=0)
+plot(ef)
+```
+Environmental vectors plotted onto ordination in @fig-04
+:::
+
+### Significance of the environmental covariates
+
+```{r}
+ef
+```
+
+```{r}
+con_scrs <- as.data.frame(scores(ef, display = "vectors")) %>%
+ rownames_to_column(var = "var") %>% tibble()
+
+fct_scrs <- as.data.frame(scores(ef, display = "factors")) %>%
+ rownames_to_column(var = "var") %>% tibble()
+
+scale_factor <- 15
+```
+
+### Plot with prey history highlighted
+
+```{r}
+ppca <- ggplot(pca2plot) +
+ geom_point(aes(
+ x = PC1,
+ y = PC2,
+ color = prey_history,
+ shape = predator_history), size=3 ) +
+ geom_segment(data = con_scrs,
+ aes(x = 0, xend = PC1*scale_factor, y = 0, yend = PC2*scale_factor),
+ arrow = arrow(length = unit(0.25, "cm")), colour = "black") +
+ geom_text_repel(data = con_scrs, aes(x = PC1*scale_factor, y = PC2*scale_factor, label = var),
+ size = 3) +
+ labs(x = paste0("PC1 (", round(pca_ord_aitc_importance[2,2]*100, 1),"%)"),
+ y = paste0("PC2 (", round(pca_ord_aitc_importance[2,3]*100, 1),"%)")) +
+ stat_ellipse(aes(x = PC1, y = PC2, color = prey_history)) +
+ coord_fixed() +
+ scale_color_manual(values = c("#88CCEE", "#CC6677")) +
+ scale_shape_manual(values = c(1, 0, 15)) +
+ theme_bw() +
+ theme(
+ panel.grid.major = element_blank(),
+ panel.grid.minor = element_blank(),
+ panel.background = element_blank(),
+ )
+```
+
+::: {#fig-06}
+```{r}
+#| echo: false
+ppca
+```
+Plot of principal component analysis (PCA) done in the Aitchison geometry of the simplex (i.e. centered log-ratio transform species compositions). Color shows evolutionary history of the prey and shape shows evolutionary history of the predator. Arrows and ellipses depict statistically significant (P < 0.05) experimental variables (continuous and categorical, resp) fit to the ordination axes via regression (`vegan::envfit`) and projected onto the ordination plot. Significance is assessed by permutation.
+:::
+
+```{r}
+#| echo: false
+ggsave(
+ here::here(figs, "ordination_final.png"),
+ ppca,
+ width = 12,
+ height = 6,
+ units = "in",
+ device = "png",
+ dpi = 320
+)
+
+ggsave(
+ here::here(figs, "ordination_final.pdf"),
+ ppca,
+ width = 12,
+ height = 6,
+ units = "in",
+ device = "pdf"
+)
+```
+
+### Plot with predator history highlighted
+
+::: {#fig-07}
+```{r}
+ggplot(pca2plot) +
+ geom_point(aes(
+ x = PC1,
+ y = PC2,
+ color = predator_history,
+ shape = prey_history), size=3 ) +
+ geom_segment(data = con_scrs,
+ aes(x = 0, xend = PC1*scale_factor, y = 0, yend = PC2*scale_factor),
+ arrow = arrow(length = unit(0.25, "cm")), colour = "black") +
+ geom_text_repel(data = con_scrs, aes(x = PC1*scale_factor, y = PC2*scale_factor, label = var),
+ size = 3) +
+ labs(x = paste0("PC1 (", round(pca_ord_aitc_importance[2,2]*100, 1),"%)"),
+ y = paste0("PC2 (", round(pca_ord_aitc_importance[2,3]*100, 1),"%)")) +
+ stat_ellipse(aes(x = PC1, y = PC2, color = predator_history)) +
+ coord_fixed() +
+ theme_bw() +
+ theme(
+ panel.grid.major = element_blank(),
+ panel.grid.minor = element_blank(),
+ panel.background = element_blank(),
+ )
+
+```
+As in @Fig-06 but with predator evolutionary history highlighted in color.
+:::
+
+The evolved predator seems to "tighten" up the variation and make samplles with ancestral prey "look" more like the samples with evolved prey
+
+### PERMANOVA
+
+PERMANOVA suggests significant effect of predator and prey evolutionary history
+
+```{r}
+set.seed(45781)
+adonis2(aitc_dist ~ pca2plot$prey_history + pca2plot$predator_history,
+ permutations = 1e3)
+```
+
+Check for homogeneity of variance between these two categories. Suggests that they are different although maybe not by very much...
+
+```{r}
+set.seed(45782)
+bdprey <- betadisper(aitc_dist, pca2plot$prey_history)
+permutest(bdprey)
+```
+
+```{r}
+set.seed(45783)
+bdpred <- betadisper(aitc_dist, pca2plot$predator_history)
+permutest(bdpred)
+```
+
+
+## Geometric analysis of temporal trajectories in PCA
+
+Function to calculate euclidean distance between points
+```{r}
+euclidean_dist <- function(x1, x2, y1, y2){
+ sqrt((x1 - x2)^2 + (y1 - y2)^2)
+}
+```
+
+```{r}
+eucdist <- pca2plot %>%
+ filter(prey_history != "none") %>%
+ filter(predator_history != "nopredator") %>%
+ dplyr::select(PC1, PC2, replicate, prey_history, predator_history, time_days) %>%
+ arrange(replicate, prey_history, predator_history, time_days) %>%
+ group_by(replicate, prey_history, predator_history) %>%
+ mutate(d = euclidean_dist(PC1, lag(PC1), PC2, lag(PC2)),
+ range = paste0(lag(time_days),"-",time_days)) %>%
+ mutate(range = forcats::fct_reorder(range, as.numeric(as.character((time_days))))) %>%
+ filter(range != "NA-4")
+
+write_tsv(eucdist, here::here(data, "aitchison_pca_distances.tsv"))
+```
+
+::: {#fig-08}
+```{r}
+ggplot(eucdist) +
+ geom_line(aes(x = range, y = d, color = replicate, group=replicate)) +
+ labs(x = "Time range (days)", y = "Euclidean distance between days") +
+ scale_x_discrete(guide = guide_axis(angle = 45)) +
+ facet_grid(prey_history ~ predator_history, labeller = label_both, scales = "free_x") +
+ theme_bw() +
+ theme(
+ legend.position = "bottom",
+ strip.placement = 'outside',
+ strip.background = element_blank(),
+ panel.grid = element_blank())
+```
+Geometric analysis of community trajectories based on PCA. Segment lengths between consecutive sampling days in the two PCA dimensions were calculated and plotted chronologically for each biological replicate in each treatment combination. As sampling intervals were of equal duration (except day 16 - 24), segment length represents relative speed of compositional change.
+:::
\ No newline at end of file
diff --git a/R/metagenome/01_format_variants.qmd b/R/metagenome/01_format_variants.qmd
new file mode 100644
index 0000000..ff1e695
--- /dev/null
+++ b/R/metagenome/01_format_variants.qmd
@@ -0,0 +1,412 @@
+---
+title: "Format metagenome variants"
+author: "Shane Hogle"
+date: today
+abstract: "This notebook formats for downstream use the table and VCF output from mutect2 called on metagenome data. It does some filtering to remove variants that are likely spurious and connects variants to functional annotations and gene identities. Finally it joins the WGS variants formatted earlier as the initial sampling point so that we can make time series from 0, 8, 28, and 60 days."
+---
+
+# Setup
+
+Libraries and global variables
+
+```{r}
+#| output: false
+library(here)
+library(tidyverse)
+library(Polychrome)
+library(withr)
+library(fs)
+library(archive)
+source(here::here("R", "utils_generic.R"))
+```
+
+Set up some directories
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+data_raw <- here::here("_data_raw", "metagenome")
+shared <- here::here("_data_raw", "shared")
+vcftar <- here::here(data_raw, "tables.tar.gz")
+covtar <- here::here(data_raw, "coverage.tar.gz")
+
+# create temporary location to decompress
+tmpdir <- fs::file_temp()
+
+# make processed data directory if it doesn't exist
+data <- here::here("data", "metagenome")
+fs::dir_create(data)
+```
+
+## Untar and decompress
+
+```{r}
+# untar directory containing variant tables
+archive::archive_extract(
+ vcftar,
+ dir = tmpdir,
+ files = NULL,
+ options = character(),
+ strip_components = 0L
+)
+
+vcfdir <- here::here(tmpdir, "tables")
+
+# untar directory containing variant tables
+archive::archive_extract(
+ covtar,
+ dir = tmpdir,
+ files = NULL,
+ options = character(),
+ strip_components = 0L
+)
+
+covdir <- here::here(tmpdir, "coverage")
+```
+
+
+# Reading and small formatting of data
+
+## Coverage data
+```{r}
+#| eval: false
+#| echo: true
+covfiles <- fs::dir_ls(
+ path = covdir,
+ all = FALSE,
+ recurse = TRUE,
+ type = "file",
+ glob = "*.coverage.tsv",
+ regexp = NULL,
+ invert = FALSE,
+ fail = TRUE
+)
+
+covslurped <- readr::read_tsv(
+ covfiles,
+ skip = 1,
+ col_names = c(
+ "scaffold",
+ "mean",
+ "trimmed_mean",
+ "covered_fraction",
+ "covered_bases",
+ "variance",
+ "length",
+ "read_count",
+ "reads_per_base"
+ ),
+ col_types = "cdddddddd",
+ id = "file_name"
+)
+
+covslurpedfmt <- covslurped %>%
+ mutate(sample = str_extract(file_name, "SH-MET-[:digit:]{3}"),
+ strainID = str_extract(file_name, "HAMBI_[:digit:]{4}")) %>%
+ dplyr::select(-file_name) %>%
+ relocate(strainID, sample)
+
+covslurpedfmtflt <- covslurpedfmt %>%
+ mutate(bamname = paste0(strainID,"-",sample)) %>%
+ # exclude species with trimmed mean (Dt) < 5
+ mutate(FILTER = if_else(trimmed_mean >= 5, FALSE, TRUE))
+
+write_tsv(covslurpedfmtflt, here::here(data, "coverage.tsv"))
+
+# cleanup
+fs::dir_delete(covdir)
+```
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+covslurpedfmtflt <- read_tsv(here::here(data, "coverage.tsv"))
+```
+
+## Variants
+
+### SnpEff annotations
+
+```{r}
+snpefffiles <- fs::dir_ls(
+ path = vcfdir,
+ all = FALSE,
+ recurse = TRUE,
+ type = "file",
+ glob = "*.snpeff.tsv",
+ regexp = NULL,
+ invert = FALSE,
+ fail = TRUE
+)
+
+snpeffslurped <- readr::read_tsv(
+ snpefffiles,
+ skip = 1,
+ col_names = c(
+ "chrom",
+ "pos",
+ "ref",
+ "alt",
+ "filter",
+ # Annotated using Sequence Ontology terms. Multiple effects can be concatenated using '&'.
+ "effect",
+ # A simple estimation of putative impact / deleteriousness : HIGH, MODERATE, LOW, or MODIFIER
+ "impact",
+ # gene id from prokka
+ "locus_tag",
+ # whether variant is coding or noncoding
+ "biotype",
+ # Variant using HGVS notation (DNA level)
+ "hgvs_c",
+ # Variant using HGVS notation (Protein level).
+ "hgvs_p",
+ # Position in coding bases (one based includes START and STOP codons).
+ "cds_pos",
+ # Total number of coding bases (one based includes START and STOP codons).
+ "cds_len",
+ # Position in translated product (one based includes START and STOP codons).
+ "aa_pos",
+ # Total length of translated product (one based includes START and STOP codons).
+ "aa_len"
+ ),
+ col_types = c("cdcccccccccdddd"),
+ id = "sample"
+)
+
+snpeffslurpedfmt <- snpeffslurped %>%
+ # this is weird, but joining on column like filter that contains characters like ';' doesnt work
+ dplyr::select(-filter) %>%
+ mutate(sample = str_extract(sample,
+ regex("(HAMBI_[:digit:]{4}-SH-MET-[:digit:]{3})"))) %>%
+ separate(sample, into = c("strainID", "sample"), sep="-", extra = "merge")
+```
+
+
+### Allele frequencies
+
+```{r}
+varfiles <- fs::dir_ls(
+ path = vcfdir,
+ all = FALSE,
+ recurse = TRUE,
+ type = "file",
+ glob = "*.variants.tsv",
+ regexp = NULL,
+ invert = FALSE,
+ fail = TRUE
+)
+
+varslurped <- readr::read_tsv(
+ varfiles,
+ skip = 1,
+ col_names = c(
+ "chrom",
+ "pos",
+ "ref",
+ "alt",
+ "filter",
+ "type",
+ # Allelic depths for the ref and alt alleles in the order listed
+ "ad",
+ # Allele fractions of alternate alleles. Excludes filtering
+ "freq_alt_raw",
+ # Approximate read depth (reads are filtered if MQ=255 or with bad mates)
+ "depth_total",
+ # Count of fragments supporting each allele.
+ "frag_allele_depth",
+ # Genotype Quality
+ "genotype_qual"
+ ),
+ col_types = c("cdccccccccc"),
+ id = "sample"
+)
+
+varslurpedfmt <- varslurped %>%
+ mutate(sample = str_extract(sample,
+ regex("(HAMBI_[:digit:]{4}-SH-MET-[:digit:]{3})"))) %>%
+ separate(sample, into = c("strainID", "sample"), sep="-", extra = "merge") %>%
+ # taken only first two most abundant alleles
+ separate(ad, into = c("depth_ref", "depth_alt"),
+ sep=",", extra="drop") %>%
+ # format depths to numberic
+ mutate(depth_ref = as.numeric(depth_ref),
+ depth_alt = as.numeric(depth_alt)) %>%
+ select(-frag_allele_depth, -genotype_qual) %>%
+ mutate(freq_alt = depth_alt/(depth_ref + depth_alt),
+ freq_ref = 1 - freq_alt) %>%
+ # final format depths to numeric
+ mutate(freq_alt_raw = as.numeric(freq_alt_raw),
+ depth_total = as.numeric(depth_total))
+```
+
+### Clean up tmp directories
+```{r}
+# remove decompressed vcf and tables directory from temp location
+fs::dir_delete(vcfdir)
+```
+
+## Mutations from WGS analysis
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+wgs_freq <- read_tsv(here::here("data", "wgs", "wgsvars_filt_mb_nonsyn.tsv"))
+```
+
+## Annotations
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+#annotations <- read_rds(here::here(data_raw, "annotations_codon_degeneracy.rds"))
+mgefinder <- read_tsv(here::here(shared, "MGE_finder_HAMBI_combined.tsv"))
+genomad <- read_tsv(here::here(shared, "genomad_HAMBI_combined.tsv")) %>%
+ # exclude very large plasmids over 1Mbp from the filtering
+ filter((end - start) < 1000000)
+```
+
+# Formatting and filtering
+
+```{r}
+# Combine snpeff and var freqs
+mgvars <- left_join(varslurpedfmt, snpeffslurpedfmt,
+ by = join_by(strainID, sample, chrom, pos, ref, alt))
+```
+
+```{r}
+# get species called in metagenomes
+sps <- pull(mgvars, chrom) %>%
+ str_extract("HAMBI_\\d{4}") %>%
+ unique()
+
+sps
+```
+
+## Exclude regions with mobile elements in metagenomes
+```{r}
+mgvars_filt_mb <- mgvars %>%
+ left_join(mgefinder, by = join_by(strainID, chrom), relationship = "many-to-many") %>%
+ filter(!(pos >= start & pos <= end)) %>%
+ dplyr::select(-name, -start, -end, -prediction_tool) %>%
+ distinct() %>%
+ left_join(genomad, by = join_by(strainID, chrom), relationship = "many-to-many") %>%
+ filter(!(pos >= start & pos <= end)) %>%
+ dplyr::select(-name, -start, -end, -prediction_tool) %>%
+ distinct()
+```
+
+## Filter WGS on species only included in metaG
+```{r}
+wgs_filt_mb <- wgs_freq %>%
+ filter(strainID %in% sps) %>%
+ mutate(anc = "anc", evo = "evo") %>%
+ pivot_longer(cols = c(anc, evo), names_to = "predator_history", values_to = "tmp") %>% select(-tmp) %>%
+ mutate(prey_history = "evo",
+ transfer_category = "hi",
+ transfer_volume_ul = 1800,
+ time_days = 0)
+```
+
+## Combine WGS and metaG
+
+This includes all variants - both amino acid changing and non-changing
+```{r}
+freq_mg_wgs <- left_join(mgvars_filt_mb,
+ read_tsv(here::here(data, "metadata.tsv"))) %>%
+ filter(transfer_category == "hi") %>%
+ bind_rows(wgs_filt_mb) %>%
+ mutate(pos = as.numeric(pos),
+ depth_total = as.numeric(depth_total)) %>%
+ select(-hgvs_c, -aa_pos, -aa_len)
+```
+
+## Expand grid for variants
+
+Make a dataframe where all times/species/replicates/genomic position/experimental conditions are present
+```{r}
+vars_united <- freq_mg_wgs %>%
+ select(chrom, pos, ref, alt) %>%
+ distinct() %>%
+ unite("combo", c(chrom, pos, ref, alt), sep = "|")
+
+times_full <- expand_grid(combo = vars_united$combo,
+ time_days = c(0, 8, 28, 60),
+ replicate = c("A", "C", "E"),
+ prey_history = c("anc", "evo"),
+ predator_history = c("anc", "evo")) %>%
+ separate_wider_delim(combo, delim = "|", names = c("chrom", "pos", "ref", "alt")) %>%
+ mutate(pos = as.numeric(pos),
+ transfer_category = "hi",
+ transfer_volume_ul = 1800)
+```
+
+## Format metagenome coverage data
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+mg_cov <- covslurpedfmtflt %>%
+ filter(strainID %in% sps) %>%
+ left_join(read_tsv(here::here(data, "metadata.tsv"))) %>%
+ filter(transfer_category == "hi") %>%
+ # samples with trimmed mean coverage >=10 with over 99% of genome covered
+ # are to be trusted
+ mutate(cov_thresh_met = if_else(trimmed_mean >= 10 & covered_fraction > 0.99, 1, 0)) %>%
+ select(sample, strainID, chrom=scaffold, trimmed_mean, covered_fraction, transfer_category:cov_thresh_met)
+```
+
+# Combine variants into complete time series
+```{r}
+freq_full <- left_join(times_full, freq_mg_wgs, relationship = "one-to-one") %>%
+ arrange(chrom, pos, ref, alt, replicate, predator_history, prey_history, time_days) %>%
+ # group_by(chrom) %>%
+ fill(strainID, .direction = "updown") %>%
+ ungroup() %>%
+ group_by(time_days, replicate, prey_history, predator_history) %>%
+ fill(sample, .direction = "updown") %>%
+ ungroup() %>%
+ left_join(mg_cov) %>%
+ mutate(cov_thresh_met = if_else(time_days == 0, 1, cov_thresh_met)) %>%
+ mutate(freq_alt_complete = case_when(cov_thresh_met == 1 & !is.na(freq_alt) ~ freq_alt,
+ cov_thresh_met == 1 & is.na(freq_alt) ~ 0,
+ TRUE ~ NA_real_))
+```
+
+Some filtering to reduce size. Also record the number of mutect2 passing variants and add a time-lagged variable. We remove variants that only have an observation at T0 and never in any metagenome aample due to low coverage (i.e. they are NA). This will reduce the size of the dataset and make sure we focus on more than just signals present in the WGS analysis
+```{r}
+freq_full_sm01 <- freq_full %>%
+ group_by(chrom, pos, ref, alt, replicate, prey_history, predator_history) %>%
+ mutate(id = cur_group_id()) %>%
+ ungroup() %>%
+ group_by(id) %>%
+ # can't be all zeros
+ filter(sum(freq_alt_complete, na.rm=TRUE) > 0) %>%
+ # exclude only NAs
+ filter(sum(is.na(freq_alt_complete[time_days != 0])) < 3) %>%
+ # mark some statistics that may be useful for later filtering
+ mutate(n_gt_fthresh = sum(freq_alt_complete >= 0.05, na.rm = T),
+ n_pass = sum(filter == "PASS", na.rm = T),
+ n_cov_gt = sum(cov_thresh_met, na.rm =T)) %>%
+ # freq_alt_complete_t1=lag(freq_alt_complete, order_by = time_days, n=1)
+ ungroup()
+```
+
+Here we require at least one observation with greater than 0.05 alt frequency and at least one variant passing mutect filters in the variant time series
+
+```{r}
+freq_full_sm02 <- freq_full_sm01 %>%
+ filter(n_gt_fthresh >= 1 & n_pass >= 1) %>%
+ group_by(id) %>%
+ fill(c(type, effect, impact, locus_tag, biotype, hgvs_p, cds_pos, cds_len), .direction = "updown")
+```
+
+# Export
+```{r}
+# save table for later use
+write_tsv(freq_full_sm02, here::here(data, "metagenome_variant_timeseries.tsv"))
+```
diff --git a/R/metagenome/02_parallelism.qmd b/R/metagenome/02_parallelism.qmd
new file mode 100644
index 0000000..6b72093
--- /dev/null
+++ b/R/metagenome/02_parallelism.qmd
@@ -0,0 +1,793 @@
+---
+title: "Metagenome genetic parallelism"
+author: "Shane Hogle"
+date: today
+link-citations: true
+bibliography: references.bib
+abstract: "This notebook reads in the formatted metagenome variant frequencies and annotations. It then uses statistical tests to determine which mutations and which functional categories appeared across biological replicates within condition more often than would be expected by chance. It also looks for mutations/functions that arise within treatment categories more than expected by chance."
+---
+
+# Setup
+
+Libraries and global variables
+
+```{r}
+#| output: false
+library(tidyverse)
+library(here)
+library(fs)
+library(patchwork)
+library(ggforce)
+library(latex2exp)
+library(vctrs)
+
+source(here::here("R", "utils_generic.R"))
+source(here::here("R", "metagenome", "utils_parallelism.R"))
+```
+
+Set up some directories
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+data_raw <- here::here("_data_raw", "metagenome")
+data <- here::here("data", "metagenome")
+shared <- here::here("_data_raw", "shared")
+figs <- here::here("figs", "metagenome")
+
+# make processed data directory if it doesn't exist
+fs::dir_create(data)
+fs::dir_create(figs)
+```
+
+# Read data
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+
+# these were already filtered in the last step
+mgvars <- read_tsv(here::here(data, "metagenome_variant_timeseries.tsv"))
+degentab <- read_rds(here::here(shared, "annotations_codon_degeneracy.rds"))
+genome_len <- read_tsv(here::here(shared, "HAMBI_genome_len.tsv"))
+# annotations
+annotations <- read_rds(here::here(shared, "annotations_codon_degeneracy.rds"))
+```
+
+# Parallelism at nucleotide level
+
+```{r}
+# exclude fixed variants from day 0
+tofilter <- mgvars %>%
+ filter(time_days == 0) %>%
+ filter(freq_alt_complete == 1) %>%
+ dplyr::select(chrom, pos, ref, alt) %>%
+ distinct()
+
+# mobile regions have already been masked in the data wrangling step
+
+mgvars_filt_mb <- anti_join(mgvars, tofilter, by = join_by(chrom, pos, ref, alt)) %>%
+ # exclude variants that were added to make complete time course
+ filter(!is.na(freq_alt_complete)) %>%
+ filter(freq_alt_complete > 0) %>%
+ filter(time_days != 0)
+```
+
+Looking at all mutations (inside and outside of coding sequences, synonymous and nonsynonymous)
+
+```{r}
+mgvars_filt_mb %>%
+ group_by(strainID, prey_history, predator_history, chrom, pos, ref, alt) %>%
+ mutate(mi=n_distinct(replicate)) %>%
+ group_by(strainID) %>%
+ count(mi) %>%
+ group_by(mi) %>%
+ mutate(tot = sum(n)) %>%
+ arrange(desc(mi))
+```
+
+We identified 13 species $\times$ treatment combinations with detectable mutations. To put these observations in context, we can compare them to a simple null model in which mutations are uniformly distributed across the sites in the genome. We can then check how many sites with 1, 2, and 3-fold multiplicity that we would expect by chance. We define nucleotide parallelism as the number of mutations occurring at the same site in the genome in independent populations for a given species $\times$ treatment combination. For each site, we define the multiplicity, $m_{i}$, as the number of populations (i.e., replicates) with a mutation detected at that site, so that the multiplicity in this experiment can range from one to three.
+
+The expected number of mutations with $m_{i} \geq m$ in a sample of total mutations size $n_{tot}$ is:
+
+$$ S(m) \approx \sum_{n \geq m} \frac{n}{n_{tot}} \cdot L_{tot} \cdot \frac{ \left( \frac{n_{tot}}{L_{tot}} \right) ^n}{n!} e^{-n_{tot}/L_{tot}} $$
+
+where $L_{tot}$ is the total number of bases in the genome.
+
+## Nucleotide parallelism by treatment
+
+Here we will do the estimation separately for each species $\times$ treatment combination.
+
+```{r}
+nuc_survival <- mgvars_filt_mb %>%
+ dplyr::select(strainID, prey_history, predator_history, pos, ref, alt, replicate) %>%
+ # multiplicity = number of replicate populations each unique mutation is
+ # observed in
+ summarize(m = n_distinct(replicate), .by = c(strainID, prey_history, predator_history, pos, ref, alt)) %>%
+ # now calculate the total number of mutations across all replicates so we need
+ # to ungroup by mutation position/alt allele but because we still want to
+ # determine this value by treatment category we keep the treatment category
+ # grouping. However, this should be changed if you want to for example average
+ # over all the treatment conditions on a species basis
+ group_by(strainID, prey_history, predator_history) %>%
+ count(m, name = "m_count") %>%
+ mutate(n = m * m_count,
+ Ntot = sum(n),
+ perc = n / Ntot * 100) %>%
+ left_join(genome_len, by = join_by(strainID)) %>%
+ arrange(cur_group_id(), desc(m)) %>%
+ # dpois() tells the probability mass at a given number of counts. Here we
+ # want to get the probability of observing n mutations with multiplicity
+ # = mi (i.e. the counts of mi in the observed data). We assume that
+ # mutations independently occur on the genome of size Ltot at a rate of
+ # lambda = Ntot/Ltot and that generally the events are rare. Thus this
+ # situation can be modeled by the Poisson distribution. We can get the
+ # binned number of mutations per level of multiplicity m by multiplying
+ # the probability by the length of the genome and the binned mutations
+ # divided by the total number of mutations.
+ mutate(m_count_expected = cumsum((m_count / Ntot) *
+ total_len *
+ dpois(m, lambda = Ntot / total_len))) %>%
+ dplyr::select(-num_contigs) %>%
+ relocate(m, n, Ntot, perc, m_count, m_count_expected) %>%
+ ungroup()
+
+# setup for plotting
+nuc_survival_plot <- nuc_survival %>%
+ group_by(strainID, prey_history, predator_history) %>%
+ # when there is only one multiplicity observed for a mutation filter such
+ # that the multiplicty of that mutation must be greater than one.
+ # Otherwise include all remaining mutations (m > 0)
+ filter(case_when(n() == 1 ~ m > 1,
+ TRUE ~ m > 0)) %>%
+ pivot_longer(cols = c("m_count", "m_count_expected")) %>%
+ mutate(label = paste(strainID, "prey:", prey_history, "\npredator:", predator_history)) %>%
+ # and make final plot
+ plot_nuc_survival(., 5000, c(1, 10, 100, 1000, 10000), 4)
+```
+
+::: {#fig-01}
+```{r}
+#| fig.width: 8
+#| fig.height: 6
+#| echo: false
+#| warning: false
+nuc_survival_plot
+```
+
+Distribution of nucleotide multiplicity $(m_{i})$ - i.e., the number of replicates with with the same genomic mutation including site and alternative allele - for species $\times$ treatment combination with sufficient sequencing depth and multiplicities $\gt 1$. The observed data is shown in blue while the null expectation is shown in red. In all cases the observed $m_{i}$ exceeded the null expectation meaning that we must reject the simple null model of uniform mutation distribution in favor of the alternative model where mutations are nonrandom and cluster across replicate populations.
+:::
+
+## Conclusion
+
+Nucleotide parallelism results are presented in Figure @fig-01. For these species and evolution treatments this very simple null model mostly predicts that we should expect fewer than two parallel mutations (same position, same alternative allele) across the three replicate populations. For 8 species $times$ treatment combinations the observed data show an excess of nucleotide parallelism relative to this simple null expectation. In particular all species have at least 1 mutation that is identical across all three replicate evolved populations. HAMBI_1287, HAMBI_1972, and HAMBI_2659 did evolve some new parallel nucleotide mutation de novo in the ancestral prey treatment. These will be interesting to follow up on. Still, multi-hit sites constitute only a small fraction of the total observed mutations across all species (\~ 14% across species with multi-hit sites). Thus, we will continue our analysis but looking at the gene level.
+
+```{r}
+#| warning: false
+nuc_survival %>%
+ mutate(Ntot_f = sum(n)) %>%
+ summarize(perc = sum(n)/Ntot_f*100, .by = "m") %>%
+ distinct() %>%
+ mutate(cum_perc = cumsum(perc))
+```
+
+# Parallelism at gene level
+
+If selection pressures and mutation rates did not vary between genes, the number of mutations in each gene should be proportional to the target gene size. We assume that gene length is a primary driver of the target size. We then define a multiplicity for each gene according to:
+
+$$m_{i} = n_{i} \cdot \frac{\overline{L}}{L_{i}}$$ {#eq-01}
+
+where $n_{i}$ is the number of nonsynonymous mutations (i.e., amino acid changing) in $gene_{i}$ across all replicate populations per species, $L_{i}$ is the total number of nonsynonymous sites (i.e., excluding 4-fold degenerate sites) in $gene_{i}$, and $\overline{L}$ is the average value of $L_{i}$ across all genes in the genome. This definition ensures that under the null hypothesis, all genes have the same expected multiplicity $\overline{m} = n_{tot}/n_{genes}$.
+
+We next define a null model for gene multiplicity that assumes mutations are assigned to genes with probability:
+
+$$p_{i} \propto L_{i} r_{i}$$ {#eq-02}
+
+for some set of enrichment factors $r_{i}$. In the alternate model, the maximum likelihood estimator of the enrichment factors $r_{i}$ is the ratio of observed to expected gene multiplicities, $r_{i} = m_{i}/\overline{m}$. The net increase of the log-likelihood relative to the null model ($r_{i} = 1$) is then given by:
+
+$$\Delta \ell = \sum_{i} n_{i} \log \left ( \frac{m_{i}}{\overline{m}} \right )$$ {#eq-03}
+
+This likelihood ratio estimator is equivalent to the total G-score introduced in earlier work [@tenaillon2016]. Here we assess statistical significance using permutation tests following Shoemaker *et al.* [@shoemaker2021]. For permutations tests (n = 10,000) we randomly resample mutations to the observed $n_{tot}$ set size from each species using the multinomial distribution, where the probability of sampling a non-synonymous mutation at $gene_{i}$ was given by $p_{i} = {L_{i}} / {L_{tot}}$ where $L_{i}$ is the total number of nonsynonymous sites in $gene_{i}$ and $L_{tot}$ is the total number of nonsynonymous sites in the genome.
+
+We next identify specific genes enriched for mutations following the criteria of Good *et. al* . [@good2017]. We calculate a \textit{p} value for each gene using the equation:
+
+$$P_{i} = \sum_{n \geq n_{i}} \frac{ \left ( \frac{n_{tot} L_{i}}{ \overline{L} N_{genes}} \right )^{n}}{n!} e^{- \frac{n_{tot} L_{i} }{\overline{L} N_{genes}}}$$ {#eq-04}
+
+Here we only consider genes with $n_{i} \geq 3$ to exclude low \textit{p} values driven primarily by gene length. Under the null hypothesis, the expected number of genes with $P_{i} \geq p$ can be found using a Poisson survival curve given by:
+
+$$\overline{N}(P) \approx \sum_{i=1}^{N_{genes}} \sum_{n=3}^{\infty} \theta (P - P_{i}(n, L_{i}) \cdot \frac{ \left( \frac{n_{tot} L_{i}}{\overline{L} N_{genes}} \right)^{n}}{n!} e^{-\left( \frac{n_{tot} L_{i}}{\overline{L} N_{genes}} \right)}$$ {#eq-05}
+
+We can compare this expected number to the observed number of genes $N(P)$ using a critical \textit{p} value ($P^{\ast}$) such that
+
+$$\frac{\overline{N}(P^{\ast})}{N(P^{\ast})} \leq \alpha$$ {#eq-06}
+
+for a given FDR $\alpha$ value of 0.05. For this value we then define the set of significantly enriched genes as:
+
+$$I = \left\{ i : P_{i} \leq P^{\ast} \left(\alpha \right) \right\}$$ {#eq-07}
+
+## Prepare input data for non-syn mutations only
+
+
+```{r}
+mgvars_filt_mb_ns <- mgvars_filt_mb %>%
+ # exclude intragenic, intergenic, and synonymous mutations. Also exclude
+ # fusion functions because these are weird and also rare. Excluding the
+ # modifier category also ensures that we filter out any tRNAs with mutations
+ filter(!str_detect(effect, "intergenic|intragenic|synonymous|fusion")) %>%
+ filter(!str_detect(impact, "MODIFIER"))
+```
+
+## G score
+
+Warning! this takes some time
+
+```{r}
+#| eval: false
+#| echo: true
+muts_by_sp_treat <- mgvars_filt_mb_ns %>%
+ dplyr::select(-c(freq_alt_complete, time_days)) %>%
+ group_by(strainID, prey_history, predator_history) %>%
+ distinct() %>%
+ mutate(groupid = cur_group_id()) %>%
+ group_by(groupid, strainID, prey_history, predator_history)
+
+# doing 10000 permutations
+result_sp_treat <- run_full_gene_parallelism_pipeline(muts_by_sp_treat, degentab, 10000)
+
+# Save this for later
+write_rds(result_sp_treat, here::here(data, "parallelism_results_sp_treat.rds"))
+```
+
+```{r}
+# Read in previous results
+result_sp_treat <- read_rds(here::here(data, "parallelism_results_sp_treat.rds"))
+```
+
+### Simulating the number of replicates with a parallel gene hit
+
+As another sanity check, we took the resamplings from the null distribution and calculated the number of replicates with a parallel gene hit (i.e. the same gene hit in $\geq 2$ experimental replicates) and compared that to the observed distribution. The null distribution reflected the process of sampling a mutation under a multinomial distribution with probability proportional to the total number of nonsynonymous sites $L_{i}$ in $gene_{i}$. We observed that in 9/16 cases the null distribution produced fewer parallel gene hits than observed in the data from 2 experimental replicates. In all 16 cases the null distribution produced fewer parallel gene hits than observed in the data from 3 experimental replicates. Thus, for some suspicious genes with very high mutational density (see below) we required these genes to be present in all three replicates to include them in our final significant gene lists.
+
+::: {#fig-03}
+```{r}
+#| fig.width: 8
+#| fig.height: 8
+#| echo: false
+#| warning: false
+plot_sims(result_sp_treat, 1)
+```
+
+Number of genes (y axis) with nonsynonymous mutation hits in $N_{replicates} = \{n : 1 \leq n \leq 3\}$ (x axis). Blue bars depict the distribution observed in the data, while red bars depict the null distribution from resampling mutations to the observed $n_{tot}$ set size from each species using the multinomial distribution, where the probability of sampling a non-synonymous mutation at $gene_{i}$ was dependent only on the number of nonsynonymous sites in the gene.
+:::
+
+### Visualize the survival curves for each treatment combination
+
+```{r}
+survcurv <- result_sp_treat$output_df2plot %>%
+ filter(!is.na(xend)) %>%
+ mutate(label = paste(strainID, "prey:", prey_history, "\npredator:", predator_history)) %>%
+ ggplot() +
+ geom_segment(
+ aes(
+ x = x,
+ y = y,
+ xend = xend,
+ yend = yend
+ ),
+ linetype = "dashed",
+ linewidth = 0.25, color = "red"
+ ) +
+ geom_step( aes(x = obs_p, y = value, color = name)) +
+ facet_wrap(~ label, ncol = 4) +
+ labs(x = TeX("$-log_{10}P$"), y = "Number of genes") +
+ scale_color_manual(values = c("blue", "grey50")) +
+ scale_y_log10(
+ breaks = c(1, 10, 100),
+ labels = scales::trans_format("log10", scales::math_format(10 ^.x))
+ ) +
+ annotation_logticks(sides = "l", color = "grey40") +
+ coord_cartesian(ylim = c(1, 200)) +
+ theme_bw() +
+ theme(
+ strip.background = element_blank(),
+ panel.grid = element_blank(),
+ legend.position = "bottom"
+ )
+```
+
+::: {#fig-04}
+```{r}
+#| fig.width: 8
+#| fig.height: 8
+#| echo: false
+#| warning: false
+survcurv
+```
+
+The observed $(N)$ (blue line) and expected number $(\overline{N})$ (grey line) of genes with $P_{i} \leq P$, as a function of $P$. The red dashed line denotes the genome-wide significance threshold $P^{*}$ for $\alpha = 0.05$ defined in Equation @eq-06.
+:::
+
+
+### Overall G scores
+```{r}
+result_sp_treat$output_gscore %>%
+ filter(pvalue <= 0.05)
+```
+
+### Inspecting individual genes
+
+Some genes tend to have a large number of consecutive mutations which is probably due to incorrect read mapping and not real mutations so we will try and filter these out. We'll do this by filtering on genes with a mutational density $d_{g} \gt 0.15$, where the mutational density in gene $g$ was defined as
+
+$$d_{g} = \frac{|A|_{g}}{b_{g} - a_{g}}$$
+
+for all genome positions $x$ on the interval $\{x | a_{g} \leq x \leq b_{g}\}$ where $b_{g}$ is the greatest position and $a_{g}$ is the smallest position of observed mutations in gene $g$ and $|A|_{g}$ is the cardinality of the mutation set.
+
+```{r}
+high_density_genes <- mgvars_filt_mb_ns %>%
+ group_by(strainID, locus_tag, replicate, prey_history, predator_history) %>%
+ mutate(mutwindox = pos - lag(pos),
+ mut_num = n(),
+ mut_dens = n()/(max(pos) - min(pos))) %>%
+ ungroup() %>%
+ filter(mut_dens != Inf) %>%
+ filter(mut_dens > 0.15) %>%
+ distinct(locus_tag) %>%
+ pull()
+```
+
+```{r}
+output_gene_table_filt <- result_sp_treat$output_gene_table %>%
+ # only genes exceeding the critical pstar value from the survival curves above
+ filter(neg_log10P >= pstar) %>%
+ # only include genes with at least 3 hits in the same gene or found in more than one replicate.
+ filter(observed_hits_n_i >= 3 | n_replicate > 1) %>%
+ # filters out the problematic genes above unless the number of replicates detected is 2 or more
+ filter(case_when(n_replicate < 2 ~ !(locus_tag %in% high_density_genes),
+ TRUE ~ n_replicate > 1)) %>%
+ left_join(annotations, by = join_by(locus_tag, strainID))
+```
+
+# Convergent/divergent evolution
+
+So now we have a list of enriched genes for each condition. We can see that, for example, "H0403_00778" occurs in the significant hit lists of multiple conditions. Now we would like to determine whether a significant gene popping up in these multiple conditions happens more than would be expected by chance.
+
+## Jaccard Index for convergence between treatments
+
+We follow the same approach as in [@shoemaker2021] where we simulate the process of randomly sampling $n = 100,000$ draws of significant gene set pairs $(I_{i},I_{j})$ from every species $\times$ treatment pair where the probability of sampling a gene is proportional to its length. We then calculate the Jaccard index
+
+$$J(I_{i},I_{j}) = \left|I_{i} \cap I_{j}\right|/ \left|I_{i} \cup I_{j}\right|$$ for each null simulation $J(I_{i},I_{j})_{null}$ and calculate an empirical $P$ value for the number of null simulation results exceeding the observed $J(I_{i},I_{j})_{observed}$ to determine whether the intersection of the set of genes was greater (convergence) or less than (divergence) expected by chance.
+
+### HAMBI_1287
+
+```{r}
+#| eval: false
+#| echo: true
+H1287_cvdv <- resample_jaccard(100000,
+ prep4jaccard_resample(output_gene_table_filt, degentab,
+ "HAMBI_1287", prey_history, predator_history)
+ )
+
+write_rds(H1287_cvdv, here::here(data, "jaccard_H1287_cvdv.rds"))
+```
+
+```{r}
+#| eval: true
+#| echo: false
+read_rds(here::here(data, "jaccard_H1287_cvdv.rds"))
+```
+
+### HAMBI_1972
+
+```{r}
+#| eval: false
+#| echo: true
+H1972_cvdv <- resample_jaccard(100000,
+ prep4jaccard_resample(output_gene_table_filt, degentab,
+ "HAMBI_1972", prey_history, predator_history)
+ )
+
+write_rds(H1972_cvdv, here::here(data, "jaccard_H1972_cvdv.rds"))
+```
+
+```{r}
+#| eval: true
+#| echo: false
+read_rds(here::here(data, "jaccard_H1972_cvdv.rds"))
+```
+
+### HAMBI_1977
+
+```{r}
+#| eval: false
+#| echo: true
+H1977_cvdv <- resample_jaccard(100000,
+ prep4jaccard_resample(output_gene_table_filt, degentab,
+ "HAMBI_1977", prey_history, predator_history)
+ )
+
+write_rds(H1977_cvdv, here::here(data, "jaccard_H1977_cvdv.rds"))
+```
+
+```{r}
+#| eval: true
+#| echo: false
+read_rds(here::here(data, "jaccard_H1977_cvdv.rds"))
+```
+
+### HAMBI_2659
+
+```{r}
+#| eval: false
+#| echo: true
+H2659_cvdv <- resample_jaccard(100000,
+ prep4jaccard_resample(output_gene_table_filt, degentab,
+ "HAMBI_2659", prey_history, predator_history)
+ )
+
+write_rds(H2659_cvdv, here::here(data, "jaccard_H2659_cvdv.rds"))
+```
+
+```{r}
+#| eval: true
+#| echo: false
+read_rds(here::here(data, "jaccard_H2659_cvdv.rds"))
+```
+
+## Enrichment of individual COG categories
+
+So now we have a list of enriched genes for each species. We can see that, for example, some gene functions pop up multiple times (like transcription factors and sensory kinases). Are these different functions popping up across species more that we would expect by chance? We'll try and test this using a simulation based approach
+
+```{r}
+#| output: false
+# read back the formatted eggnog data
+eggnog <- read_tsv(here::here(shared, "HAMBI_all_eggnog_formatted.tsv.xz"))
+```
+
+```{r}
+cog_description <- tibble::tribble(
+ ~COG_category_single, ~COG_category_long,
+ "J", "J - Translation, ribosomal structure and biogenesis",
+ "A", "A - RNA processing and modification",
+ "K", "K – Transcription",
+ "L", "L - Replication, recombination and repair",
+ "B", "B - Chromatin structure and dynamics",
+ "D", "D - Cell cycle control, cell division, chromosome partitioning",
+ "Y", "Y - Nuclear structure",
+ "V", "V - Defense mechanisms",
+ "T", "T - Signal transduction mechanisms",
+ "M", "M - Cell wall/membrane/envelope biogenesis",
+ "N", "N - Cell motility",
+ "Z", "Z – Cytoskeleton",
+ "W", "W - Extracellular structures",
+ "U", "U - Intracellular trafficking, secretion, and vesicular transport",
+ "O", "O - Posttranslational modification, protein turnover, chaperones",
+ "X", "X - Mobilome: prophages, transposons",
+ "C", "C - Energy production and conversion",
+ "G", "G - Carbohydrate transport and metabolism",
+ "E", "E - Amino acid transport and metabolism",
+ "F", "F - Nucleotide transport and metabolism",
+ "H", "H - Coenzyme transport and metabolism",
+ "I", "I - Lipid transport and metabolism",
+ "P", "P - Inorganic ion transport and metabolism",
+ "Q", "Q - Secondary metabolites biosynthesis, transport and catabolism",
+ "R", "R - General function prediction only",
+ "S", "S - Function unknown"
+ )
+```
+
+save a copy of significantly parallel genes as a table
+
+```{r}
+output_gene_table_filt_cog <- left_join(output_gene_table_filt, eggnog) %>%
+ rename(prokka_annotation = product) %>%
+ relocate(strainID, prey_history, predator_history, COG_category_long, Preferred_name, Description, prokka_annotation)
+
+write_tsv(output_gene_table_filt_cog, here::here(data, "enriched_parallel_genes.tsv"))
+
+output_gene_table_filt_cog
+```
+
+Calculate the probability of each gene (ns_sites/total_ns_sites) and link to COG categories
+
+```{r}
+eggnog_nslen <- left_join(eggnog, degentab,
+ by = join_by(strainID, locus_tag)) %>%
+ dplyr::select(strainID, locus_tag, COG_category_single, ns_length) %>%
+ filter(!is.na(COG_category_single)) %>%
+ group_by(strainID) %>%
+ mutate(p = ns_length/sum(ns_length),
+ cog_genes = n()) %>%
+ ungroup()
+```
+
+### Jensen Shannon Divergence
+
+This function draws n genes from each species with a probability proportional to gene length, where n = the number of significant parallel genes observed from the data. We repeat this simulation 10 000 times and below we check if 1) the Jensen Shannon divergence between the distribution of the COG categories drawn randomly and the background (i.e., the distribution of COG categories in the whole genome) is greater than the Jensen Shannon divergence between the background and the observed distribution of COG categories in significantly parallel genes. The idea is that if some COG categories are enriched/depleted in the parallel genes, then the JSD divergence between the observed and the background should be larger than the JSD between the random sampled genes (sampling probability based only on gene length) and the background. We calculate an empirical P value by "counting" how many times the simulated JSD exceeds the observed.
+
+```{r}
+#| eval: false
+#| echo: true
+gene_nest <- output_gene_table_filt %>%
+ summarize(nsiggenes = n(), .by = c(strainID, prey_history, predator_history)) %>%
+ left_join(eggnog_nslen,
+ by = join_by(strainID),
+ relationship = "many-to-many") %>%
+ nest(data = -c(strainID, prey_history, predator_history))
+
+gene_draw <- function(){
+ gene_nest %>%
+ mutate(locus_tag = map(
+ data,
+ \(x) sample(
+ x = x$locus_tag,
+ size = unique(x$nsiggenes),
+ prob = x$p,
+ replace = FALSE
+ )
+ )) %>%
+ dplyr::select(-data) %>%
+ unnest(cols = locus_tag)
+}
+
+# take 10 000 random draws of the genes. Use a seed for reproducibility
+withr::with_seed(12367,
+ genes_simulated <- map(1:10000, ~gene_draw(), .progress=TRUE) %>%
+ list_rbind(names_to = "id"))
+
+# save the output so we don't need to run again
+write_rds(genes_simulated, here::here(data, "genes_simulated_nomobile.rds"))
+```
+
+Some data formatting
+
+```{r}
+# read data back
+genes_simulated <- read_rds(here::here(data, "genes_simulated_nomobile.rds"))
+
+# map the locus_tags to COG categories for the simulated
+# gene draws
+COGs_distribution_simulated <- genes_simulated %>%
+ left_join(eggnog_nslen, by = join_by(strainID, locus_tag)) %>%
+ group_by(id, strainID, prey_history, predator_history) %>%
+ count(COG_category_single, name = "nsim") %>%
+ ungroup()
+
+# get the discrete distribution of COGs in the significant parallel genes
+COGs_distribution_observed <- left_join(output_gene_table_filt, eggnog,
+ by = join_by(locus_tag, strainID)) %>%
+ group_by(strainID, prey_history, predator_history) %>%
+ count(COG_category_single, name = "nobs")
+
+# get the discrete distribution of COGs across the whole genome of all species
+# this is the null case (i.e. the background that we would expect if there was no
+# enrichment)
+COGs_distribution_nullexpectation <- eggnog %>%
+ group_by(strainID) %>%
+ count(COG_category_single, name = "nnull")
+```
+
+Now we calculate the Jensen Shannon divergence between the background and the simulated data to estimate an empirical P value
+
+```{r}
+# function to calculate the Jensen Shannon divergene of a pair of
+# distributions
+JSD_pair <- function(p, q){
+ m <- 0.5 * (p + q)
+ JS <- 0.5 * (sum(p * log(p/m)) + sum(q * log(q/m)))
+ return(JS)
+}
+
+# Calculate the Jensen Shannon divergence between the COG distribution observed in
+# enriched genes and the COG distribution in the genomic background
+JSD_observed <- left_join(COGs_distribution_observed, COGs_distribution_nullexpectation,
+ by = join_by(COG_category_single)) %>%
+ group_by(prey_history, predator_history) %>%
+ mutate(p = nobs/sum(nobs),
+ q = nnull/sum(nnull)) %>%
+ summarize(JSD = JSD_pair(p, q))
+
+# now calculate JSD between the simulated draws and the backgound distribution
+JSD_simulated <- left_join(COGs_distribution_simulated,
+ COGs_distribution_nullexpectation,
+ by = join_by(strainID, COG_category_single)) %>%
+ # drops from calculation categories that aren't in both the observed data
+ # and the resampled data. Need to do this because JSD can't handle zero values
+ drop_na() %>%
+ group_by(id, strainID, prey_history, predator_history) %>%
+ mutate(p = nsim/sum(nsim),
+ q = nnull/sum(nnull)) %>%
+ summarize(JSD = JSD_pair(p, q))
+
+# proportion of JSD measures between simulated and the background that are greater
+# than the JSD between the observed and the background
+global_p <- left_join(JSD_simulated, JSD_observed, by = join_by(prey_history, predator_history)) %>%
+ mutate(ngt = JSD.x >= JSD.y) %>%
+ group_by(strainID, prey_history, predator_history) %>%
+ summarize(p = sum(ngt)/n())
+
+global_p
+```
+
+The overall distribution of COG categories in the parallel genes for each genome in each treatment combination is not significantly different than in the rest of the genome in 9/10 cases. However, for HAMBI_1972 there does seem to be a difference in the distribution of COG categories in enriched genes relative to the genome background.
+
+### Enrichment of individual COG categories
+
+OK so the overall distribution of COG categories in the parallel genes is not significantly different than in the rest of the genome - \~40% of the time the JSD between the COG distribution in a random sample and the background is greater than the observed COG distribution and the background.
+
+#### Hypergeometric Test
+
+We can test enrichment in individual COG categories in individual species using the hypergeometric test. We will correct for multiple testings by controlling the False Discovery Rate (FDR) using q-values[@storey2003a] and choosing a maximum FDR threshold of 5%.
+
+```{r}
+library(qvalue)
+
+par2test <- left_join(output_gene_table_filt, eggnog,
+ by = join_by(locus_tag, strainID)) %>%
+ filter(!is.na(COG_category)) %>%
+ group_by(strainID, prey_history, predator_history) %>%
+ count(COG_category_single, name = "n_cog_par") %>%
+ mutate(n_par = sum(n_cog_par))
+
+back2test <- eggnog %>%
+ filter(!is.na(COG_category)) %>%
+ group_by(strainID) %>%
+ count(COG_category_single, name = "n_cog_background") %>%
+ mutate(n_background = sum(n_cog_background))
+
+phypres <- left_join(par2test, back2test, by = join_by(strainID, COG_category_single)) %>%
+ mutate(p_enrich = enricher(n_background, n_par, n_cog_background, n_cog_par, over=TRUE),
+ p_deplete = enricher(n_background, n_par, n_cog_background, n_cog_par, over=FALSE)) %>%
+ mutate(p_enrich_adj = p.adjust(p_enrich, method = "fdr"),
+ p_deplete_adj = p.adjust(p_deplete, method = "fdr"))
+
+qenrich <- qvalue_truncp(phypres$p_enrich)
+qdeplete <- qvalue_truncp(phypres$p_deplete)
+
+phypres$q_enrich <- qenrich$qvalues
+phypres$q_deplete <- qdeplete$qvalues
+
+phypres %>%
+ relocate(p_enrich, q_enrich, p_enrich_adj) %>%
+ arrange(p_enrich)
+```
+
+#### Nonparametric simulation
+
+We can also look at different COG categories by simply counting the number of random samplings that have a COG count greater/less than that the observed COG count. Since we resampled the same number of genes as in the observed we can compare the two directly. Below we do this integrated over all species in the community to see if there was any pathway enriched across all species.
+
+```{r}
+left_join(COGs_distribution_simulated, COGs_distribution_observed,
+ by = join_by(COG_category_single, strainID, prey_history, predator_history)) %>%
+ drop_na() %>%
+ mutate(gt = if_else(nsim >= nobs, 1, 0)) %>%
+ group_by(COG_category_single, strainID, prey_history, predator_history) %>%
+ summarize(p = sum(gt)/n()) %>%
+ arrange(p) %>%
+ mutate(p_adj = p.adjust(p, method = "fdr"))
+```
+
+### COG Distribution Plot
+
+```{r}
+cog_pmut <- left_join(output_gene_table_filt, eggnog,
+ by = join_by(locus_tag, strainID)) %>%
+ count(COG_category_single) %>%
+ mutate(f = n/sum(n),
+ n = n*1000,
+ type = "parallel_mutations")
+
+cog_everything <- eggnog %>%
+ count(COG_category_single) %>%
+ mutate(f = n/sum(n),
+ type = "all_genes")
+
+pcogs <- bind_rows(cog_pmut, cog_everything) %>%
+ drop_na() %>%
+ ggplot() +
+ geom_bar(aes(x = fct_reorder(COG_category_single, f, .desc = F), y=n, fill = type), stat="identity",
+ position = position_dodge( preserve = "total")) +
+ scale_fill_brewer(palette = "Set1") +
+ labs(x = "COG Category", y = "Fraction genes in COG category", fill = "Subset") +
+ scale_y_continuous(
+ "N all genes",
+ sec.axis = sec_axis(~ . * 1/1000, name = "N parallel genes")
+ ) +
+ theme_bw() +
+ theme(
+ panel.grid = element_blank(),
+ panel.background = element_blank(),
+ legend.position = "bottom"
+ )
+```
+
+::: {#fig-05}
+
+```{r}
+#| fig.width: 8
+#| fig.height: 6
+#| echo: false
+pcogs
+```
+
+Fraction of genes in each Cluster of Orthologous Groups (COG) category for genes with a COG annotation (red bars, left y-axis) and only genes exhibiting significant genomic parallelism (blue bars, right y-axis) integrated over HAMBI species from metagenomes (i.e., species exceeding the coverage threshold).
+:::
+
+### Species present in the metagenomes
+
+```{r}
+# species detected in the metagenomes
+mgsp <- c("HAMBI_2659", "HAMBI_1972", "HAMBI_1977", "HAMBI_1287")
+
+cog_pmut_mgsp <- left_join(output_gene_table_filt, eggnog,
+ by = join_by(locus_tag, strainID)) %>%
+ filter(strainID %in% mgsp) %>%
+ group_by(strainID, prey_history, predator_history) %>%
+ count(COG_category_single, name = "tot") %>%
+ # scaling factor to make ploting on 2 axes possible
+ mutate(n = tot*250,
+ f = n/sum(n),
+ type = "parallel_mutations")
+
+cog_everything_mgsp <- eggnog %>%
+ filter(strainID %in% mgsp) %>%
+ group_by(strainID) %>%
+ count(COG_category_single) %>%
+ mutate(f = n/sum(n),
+ type = "all_genes") %>%
+ expand_grid(prey_history = c("anc", "evo"), predator_history = c("anc", "evo"))
+
+pcogs_mgsp <- semi_join(cog_everything_mgsp, cog_pmut_mgsp,
+ by = join_by(strainID, prey_history, predator_history)) %>%
+ bind_rows(cog_pmut_mgsp) %>%
+ filter(!is.na(COG_category_single)) %>%
+ mutate(treat = paste0("prey:", prey_history, " | pred:", predator_history)) %>%
+ ggplot() +
+ geom_bar(aes(x = fct_reorder(COG_category_single, n, .desc = F), y=n, fill = type), stat="identity",
+ position = position_dodge( preserve = "total")) +
+ scale_fill_brewer(palette = "Set1") +
+ geom_point(data = tibble(x = c("U", "P", "D", "D", "P", "M"),
+ y = c(900, 900, 900, 900, 900, 900),
+ strainID = c("HAMBI_2659", "HAMBI_2659", "HAMBI_1972", "HAMBI_1972", "HAMBI_2659" ,"HAMBI_1972"),
+ prey_history = c("evo", "evo", "evo", "evo", "evo", "evo"),
+ predator_history = c("evo", "evo", "anc", "evo", "anc", "evo")),
+ aes(x = x, y = y), shape = 8) +
+ facet_grid(treat ~ strainID) +
+ labs(x = "COG Category", y = "Fraction genes in COG category", fill = "Subset") +
+ scale_y_continuous(
+ "N all genes",
+ sec.axis = sec_axis(~ . * 1/250, name = "N parallel genes")
+ ) +
+ theme_bw() +
+ theme(
+ panel.grid = element_blank(),
+ strip.background = element_blank(),
+ panel.background = element_blank(),
+ legend.position = "bottom"
+ )
+```
+
+::: {#fig-06}
+```{r}
+#| fig.width: 8
+#| fig.height: 8
+#| echo: false
+pcogs_mgsp
+```
+
+Fraction of genes in each Cluster of Orthologous Groups (COG) category for genes with a COG annotation (red bars, left y-axis) and only genes exhibiting significant genomic parallelism (blue bars, right y-axis) for the four most abundant HAMBI genomes. COG categories for species $\times$ treatments combinationm are denoted with \* if they are significantly enriched in parallel mutations relative to the genomic background. See statistical results in table below.
+:::
+
+```{r}
+ggsave(here::here(figs, "COG_enrich_mgs.svg"), pcogs_mgsp, width=7, height=7,
+ units="in", device="svg")
+```
+
+| strainID | prey history | predator history | COG category | q value |
+|--------------|--------------|----------------|--------------|--------------|
+| HAMBI_1972 | Evolved | Ancestral | D - Cell cycle control, cell division, chromosome partitioning | 0.019 |
+| HAMBI_1972 | Evolved | Evolved | D - Cell cycle control, cell division, chromosome partitioning | 0.026 |
+| HAMBI_1972 | Evolved | Evolved | M - Cell wall/membrane/envelope biogenesis | 0.047 |
+| HAMBI_2659 | Evolved | Evolved | U - Intracellular trafficking, secretion, and vesicular transport | 0.026 |
+| HAMBI_2659 | Evolved | Ancestral | P - Inorganic ion transport and metabolism | 0.037 |
+| HAMBI_2659 | Evolved | Evolved | P - Inorganic ion transport and metabolism | 0.037 |
diff --git a/R/metagenome/03_analyze_metagenome_variant_timeseries.qmd b/R/metagenome/03_analyze_metagenome_variant_timeseries.qmd
new file mode 100644
index 0000000..96ca98b
--- /dev/null
+++ b/R/metagenome/03_analyze_metagenome_variant_timeseries.qmd
@@ -0,0 +1,461 @@
+---
+title: "Plot variant time series"
+author: "Shane Hogle"
+date: today
+abstract: "This notebook reads in the formatted metagenome variant frequencies and annotations. It makes some plots of the variant trajectories through time."
+---
+
+# Setup
+
+Libraries and global variables
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+library(tidyverse)
+library(here)
+library(fs)
+library(fishualize)
+library(withr)
+library(scales)
+library(patchwork)
+library(Polychrome)
+
+source(here::here("R", "utils_generic.R"))
+```
+
+Set up some directories
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+data_raw <- here::here("_data_raw", "metagenome")
+data <- here::here("data", "metagenome")
+shared <- here::here("_data_raw", "shared")
+figs <- here::here("figs", "metagenome")
+
+# make processed data directory if it doesn't exist
+fs::dir_create(data)
+fs::dir_create(figs)
+```
+
+# Read data
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+
+# these were already filtered in the last step
+mgvars <- read_tsv(here::here(data, "metagenome_variant_timeseries.tsv"))
+degentab <- read_rds(here::here(shared, "annotations_codon_degeneracy.rds"))
+genome_len <- read_tsv(here::here(shared, "HAMBI_genome_len.tsv"))
+# annotations
+annotations <- read_rds(here::here(shared, "annotations_codon_degeneracy.rds"))
+# significantly parallel genes
+par_genes <- read_tsv(here::here(data, "enriched_parallel_genes.tsv"))
+
+```
+
+
+```{r}
+cog_description <- tibble::tribble(
+ ~COG_category_single, ~COG_category_long,
+ "J", "J - Translation, ribosomal structure and biogenesis",
+ "A", "A - RNA processing and modification",
+ "K", "K – Transcription",
+ "L", "L - Replication, recombination and repair",
+ "B", "B - Chromatin structure and dynamics",
+ "D", "D - Cell cycle control, cell division, chromosome partitioning",
+ "Y", "Y - Nuclear structure",
+ "V", "V - Defense mechanisms",
+ "T", "T - Signal transduction mechanisms",
+ "M", "M - Cell wall/membrane/envelope biogenesis",
+ "N", "N - Cell motility",
+ "Z", "Z – Cytoskeleton",
+ "W", "W - Extracellular structures",
+ "U", "U - Intracellular trafficking, secretion, and vesicular transport",
+ "O", "O - Posttranslational modification, protein turnover, chaperones",
+ "X", "X - Mobilome: prophages, transposons",
+ "C", "C - Energy production and conversion",
+ "G", "G - Carbohydrate transport and metabolism",
+ "E", "E - Amino acid transport and metabolism",
+ "F", "F - Nucleotide transport and metabolism",
+ "H", "H - Coenzyme transport and metabolism",
+ "I", "I - Lipid transport and metabolism",
+ "P", "P - Inorganic ion transport and metabolism",
+ "Q", "Q - Secondary metabolites biosynthesis, transport and catabolism",
+ "R", "R - General function prediction only",
+ "S", "S - Function unknown"
+ )
+
+withr::with_seed(12367,
+ cogpal <- unname(createPalette(length(unique(cog_description$COG_category_single)), c("#F3874AFF", "#FCD125FF"), M=5000))
+)
+names(cogpal) <- cog_description$COG_category_long
+```
+
+# Clustering
+
+~~One clustering approach is with the kml package - [see tutorial here](https://github.com/keithmcnulty/longitudinal_clustering). Another option would be to use the `curveRep` function from `Hmisc` - [see more here](https://stats.stackexchange.com/a/17804).~~
+
+Classic old hierarchical clustering from hclust works best and is easiest to use
+
+```{r}
+# create unique group id as a convenience for later plotting
+df_grouped <- mgvars %>%
+ #filter(!str_detect(effect, "intergenic|intragenic|synonymous|fusion")) %>%
+ #filter(!str_detect(impact, "MODIFIER")) %>%
+ group_by(strainID, chrom, pos, ref, alt, replicate, prey_history, predator_history) %>%
+ mutate(group_id = cur_group_id()) %>%
+ relocate(group_id) %>%
+ ungroup()
+
+# format the grouped dataframe in a way that can be plotted
+df2clust <- df_grouped %>%
+ mutate(day = paste0("day", time_days)) %>%
+ select(group_id, day, freq_alt_complete) %>%
+ pivot_wider(names_from = "day", values_from = "freq_alt_complete") %>%
+ as.data.frame() %>%
+ column_to_rownames(var = "group_id")
+
+# scale the dataframe for clustering
+df2clust_scaled <- scale(df2clust)
+
+# perform the hierarchcical clustering using euclidean distance and Ward's D
+hc <- hclust(dist(df2clust_scaled, method = "euclidean"), method = "ward.D2" )
+
+# get order of the clustered observations. The order is for the groups
+ord <- hc$order
+
+# get cluster membership. Again memebership is for the groups
+myclust <- cutree(hc, k = 6)
+
+df2plot <- df_grouped %>%
+ mutate(cluster = myclust[group_id],
+ order = ord[group_id]) %>%
+ # convert group_id into a factor that is ordered by the hierarchical clustering
+ # done above
+ mutate(group_id = factor(group_id, levels = ord),
+ cluster = factor(cluster)) %>%
+ mutate(day = factor(time_days),
+ pos = factor(pos),
+ mylab = interaction(replicate, prey_history, predator_history, paste0(locus_tag, "_", pos)),
+ strainID2 = case_when(strainID == "HAMBI_0403" ~ 5,
+ strainID == "HAMBI_1287" ~ 4,
+ strainID == "HAMBI_1972" ~ 2,
+ strainID == "HAMBI_1977" ~ 3,
+ strainID == "HAMBI_2659" ~ 1)) %>%
+ mutate(mylab2 = fct_reorder(group_id, strainID2)) %>%
+ relocate(mylab2, order, strainID2)
+```
+
+# Main text figure
+
+Idea is to have a one heatmap for the alt allele frequencies for every detected mutation that passes our thresholds and then have some horiztonal bars which designate the different treatment combinations, the longitudinal clusters, and the species identieis stacked on top. Then below this we will have some simple line plots show the highly parallel mutations identified in the prior analysis.
+
+## Heatmap all species together
+
+```{r}
+blank_theme01 <- function(){
+ theme(
+ panel.grid = element_blank(),
+ panel.border = element_blank(),
+ panel.background = element_blank(),
+ axis.text = element_blank(),
+ axis.ticks = element_blank(),
+ legend.title = element_blank(),
+ plot.margin = unit(c(0,0,0,0), "cm"),
+ strip.background = element_blank(),
+ strip.text = element_blank()
+ )
+}
+
+blank_theme02 <- function(){
+ theme(
+ panel.grid = element_blank(),
+ panel.border = element_blank(),
+ panel.background = element_blank(),
+ axis.ticks = element_blank(),
+ legend.title = element_blank(),
+ strip.background = element_blank(),
+ strip.text = element_blank()
+ )
+}
+
+phoriz_bar <- function(df, fillvar, mypal){
+ # only take the first time point so to make one bar
+ filter(df, day == 0) %>%
+ ggplot(aes(y = day, x=mylab2, fill = {{ fillvar }})) +
+ geom_tile() +
+ labs(x = NULL, y = NULL) +
+ facet_wrap(prey_history ~ predator_history, scales = "free", nrow = 1) +
+ scale_fill_manual(values = {{ mypal }}) +
+ blank_theme01()
+}
+
+pheat_nolb <- function(df, fillvar){
+ ggplot(df, aes(y = day, x=mylab2, fill = {{ fillvar }} )) +
+ geom_tile() +
+ labs(x = NULL, y = NULL) +
+ facet_wrap(prey_history ~ predator_history, scales = "free", nrow = 1) +
+ scale_fill_viridis_c(limits = c(0, 1), trans = "sqrt") +
+ scale_x_discrete(guide = guide_axis(angle = 90)) +
+ blank_theme01()
+}
+
+pheat_labs <- function(df, fillvar){
+ ggplot(df, aes(y = day, x=mylab2, fill = {{ fillvar }})) +
+ geom_tile() +
+ labs(x = NULL, y = NULL) +
+ facet_wrap(prey_history ~ predator_history, scales = "free", nrow = 1) +
+ scale_fill_viridis_c(limits = c(0, 1), trans = "sqrt") +
+ scale_x_discrete(guide = guide_axis(angle = 90)) +
+ blank_theme02()
+}
+```
+
+
+### All alleles
+```{r}
+#| fig.width: 8
+#| fig.height: 8
+#clupal <- unname(createPalette(length(unique(myclust)), c("#F3874AFF", "#FCD125FF"), M=5000))
+clupal <- gray.colors(length(unique(myclust)), start = 0.1, end = 0.9, gamma = 2.2, 1, rev = TRUE)
+names(clupal) <- unique(myclust)
+
+pheat_all <- phoriz_bar(filter(df2plot, strainID == "HAMBI_2659"), cluster, clupal) +
+ pheat_nolb(filter(df2plot, strainID == "HAMBI_2659"), freq_alt_complete) +
+ phoriz_bar(filter(df2plot, strainID == "HAMBI_1972"), cluster, clupal) +
+ pheat_nolb(filter(df2plot, strainID == "HAMBI_1972"), freq_alt_complete) +
+ phoriz_bar(filter(df2plot, strainID == "HAMBI_1977"), cluster, clupal) +
+ pheat_nolb(filter(df2plot, strainID == "HAMBI_1977"), freq_alt_complete) +
+ phoriz_bar(filter(df2plot, strainID == "HAMBI_1287"), cluster, clupal) +
+ pheat_nolb(filter(df2plot, strainID == "HAMBI_1287"), freq_alt_complete) +
+ phoriz_bar(filter(df2plot, strainID == "HAMBI_0403"), cluster, clupal) +
+ pheat_nolb(filter(df2plot, strainID == "HAMBI_0403"), freq_alt_complete) +
+ plot_layout(ncol = 1, nrow = 10,
+ heights = c(0.25, 1, 0.25, 1, 0.25, 1, 0.25, 1, 0.25, 1),
+ guides = "collect")
+
+pheat_all
+```
+
+#### Save
+```{r}
+ggsave(here::here(figs, "heatmap_all_alleles.svg"), pheat_all, width=11, height=4, units="in",
+ device="svg")
+```
+
+### Parallel genes
+
+```{r}
+#| fig.width: 8
+#| fig.height: 8
+#| warning: false
+
+pdfr_evo <- left_join(par_genes, df2plot,
+ by = join_by(locus_tag, strainID, prey_history, predator_history, chrom)) %>%
+ filter(prey_history == "evo") %>%
+ # some manual renaming
+ mutate(gene = case_when(locus_tag == "H1287_02172" ~ "yddV_2",
+ locus_tag == "H1977_02612" ~ "dppA/oppA",
+ locus_tag == "H1972_00299" ~ "PAP2_like",
+ locus_tag == "H1972_02826" ~ "ynjC",
+ locus_tag == "H1972_00671" ~ "yqaA",
+ locus_tag == "H1972_00421" ~ "hyp",
+ locus_tag == "H1972_02723" ~ "yifE",
+ locus_tag == "H2659_01789" ~ "hyp",
+ TRUE ~ gene)) %>%
+ mutate(gene_lab = if_else(is.na(gene), Preferred_name, gene)) %>%
+ mutate(gene_lab = if_else(gene_lab == "-", locus_tag, gene_lab)) %>%
+ mutate(gene_lab = paste0(gene_lab, " | ", replicate, " | ", cluster, "|", group_id)) %>%
+ arrange(COG_category_long, gene_lab) %>%
+ mutate(mylab2 = factor(gene_lab, unique(gene_lab)))
+
+pdfr_anc <- left_join(par_genes, df2plot,
+ by = join_by(locus_tag, strainID, prey_history, predator_history, chrom)) %>%
+ filter(prey_history == "anc") %>%
+ mutate(gene = case_when(locus_tag == "H1972_02826" ~ "ynjC",
+ locus_tag == "H1972_00421" ~ "hyp",
+ TRUE ~ gene)) %>%
+ mutate(gene_lab = if_else(is.na(gene), Preferred_name, gene)) %>%
+ mutate(gene_lab = if_else(gene_lab == "-", locus_tag, gene_lab)) %>%
+ mutate(gene_lab = paste0(gene_lab, " | ", replicate, " | ", cluster, "|", group_id)) %>%
+ arrange(COG_category_long, gene_lab) %>%
+ mutate(mylab2 = factor(gene_lab, unique(gene_lab)))
+
+# evolved
+pheat_par <- phoriz_bar(filter(pdfr_evo, strainID == "HAMBI_2659"), COG_category_long, cogpal) +
+ pheat_labs(filter(pdfr_evo, strainID == "HAMBI_2659"), freq_alt_complete) +
+ # evolved 1972
+ phoriz_bar(filter(pdfr_evo, strainID == "HAMBI_1972"), COG_category_long, cogpal) +
+ pheat_labs(filter(pdfr_evo, strainID == "HAMBI_1972"), freq_alt_complete) +
+ # ancestral 1972 - the only species that has significantly parallel genes in the ancestral treatment
+ phoriz_bar(filter(pdfr_anc, strainID == "HAMBI_1972"), COG_category_long, cogpal) +
+ pheat_labs(filter(pdfr_anc, strainID == "HAMBI_1972"), freq_alt_complete) +
+ # evolved 1977
+ phoriz_bar(filter(pdfr_evo, strainID == "HAMBI_1977"), COG_category_long, cogpal) +
+ pheat_labs(filter(pdfr_evo, strainID == "HAMBI_1977"), freq_alt_complete) +
+ # evolved 1287
+ phoriz_bar(filter(pdfr_evo, strainID == "HAMBI_1287"), COG_category_long, cogpal) +
+ pheat_labs(filter(pdfr_evo, strainID == "HAMBI_1287"), freq_alt_complete) +
+ plot_layout(ncol = 1, nrow = 10,
+ heights = c(0.25, 1, 0.25, 1, 0.25, 1, 0.25, 1, 0.25, 1),
+ guides = "collect")
+
+pheat_par
+```
+
+#### Save
+```{r}
+ggsave(here::here(figs, "heatmap_par_alleles.svg"), pheat_par, width=12, height=10, units="in",
+ device="svg")
+```
+
+## Individual allelle trajectories
+
+Read data from parallelism analysis
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+```
+
+Function for plotting trajectories for indivdidual genes
+
+```{r}
+plotpargenes <- function(strainID){
+ pdfr <- left_join(par_genes, df2plot,
+ by = join_by(locus_tag, strainID, prey_history, predator_history, chrom)) %>%
+ # to reduce size only include genes that are mutated in at least two of the replicates
+ filter(n_replicate >=2) %>%
+ # name the genes for plotting
+ mutate(gene_lab = if_else(is.na(gene), Preferred_name, gene),
+ treat = paste0("prey:", prey_history, " | pred:", predator_history)) %>%
+ mutate(gene_lab = if_else(gene_lab == "-", Description, gene_lab)) %>%
+ filter(strainID == {{ strainID }})
+
+ # make random color set to help differentiate alleles
+ mypal <- unname(createPalette(length(unique(pdfr$mylab2)), c("#F3874AFF", "#FCD125FF"), M=5000))
+
+ ggplot() +
+ geom_line(data=pdfr[!is.na(pdfr$freq_alt_complete), ],
+ aes(x=time_days, y=freq_alt_complete,
+ group = mylab2,
+ color = hgvs_p)) +
+ geom_point(data = pdfr,
+ aes(x=time_days, y=freq_alt_complete, shape = replicate),
+ alpha = 1) +
+ guides(color = "none") +
+ scale_color_manual(values = mypal) +
+ facet_wrap(gene_lab ~ treat) +
+ labs(x = "Days", y = "Allele frequency") +
+ theme_bw() +
+ theme(
+ legend.position = "bottom",
+ strip.placement = 'outside',
+ strip.background = element_blank(),
+ panel.grid = element_blank())
+}
+```
+
+### HAMBI_1287
+
+::: {#fig-01}
+```{r}
+#| fig.width: 8
+#| fig.height: 8
+#| echo: false
+plotpargenes("HAMBI_1287")
+```
+
+Nonsynonymous variant frequencies (vertical axis) over time (horizontal axis) for combinations of individual genes (title bar top row) and treatment combinations (title bar bottom row). Point shape denotes the biological replicate from the treatment and line colors indicate the amino acid changing variant (some genes have multiple non-synonymous variants at different positions). Focal genes are from *Citrobacter koserii* HAMBI_1287 and are those exhibiting significant parallelism for the duration of the experiment.
+:::
+
+### HAMBI_1972
+
+::: {#fig-02}
+```{r}
+#| fig.width: 8
+#| fig.height: 7
+#| echo: false
+plotpargenes("HAMBI_1972")
+```
+
+As in @fig-01 but for *Aeromonas caviae* HAMBI_1972.
+:::
+
+### HAMBI_1977
+
+::: {#fig-03}
+```{r}
+#| fig.width: 8
+#| fig.height: 8
+#| echo: false
+plotpargenes("HAMBI_1977")
+```
+
+As in @fig-01 but for *Pseudomonas chlororaphis* HAMBI_1977.
+:::
+
+### HAMBI_2659
+
+::: {#fig-04}
+
+```{r}
+#| fig.width: 8
+#| fig.height: 8
+#| echo: false
+plotpargenes("HAMBI_2659")
+```
+
+As in @fig-01 but for *Stenotrophomonas maltophilia* HAMBI_2659.
+:::
+
+### Selection of parallel genes to plot
+
+::: {#fig-05}
+
+```{r}
+#| fig.width: 8
+#| fig.height: 8
+
+pdfr <- par_genes %>%
+ filter(n_replicate >= 2) %>%
+ arrange(desc(gene_multiplicity_m_i)) %>%
+ # take top 20 highest multiplicitiies
+ slice(1:20) %>%
+ left_join(df2plot, by = join_by(locus_tag, strainID, prey_history, predator_history, chrom)) %>%
+ # name the genes for plotting
+ mutate(gene_lab = if_else(is.na(gene), Preferred_name, gene),
+ treat = paste0("prey:", prey_history, " | pred:", predator_history)) %>%
+ mutate(gene_lab = if_else(gene_lab == "-", Description, gene_lab))
+
+
+mypal <- unname(createPalette(length(unique(pdfr$mylab2)), c("#F3874AFF", "#FCD125FF"), M=5000))
+
+ggplot() +
+ geom_line(data=pdfr[!is.na(pdfr$freq_alt_complete), ],
+ aes(x=time_days, y=freq_alt_complete,
+ group = mylab2,
+ color = hgvs_p)) +
+ geom_point(data = pdfr,
+ aes(x=time_days, y=freq_alt_complete, shape = replicate),
+ alpha = 1) +
+ guides(color = "none") +
+ scale_color_manual(values = mypal) +
+ facet_wrap(gene_lab ~ treat) +
+ labs(x = "Days", y = "Allele frequency") +
+ theme_bw() +
+ theme(
+ legend.position = "bottom",
+ strip.placement = 'outside',
+ strip.background = element_blank(),
+ panel.grid = element_blank())
+```
+As in @fig-01 but for the genes with the top 20 higest gene multiplicities across HAMBI_1287, HAMBI_1972, HAMBI_1977, and HAMBI_2659.
+:::
diff --git a/R/metagenome/04_mutation_dynamics.qmd b/R/metagenome/04_mutation_dynamics.qmd
new file mode 100644
index 0000000..d8c917e
--- /dev/null
+++ b/R/metagenome/04_mutation_dynamics.qmd
@@ -0,0 +1,956 @@
+---
+title: "Mutational dynamics"
+author: "Shane Hogle"
+date: today
+link-citations: true
+bibliography: references.bib
+abstract: "This notebook reads in the formatted metagenome variant frequencies and annotations. It makes some plots and does some basic analysis looking at the temporal dynamics and ultimate fates (e.g., extinction or fixation) of total and nonsynomous mutations from the metagenomes."
+---
+
+# Setup
+
+Libraries and global variables
+
+```{r}
+#| output: false
+library(tidyverse)
+library(here)
+library(fs)
+library(patchwork)
+library(ggforce)
+library(latex2exp)
+library(vctrs)
+library(scales)
+library(Polychrome)
+
+source(here::here("R", "utils_generic.R"))
+source(here::here("R", "metagenome", "utils_parallelism.R"))
+```
+
+Set up some directories
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+data_raw <- here::here("_data_raw", "metagenome")
+data <- here::here("data", "metagenome")
+shared <- here::here("_data_raw", "shared")
+figs <- here::here("figs", "metagenome")
+
+# make processed data directory if it doesn't exist
+fs::dir_create(data)
+fs::dir_create(figs)
+```
+
+# Read data
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+
+# these were already filtered in the last step
+mgvars <- read_tsv(here::here(data, "metagenome_variant_timeseries.tsv"))
+degentab <- read_rds(here::here(shared, "annotations_codon_degeneracy.rds"))
+genome_len <- read_tsv(here::here(shared, "HAMBI_genome_len.tsv"))
+# annotations
+annotations <- read_rds(here::here(shared, "annotations_codon_degeneracy.rds"))
+
+# coverage
+covslurpedfmtflt <- read_tsv(here::here(data, "coverage.tsv"))
+
+# parallel genes
+par_genes <- read_tsv(here::here(data, "enriched_parallel_genes.tsv"))
+```
+
+```{r}
+cog_description <- tibble::tribble(
+ ~COG_category_single, ~COG_category_long,
+ "J", "J - Translation, ribosomal structure and biogenesis",
+ "A", "A - RNA processing and modification",
+ "K", "K – Transcription",
+ "L", "L - Replication, recombination and repair",
+ "B", "B - Chromatin structure and dynamics",
+ "D", "D - Cell cycle control, cell division, chromosome partitioning",
+ "Y", "Y - Nuclear structure",
+ "V", "V - Defense mechanisms",
+ "T", "T - Signal transduction mechanisms",
+ "M", "M - Cell wall/membrane/envelope biogenesis",
+ "N", "N - Cell motility",
+ "Z", "Z – Cytoskeleton",
+ "W", "W - Extracellular structures",
+ "U", "U - Intracellular trafficking, secretion, and vesicular transport",
+ "O", "O - Posttranslational modification, protein turnover, chaperones",
+ "X", "X - Mobilome: prophages, transposons",
+ "C", "C - Energy production and conversion",
+ "G", "G - Carbohydrate transport and metabolism",
+ "E", "E - Amino acid transport and metabolism",
+ "F", "F - Nucleotide transport and metabolism",
+ "H", "H - Coenzyme transport and metabolism",
+ "I", "I - Lipid transport and metabolism",
+ "P", "P - Inorganic ion transport and metabolism",
+ "Q", "Q - Secondary metabolites biosynthesis, transport and catabolism",
+ "R", "R - General function prediction only",
+ "S", "S - Function unknown"
+ )
+
+withr::with_seed(12367,
+ cogpal <- unname(createPalette(length(unique(cog_description$COG_category_single)), c("#F3874AFF", "#FCD125FF"), M=5000))
+)
+names(cogpal) <- cog_description$COG_category_long
+```
+
+# Metagenomic coverage
+
+Only looking at the 5 species for which we called variants. Everything else was below a mean coverage of 5 and even calling variants at a coverage of 5 is a bit sketchy. Generally, the higher the coverage your genome is the confidence you can have calling polymorphisms where the alternate allele is rare. Basically this shows that we can really only same something of confidence for HAMBI_1287, HAMBI_1972, HAMBI_1977, and HAMBI_2659.
+
+::: {#fig-01}
+```{r}
+#| fig.width: 8
+#| fig.height: 7
+#| warning: false
+#| error: false
+covslurpedfmtflt %>%
+ filter(strainID %in% c("HAMBI_1287", "HAMBI_1972", "HAMBI_1977", "HAMBI_2659", "HAMBI_0403")) %>%
+ filter(!str_detect(scaffold, "plas")) %>%
+ left_join(distinct(dplyr::select(mgvars, sample, time_days, replicate, prey_history, predator_history)),
+ by = join_by(sample)) %>%
+ group_by(strainID, time_days, prey_history, predator_history) %>%
+ mutate(replicate_mean = mean(trimmed_mean)) %>%
+ ggplot(aes(x = time_days, y = trimmed_mean)) +
+ geom_hline(yintercept = 5, linetype = 2) +
+ geom_line(aes(color = strainID, group = interaction(strainID, prey_history, predator_history, replicate)),
+ alpha = 0.35, linetype = "dashed") +
+ geom_point(aes(color = strainID, shape = replicate),
+ alpha = 0.35) +
+ geom_line(aes(y = replicate_mean, x= time_days, color = strainID)) +
+ geom_point(aes(y = replicate_mean, x= time_days, color = strainID), size = 3) +
+ scale_color_manual(values = hambi_colors) +
+ facet_grid(prey_history ~ predator_history, labeller = label_both) +
+ labs(x = "Experiment time (days)", y = "Mean genome coverage", color = NULL, shape = NULL) +
+ scale_y_log10(breaks = c(1, 10, 100), labels = label_log(base = 10, digits = 3)) +
+ coord_cartesian(ylim = c(1, 300)) +
+ annotation_logticks(sides = "l", color = "grey40") +
+ scale_x_continuous(breaks = c(0, 8, 28, 60)) +
+ theme_bw() +
+ theme(
+ legend.position = "bottom",
+ strip.placement = 'outside',
+ strip.background = element_blank(),
+ panel.grid = element_blank())
+```
+
+Genomic coverage in the metagenomes over time. The genomes of these five species exceeded a coverage of at least 5X (black dashed horizontal line) at at least one sampling time from the experiment. Transparent points/lines represent individual replicates, while opaque points/lines represent the mean over the three replicates.
+:::
+
+# Mutational spectra
+
+Some broad scale patterns in how mutations trajectories vary between species and treatments can be observed from the mutational spectra. Below we examine how the fraction of mutations exceeding the maximum observed allele frequency $f_{max}$ relates to different levels of $f_{max}$ and compare this between different evolution treatments and across species.
+
+```{r}
+allelespectra <- mgvars %>%
+ dplyr::select(strainID, replicate, prey_history, predator_history, freq_alt_complete) %>%
+ drop_na() %>%
+ group_by(strainID, replicate, prey_history, predator_history) %>%
+ # the rbeta is to just add some noise at the lower and upper ends where there are ties
+ mutate(freq_alt_complete = case_when(freq_alt_complete == 1 ~ freq_alt_complete - rbeta(n(), 0.1, 10, ncp = 0),
+ freq_alt_complete == 0 ~ freq_alt_complete + rbeta(n(), 0.1, 10, ncp = 0),
+ TRUE ~ freq_alt_complete)) %>%
+ filter(freq_alt_complete >= 0.1) %>%
+ arrange(strainID, replicate, prey_history, predator_history, desc(freq_alt_complete)) %>%
+ group_by(strainID, replicate, prey_history, predator_history) %>%
+ # get 1 - emiprical cumulative distribution
+ mutate(f_gt_fmax = 1.001 - ecdf(freq_alt_complete)(freq_alt_complete))
+```
+
+::: {#fig-02}
+```{r}
+ggplot(allelespectra, aes(x = freq_alt_complete, y = f_gt_fmax)) +
+ geom_step(aes(color = predator_history, linetype = replicate, group = interaction(predator_history, replicate))) +
+ labs(x = TeX("Maximum observed allele frequency $\\textit{f_{max}}$"), y = TeX("Fraction of mutations $\\geq \\textit{f_{max}}$")) +
+ facet_grid(prey_history ~ strainID, labeller = label_both) +
+ annotation_logticks(sides = "bl", color = "grey40") +
+ scale_y_continuous(trans = "log10", breaks = c(1, 0.1, 0.01, 0.001)) +
+ scale_x_continuous(trans = "log10", breaks = c(1, 0.1), limits = c(1e-1, 1)) +
+ theme_bw() +
+ theme(
+ legend.position = "bottom",
+ strip.placement = 'outside',
+ strip.background = element_blank(),
+ panel.grid = element_blank()
+ )
+```
+
+The empirical cumulative distribution of the maximum frequency that mutations reached $f_{max}$ for each taxon-treatment combination.
+:::
+
+Mutational spectra from the ancestral bacteria are of lower resolution because there are fewer mutations in those populations during the course of the experiment. Spectra from evolved bacteria are of higher resolution but include both standing genetic variation from the start of the experiment and de novo mutions. Thus, comparing ancestral to evolved bacteria spectra is not particularly meaningful. However, some insights may be gained from comparing different predator evolutionary histories nested with bacterial evolutionary histories. For evolved bacteria it appears that for 1972, 1977, and 2659 that the trajectory of $f_{max}$ in the ancestral predator treatment drops relative to the evolved predator treatment. This would then imply that, on the whole, bacterial alleles from the evolved predator treatments increased to higher frequencies than alleles in the ancestral predator treatments. However, I am not sure there is strong support for this assertion - there are only three replicates and usually at least one replicate from the evolved predator treatment overlaps with the ancestral predator treatment. A comparison of the spectra between species indicates that, generally, some species (HAMBI_1972 and HAMBI_0403) have more high frequency mutations than others (HAMBI_1287 and HAMBI_2659).
+
+# Mutational dynamics
+
+We calculated the accumulation of mutations by time t as the sum of derived allele frequencies
+
+$$ M\left(t\right)\equiv\sum_m {\widehat f}_{pmt} $$ where ${\widehat f}_{pmt}=A_{pmt}/D_{pmt}$ and A is the coverage of the alternative allele and D is the total depth of coverage. To partially mitigate noise associated with calling low frequency variants in medium-low coverage samples (\~20X) we restricted our analysis to variants where ${\widehat f}_{pmt} \geq 0.15$.
+
+## Some data formatting
+
+```{r}
+# only denovo (i.e. not observed at time 0)
+mgvars <- mgvars %>%
+ group_by(strainID, chrom, pos, ref, alt, replicate, prey_history, predator_history) %>%
+ mutate(group_id = cur_group_id()) %>%
+ ungroup()
+
+# only denovo (i.e. not observed at time 0)
+mgvars_denovo <- mgvars %>%
+ group_by(strainID, chrom, pos, ref, alt, replicate, prey_history, predator_history) %>%
+ mutate(group_id = cur_group_id()) %>%
+ mutate(denovo = if_else(freq_alt_complete == 0 & time_days == 0, 1, NA_real_)) %>%
+ fill(denovo, .direction = "down") %>%
+ relocate(group_id, freq_alt_complete, denovo) %>%
+ ungroup() %>%
+ filter(!is.na(denovo))
+
+# only nonsynonymous variants
+mgvars_ns <- mgvars %>%
+ filter(!str_detect(effect, "intergenic|intragenic|synonymous|fusion")) %>%
+ filter(!str_detect(impact, "MODIFIER"))
+
+# only denovo (i.e. not observed at time 0) and nonsynonymous variants
+mgvars_ns_denovo <- mgvars_ns %>%
+ group_by(strainID, chrom, pos, ref, alt, replicate, prey_history, predator_history) %>%
+ mutate(group_id = cur_group_id()) %>%
+ mutate(denovo = if_else(freq_alt_complete == 0 & time_days == 0, 1, NA_real_)) %>%
+ fill(denovo, .direction = "down") %>%
+ relocate(group_id, freq_alt_complete, denovo) %>%
+ ungroup() %>%
+ filter(!is.na(denovo))
+
+cummt <- mgvars %>%
+ summarize(Mt = sum(freq_alt_complete[freq_alt_complete >= 0.15], na.rm = TRUE),
+ .by = c(strainID, time_days, replicate, prey_history, predator_history))
+
+cummt_denovo <- mgvars_denovo %>%
+ summarize(Mt = sum(freq_alt_complete[freq_alt_complete >= 0.15], na.rm = TRUE),
+ .by = c(strainID, time_days, replicate, prey_history, predator_history))
+
+cummt_ns <- mgvars_ns %>%
+ summarize(Mt = sum(freq_alt_complete[freq_alt_complete >= 0.15], na.rm = TRUE),
+ .by = c(strainID, time_days, replicate, prey_history, predator_history))
+
+cummt_ns_denovo <- mgvars_ns_denovo %>%
+ summarize(Mt = sum(freq_alt_complete[freq_alt_complete >= 0.15], na.rm = TRUE),
+ .by = c(strainID, time_days, replicate, prey_history, predator_history))
+```
+
+```{r}
+plotmuttraj <- function(df, x, alpha, ylab){
+ ggplot(df, aes(x = time_days, y = {{ x }})) +
+ geom_point(aes(color = strainID, shape = replicate), alpha = alpha) +
+ geom_line(aes(color = strainID,
+ group = interaction(strainID, prey_history, predator_history, replicate)),
+ alpha = alpha,
+ linetype = "dashed") +
+ scale_color_manual(values = hambi_colors) +
+ facet_grid(prey_history ~ predator_history, labeller = label_both) +
+ labs(x = "Experiment time (days)", y = ylab, color = NULL, shape = NULL) +
+ scale_x_continuous(breaks = c(0, 8, 28, 60)) +
+ theme_bw() +
+ theme(
+ legend.position = "bottom",
+ strip.placement = 'outside',
+ strip.background = element_blank(),
+ panel.grid = element_blank()
+ )
+}
+```
+
+## All variants {#sec-allvars}
+
+This section considers all variants from the experiment including standing genetic variation present in the evolved populations at T0 and de novo mutations that arise later. It also includes both protein changing and protein non-changing variants.
+
+### Cumulative mutation trajectories (M(t))
+
+::: {#fig-03}
+```{r}
+#| warning: false
+#| fig.width: 8
+#| fig.height: 5
+cummt %>%
+ group_by(strainID, time_days, prey_history, predator_history) %>%
+ mutate(mn = mean(Mt)) %>%
+ plotmuttraj(Mt, 0.35, "M(t)") +
+ geom_line(aes(y = mn, x= time_days, color = strainID)) +
+ geom_point(aes(y = mn, x= time_days, color = strainID), size = 3)
+```
+
+Cumulative mutation (M(t)) trajectories of both amino acid altering and amino acid non-altering variants in HAMBI species genomes with metagenomic coverage $\geq$ 5X in at least one time point (@fig-01). For evolved prey treatments this includes both standing mutations at T0 and de novo mutations. Transparent points/lines represent individual replicates, while opaque points/lines represent the mean over the three replicates.
+:::
+
+### Total mutation trajectories
+
+::: {#fig-04}
+```{r}
+#| warning: false
+#| fig.width: 8
+#| fig.height: 5
+mgvars %>%
+ summarize(n = sum(freq_alt_complete > 0, na.rm = TRUE),
+ .by = c(strainID, time_days, replicate, prey_history, predator_history)) %>%
+ group_by(strainID, time_days, prey_history, predator_history) %>%
+ mutate(mn = mean(n)) %>%
+ plotmuttraj(n, 0.35, "Number of mutations") +
+ geom_line(aes(y = mn, x= time_days, color = strainID)) +
+ geom_point(aes(y = mn, x= time_days, color = strainID), size = 3) +
+ annotation_logticks(sides = "l", color = "grey40") +
+ scale_y_log10(breaks = c(1, 10, 100),
+ labels = label_log(base = 10, digits = 3))
+```
+
+Trajectories of the **total number of non-zero mutations** detected at each time point for both amino acid altering and amino acid non-altering variants in HAMBI genomes with metagenomic coverage $\geq$ 5X in at least one time point (@fig-01). For evolved prey treatments this includes both standing mutations at T0 and de novo mutations. Transparent points/lines represent individual replicates, while opaque points/lines represent the mean over the three replicates.
+:::
+
+### Fixed mutation trajectories
+
+::: {#fig-05}
+```{r}
+#| warning: false
+#| fig.width: 8
+#| fig.height: 5
+mgvars %>%
+ summarize(n = sum(freq_alt_complete >= 1, na.rm = TRUE),
+ .by = c(strainID, time_days, replicate, prey_history, predator_history)) %>%
+ group_by(strainID, time_days, prey_history, predator_history) %>%
+ mutate(mn = mean(n)) %>%
+ plotmuttraj(n, 0.35, "Number of fixed mutations") +
+ geom_line(aes(y = mn, x= time_days, color = strainID)) +
+ geom_point(aes(y = mn, x= time_days, color = strainID), size = 3)
+```
+
+Trajectories of the total number of **fixed** mutations detected at each time point for both amino acid altering and amino acid non-altering variants in HAMBI genomes with metagenomic coverage $\geq$ 5X in at least one time point (@fig-01). For evolved prey treatments this includes both standing mutations at T0 and de novo mutations. Transparent points/lines represent individual replicates, while opaque points/lines represent the mean over the three replicates.
+:::
+
+## De novo variants
+
+This section considers only de novo mutations that did not exist within the evolved populations at T0. It includes both protein changing and protein non-changing variants. Thus, the results here for the ancestral bacteria populations here will be the same as in @sec-allvars.
+
+### Cumulative mutation trajectories (M(t))
+
+::: {#fig-06}
+```{r}
+#| warning: false
+#| fig.width: 8
+#| fig.height: 5
+fig06 <- cummt_denovo %>%
+ group_by(strainID, time_days, prey_history, predator_history) %>%
+ mutate(mn = mean(Mt)) %>%
+ plotmuttraj(Mt, 0.35, "M(t)") +
+ geom_line(aes(y = mn, x= time_days, color = strainID)) +
+ geom_point(aes(y = mn, x= time_days, color = strainID), size = 3)
+
+ggsave(here::here(figs, "Mt_denovo.svg"), fig06, width=7, height=5, units="in",
+ device="svg")
+
+fig06
+```
+
+Cumulative mutation (M(t)) trajectories of de novo amino acid altering and amino acid non-altering variants in HAMBI species genomes with metagenomic coverage $\geq$ 5X in at least one time point (@fig-01). Transparent points/lines represent individual replicates, while opaque points/lines represent the mean over the three replicates. Plot is analogous to @fig-03 but only includes de novo variants (i.e. excludes standing genetic variation in the evolved populations).
+:::
+
+### Total mutation trajectories
+
+::: {#fig-07}
+```{r}
+#| warning: false
+#| fig.width: 8
+#| fig.height: 5
+mgvars_denovo %>%
+ summarize(n = sum(freq_alt_complete > 0, na.rm = TRUE),
+ .by = c(strainID, time_days, replicate, prey_history, predator_history)) %>%
+ group_by(strainID, time_days, prey_history, predator_history) %>%
+ mutate(mn = mean(n)) %>%
+ plotmuttraj(n, 0.35, "Number of mutations") +
+ geom_line(aes(y = mn, x= time_days, color = strainID)) +
+ geom_point(aes(y = mn, x= time_days, color = strainID), size = 3) +
+ annotation_logticks(sides = "l", color = "grey40") +
+ scale_y_log10(breaks = c(1, 10, 100),
+ labels = label_log(base = 10, digits = 3))
+```
+
+Trajectories of the **total number of non-zero mutations** detected at each time point for de novo amino acid altering and amino acid non-altering variants in HAMBI genomes with metagenomic coverage $\geq$ 5X in at least one time point (@fig-01). For evolved prey treatments this includes both standing mutations at T0 and de novo mutations. Transparent points/lines represent individual replicates, while opaque points/lines represent the mean over the three replicates. Plot is analogous to @fig-04 but only includes de novo variants (i.e. excludes standing genetic variation in the evolved populations).
+:::
+
+### Fixed mutation trajectories
+
+::: {#fig-08}
+```{r}
+#| warning: false
+#| fig.width: 8
+#| fig.height: 5
+mgvars_denovo %>%
+ summarize(n = sum(freq_alt_complete >= 1, na.rm = TRUE),
+ .by = c(strainID, time_days, replicate, prey_history, predator_history)) %>%
+ group_by(strainID, time_days, prey_history, predator_history) %>%
+ mutate(mn = mean(n)) %>%
+ plotmuttraj(n, 0.35, "Number of fixed mutations") +
+ geom_line(aes(y = mn, x= time_days, color = strainID)) +
+ geom_point(aes(y = mn, x= time_days, color = strainID), size = 3)
+```
+
+Trajectories of the total number of **fixed** mutations detected at each time point for de novo amino acid altering and amino acid non-altering variants in HAMBI genomes with metagenomic coverage $\geq$ 5X in at least one time point (@fig-01). Transparent points/lines represent individual replicates, while opaque points/lines represent the mean over the three replicates. Plot is analogous to @fig-05 but only includes de novo variants (i.e. excludes standing genetic variation in the evolved populations).
+:::
+
+## Non-synonymous variants {#sec-nsvars}
+
+This section considers all nonsynonymous (i.e., amino acid altering) variants from the experiment including standing genetic variation present in the evolved populations at T0 and de novo mutations that arise later.
+
+### Cumulative mutation trajectories (M(t))
+
+::: {#fig-09}
+```{r}
+#| warning: false
+#| fig.width: 8
+#| fig.height: 5
+cummt_ns %>%
+ group_by(strainID, time_days, prey_history, predator_history) %>%
+ mutate(mn = mean(Mt)) %>%
+ plotmuttraj(Mt, 0.35, "M(t) (nonsynonymous)") +
+ geom_line(aes(y = mn, x= time_days, color = strainID)) +
+ geom_point(aes(y = mn, x= time_days, color = strainID), size = 3)
+```
+
+Cumulative mutation (M(t)) trajectories of amino acid altering variants in HAMBI species genomes with metagenomic coverage $\geq$ 5X in at least one time point (@fig-01). For evolved prey treatments this includes both standing mutations at T0 and de novo mutations. Transparent points/lines represent individual replicates, while opaque points/lines represent the mean over the three replicates. Plot is analogous to @fig-03 but only includes nonsynonymous (amino acid changing) variants.
+:::
+
+### Total mutation trajectories
+
+::: {#fig-10}
+```{r}
+#| warning: false
+#| fig.width: 8
+#| fig.height: 5
+mgvars_ns %>%
+ summarize(n = sum(freq_alt_complete > 0, na.rm = TRUE),
+ .by = c(strainID, time_days, replicate, prey_history, predator_history)) %>%
+ group_by(strainID, time_days, prey_history, predator_history) %>%
+ mutate(mn = mean(n)) %>%
+ plotmuttraj(n, 0.35, "Number of nonsynonymous mutations") +
+ geom_line(aes(y = mn, x= time_days, color = strainID)) +
+ geom_point(aes(y = mn, x= time_days, color = strainID), size = 3) +
+ annotation_logticks(sides = "l", color = "grey40") +
+ scale_y_log10(breaks = c(1, 10, 100),
+ labels = label_log(base = 10, digits = 3))
+```
+
+Trajectories of the **total number of non-zero mutations** detected at each time point for amino acid altering variants in HAMBI genomes with metagenomic coverage $\geq$ 5X in at least one time point (@fig-01). For evolved prey treatments this includes both standing mutations at T0 and de novo mutations. Transparent points/lines represent individual replicates, while opaque points/lines represent the mean over the three replicates. Plot is analogous to @fig-04 but only includes nonsynonymous (amino acid changing) variants.
+:::
+
+### Fixed mutation trajectories
+
+::: {#fig-11}
+```{r}
+#| warning: false
+#| fig.width: 8
+#| fig.height: 5
+mgvars_ns %>%
+ summarize(n = sum(freq_alt_complete >= 1, na.rm = TRUE),
+ .by = c(strainID, time_days, replicate, prey_history, predator_history)) %>%
+ group_by(strainID, time_days, prey_history, predator_history) %>%
+ mutate(mn = mean(n)) %>%
+ plotmuttraj(n, 0.35, "Number of nonsynonymous fixed mutations") +
+ geom_line(aes(y = mn, x= time_days, color = strainID)) +
+ geom_point(aes(y = mn, x= time_days, color = strainID), size = 3)
+```
+
+Trajectories of the total number of **fixed** mutations detected at each time point for amino acid altering variants in HAMBI genomes with metagenomic coverage $\geq$ 5X in at least one time point (@fig-01). For evolved prey treatments this includes both standing mutations at T0 and de novo mutations. Transparent points/lines represent individual replicates, while opaque points/lines represent the mean over the three replicates. Plot is analogous to @fig-05 but only includes non-synonymous (amino acid changing) variants.
+:::
+
+## De novo, non-synonymous variants
+
+This section considers only de novo mutations that did not exist within the evolved populations at T0 for only non-synonymous (i.e., amino acid altering) variants. Thus, the results here for the ancestral bacteria populations will be the same as in @sec-nsvars
+
+### Cumulative mutation trajectories (M(t))
+
+::: {#fig-12}
+```{r}
+#| warning: false
+#| fig.width: 8
+#| fig.height: 5
+cummt_ns_denovo %>%
+ group_by(strainID, time_days, prey_history, predator_history) %>%
+ mutate(mn = mean(Mt)) %>%
+ plotmuttraj(Mt, 0.35, "M(t) (de novo, nonsynonymous)") +
+ geom_line(aes(y = mn, x= time_days, color = strainID)) +
+ geom_point(aes(y = mn, x= time_days, color = strainID), size = 3)
+```
+
+Cumulative mutation (M(t)) trajectories of de novo amino acid altering variants in HAMBI species genomes with metagenomic coverage $\geq$ 5X in at least one time point (@fig-01). Transparent points/lines represent individual replicates, while opaque points/lines represent the mean over the three replicates. Plot is analogous to @fig-03 but only includes de novo (i.e., not present at T0) and non-synonymous (amino acid changing) variants.
+:::
+
+### Total mutation trajectories
+
+::: {#fig-13}
+```{r}
+#| warning: false
+#| fig.width: 8
+#| fig.height: 5
+fig13 <- mgvars_ns_denovo %>%
+ summarize(n = sum(freq_alt_complete > 0, na.rm = TRUE),
+ .by = c(strainID, time_days, replicate, prey_history, predator_history)) %>%
+ group_by(strainID, time_days, prey_history, predator_history) %>%
+ mutate(mn = mean(n)) %>%
+ plotmuttraj(n, 0.35, "Number of de novo nonsynonymous mutations") +
+ geom_line(aes(y = mn, x= time_days, color = strainID)) +
+ geom_point(aes(y = mn, x= time_days, color = strainID), size = 3) +
+ annotation_logticks(sides = "l", color = "grey40") +
+ scale_y_log10(breaks = c(1, 10, 100),
+ labels = label_log(base = 10, digits = 3))
+
+ggsave(here::here(figs, "cummuts_denovo_nonsyn.svg"), fig13, width=7, height=5, units="in",
+ device="svg")
+
+fig13
+```
+
+Trajectories of the **total number of non-zero mutations** detected at each time point for de novo amino acid altering variants in HAMBI genomes with metagenomic coverage $\geq$ 5X in at least one time point (@fig-01). Transparent points/lines represent individual replicates, while opaque points/lines represent the mean over the three replicates. Plot is analogous to @fig-04 but only includes de novo (i.e., not present at T0) and non-synonymous (amino acid changing) variants.
+:::
+
+### Fixed mutation trajectories
+
+::: {#fig-14}
+```{r}
+#| warning: false
+#| fig.width: 8
+#| fig.height: 5
+mgvars_ns_denovo %>%
+ summarize(n = sum(freq_alt_complete >= 1, na.rm = TRUE),
+ .by = c(strainID, time_days, replicate, prey_history, predator_history)) %>%
+ group_by(strainID, time_days, prey_history, predator_history) %>%
+ mutate(mn = mean(n)) %>%
+ plotmuttraj(n, 0.35, "Number of de novo, nonsynonymous fixed mutations") +
+ geom_line(aes(y = mn, x= time_days, color = strainID)) +
+ geom_point(aes(y = mn, x= time_days, color = strainID), size = 3)
+```
+
+Trajectories of the total number of **fixed** mutations detected at each time point for de novo amino acid altering variants in HAMBI genomes with metagenomic coverage $\geq$ 5X in at least one time point (@fig-01). Transparent points/lines represent individual replicates, while opaque points/lines represent the mean over the three replicates. Plot is analogous to @fig-05 but only includes de novo (i.e., not present at T0) and non-synonymous (amino acid changing) variants.
+:::
+
+# Variant time to majority
+
+Does the evolutionary history of prey/predator impact how long it takes alleles to reach majority status ($f_{max} \geq 0.5$). For every alternative allele that emerged in the experiment de novo, we plot the time that it took for that alt allele to reach majority status (i.e., the alt allele dominates the reference allele).
+
+## De novo variants (both amino acid chaning and non-changing)
+
+::: {#fig-15}
+```{r}
+#| warning: false
+#| fig.width: 8
+#| fig.height: 7
+mgvars_denovo %>%
+ group_by(group_id) %>%
+ filter(!is.na(freq_alt_complete) & freq_alt_complete != 0) %>%
+ filter(freq_alt_complete >= 0.50) %>%
+ slice_min(time_days) %>%
+ slice(1) %>%
+ ggplot(aes(x = factor(time_days))) +
+ geom_bar(aes(fill = strainID)) +
+ scale_fill_manual(values = hambi_colors) +
+ facet_grid(prey_history ~ predator_history, labeller = label_both) +
+ labs(x = TeX("Time to major allele ( $f_{max} \\geq 0.5$ )"), y = TeX(" $n_{allele}$ with $f_{max} \\geq 0.5$ "), fill = NULL, shape = NULL) +
+ theme_bw() +
+ theme(
+ legend.position = "bottom",
+ strip.placement = 'outside',
+ strip.background = element_blank(),
+ panel.grid = element_blank())
+```
+
+Count of alternative alleles with $f_{max} \geq 0.5$ plotted against the time (days) when the alternative allele first reaches major allele status $(f_{max} \geq 0.5)$. Colors represent the proportion of alleles from different HAMBI species. Plot includes both amino acid changing and non-changing de novo mutations (i.e. excludes standing genetic variation from T0 in the evolved bacterial populations).
+:::
+
+## De novo variants, non-synonymous
+
+::: {#fig-16}
+```{r}
+#| warning: false
+#| fig.width: 8
+#| fig.height: 7
+mgvars_ns_denovo %>%
+ group_by(group_id) %>%
+ filter(!is.na(freq_alt_complete) & freq_alt_complete != 0) %>%
+ filter(freq_alt_complete >= 0.50) %>%
+ slice_min(time_days) %>%
+ slice(1) %>%
+ ggplot(aes(x = factor(time_days))) +
+ geom_bar(aes(fill = strainID)) +
+ scale_fill_manual(values = hambi_colors) +
+ facet_grid(prey_history ~ predator_history, labeller = label_both) +
+ labs(x = TeX("Time to major allele ( $f_{max} \\geq 0.5$ )"), y = TeX(" $n_{allele}$ with $f_{max} \\geq 0.5$ "), fill = NULL, shape = NULL) +
+ theme_bw() +
+ theme(
+ legend.position = "bottom",
+ strip.placement = 'outside',
+ strip.background = element_blank(),
+ panel.grid = element_blank())
+```
+
+Count of alternative alleles with $f_{max} \geq 0.5$ plotted against the time (days) when the alternative allele first reaches major allele status $(f_{max} \geq 0.5)$. Colors represent the proportion of alleles from different HAMBI species. Plot includes only amino acid changing de novo variants (i.e. excludes standing genetic variation from T0 in the evolved bacterial populations).
+:::
+
+This plot suggests that most de novo alternative alleles (either amino acid changing or non chaning) become the major allele by either day 28 or 60 for the ancestral bacteria populations. For the evolved populations this evolution seems to happen faster where some alt alleles become the major allele already by day 28 but this is only for HAMBI_2659.
+
+# Time to mutation emergence in parallel genes
+
+```{r}
+df_ttm <- left_join(par_genes, mgvars_ns_denovo) %>%
+ group_by(group_id) %>%
+ filter(!is.na(freq_alt_complete) & freq_alt_complete != 0) %>%
+ # uncomment to only look at emergence of minor-to-major allele transition
+ #filter(freq_alt_complete >= 0.50) %>%
+ slice_min(time_days) %>%
+ slice(1) %>%
+ ungroup() %>%
+ # some manual renaming
+ mutate(gene = case_when(locus_tag == "H1287_02172" ~ "yddV_2",
+ locus_tag == "H1977_02612" ~ "dppA/oppA",
+ locus_tag == "H1972_00299" ~ "PAP2_like",
+ locus_tag == "H1972_02826" ~ "ynjC",
+ locus_tag == "H1972_00671" ~ "yqaA",
+ locus_tag == "H1972_00421" ~ "hyp",
+ locus_tag == "H1972_02723" ~ "yifE",
+ locus_tag == "H2659_01789" ~ "hyp",
+ TRUE ~ gene)) %>%
+ mutate(gene_lab = if_else(is.na(gene), Preferred_name, gene),
+ treat = paste0("prey:", prey_history, " | pred:", predator_history)) %>%
+ mutate(gene_lab = if_else(gene_lab == "-", Description, gene_lab)) %>%
+ mutate(gene_lab = paste0(treat,"-", gene_lab)) %>%
+ mutate(gene_lab2 = fct_reorder(factor(gene_lab), time_days))
+```
+
+
+```{r}
+pbar <- df_ttm %>%
+ distinct(gene_lab2, COG_category_long) %>%
+ ggplot(aes(x = 1, y = gene_lab2, fill = COG_category_long)) +
+ geom_tile() +
+ scale_fill_manual(values = cogpal) +
+ theme_void()
+
+ptext <- df_ttm %>%
+ distinct(gene_lab2, COG_category_long, COG_category_single) %>%
+ replace_na(list(COG_category_long = "S - Function unknown", COG_category_single = "S")) %>%
+ ggplot(aes(x = 1, y = gene_lab2)) +
+ geom_text(aes(label = COG_category_single, color = COG_category_long)) +
+ scale_color_manual(values = cogpal) +
+ theme_void()
+```
+
+
+```{r}
+ppoint <- ggplot(df_ttm, aes(x = factor(time_days), y = gene_lab2)) +
+ geom_jitter(aes(color = treat, shape = strainID), size = 2, width = 0.15, height = 0) +
+ labs(x = "Time to mutation emergence", y = "") +
+ scale_color_brewer(palette = "Set1") +
+ theme_bw() +
+ theme(
+ legend.position = "left",
+ strip.placement = 'outside',
+ strip.background = element_blank(),
+ panel.grid = element_blank())
+```
+
+::: {#fig-17}
+```{r}
+#| warning: false
+#| fig.width: 9
+#| fig.height: 7
+ptogether <- ppoint + ptext +
+ plot_layout(ncol = 2, widths = c(1, 0.15), guides = "collect")
+
+ggsave(here::here(figs, "par_gene_time2emergence.svg"), ptogether, width=9, height=7, units="in",
+ device="svg")
+
+ptogether
+```
+
+:::
+
+dhaR has two associated COG categories
+1. K – Transcription
+2. Q - Secondary metabolites biosynthesis transport and catabolism
+
+smf-1 (Major fimbrial subunit SMF-1) has two associated COG categoires
+1. N - Cell motility
+2. U - intracellular trafficiking, secretion, and vesicular transport
+
+
+# Relationship between gene parallelism and $f_{max}$
+Look if there is a relationship between extent of gene parallelism ($\Delta L$) and the max frequency that variants reach during the experiment.
+
+As seen from the results below we observe that the degree of parallelism $\Delta l$ or $G_{i}$ increases with $f_{max}$ across experimental treatment combinations and species, which is consistent with the idea that mutations with higher $f_{max}$ are primarily being driven by positive selection across these treatments.
+
+## $\Delta l$
+
+This is the same formatting that is done in [the prior parallelism analysis.](02_parallelism.qmd)
+```{r}
+tofilter <- mgvars %>%
+ filter(time_days == 0) %>%
+ filter(freq_alt_complete == 1) %>%
+ dplyr::select(chrom, pos, ref, alt) %>%
+ distinct()
+
+mgvars_filt_mb_ns <- anti_join(mgvars, tofilter, by = join_by(chrom, pos, ref, alt)) %>%
+ # exclude variants that were added to make complete time course
+ filter(!is.na(freq_alt_complete)) %>%
+ filter(freq_alt_complete > 0) %>%
+ filter(time_days != 0) %>%
+ filter(!str_detect(effect, "intergenic|intragenic|synonymous|fusion")) %>%
+ filter(!str_detect(impact, "MODIFIER")) %>%
+ # double check to remove any noncoding mutations
+ filter(locus_tag %in% pull(filter(degentab, !is.na(ns_length)), locus_tag))
+```
+
+We now want to look at the extent of multiplicity at different binned $f_{max}$
+```{r}
+multiplicity_binned <- mgvars_filt_mb_ns %>%
+ # automatically bin f_max into 5 categories
+ mutate(fbin = cut_interval(freq_alt_complete, 5)) %>%
+ summarize(n_i = n(),
+ n_replicate = n_distinct(replicate),
+ .by = c("locus_tag", "strainID", "fbin", "prey_history", "predator_history")) %>%
+ group_by(locus_tag, strainID) %>%
+ complete(fbin, prey_history, predator_history,
+ fill = list(n_i = 0, n_replicate = 0)) %>%
+ ungroup() %>%
+ group_by(strainID, fbin, prey_history, predator_history) %>%
+ mutate(Ntot = sum(n_i)) %>%
+ ungroup()
+```
+
+Make df with information about gene lengths
+```{r}
+gene_tab <- degentab %>%
+ dplyr::select(strainID, locus_tag, l_i = ns_length) %>%
+ drop_na() %>%
+ group_by(strainID) %>%
+ mutate(
+ # compute the number of genes in the genome
+ n_genes = n(),
+ # compute the sum of non-syn sites in genome
+ Ltot = sum(l_i),
+ # compute the average non-syn sites per gene
+ Lavg = mean(l_i)) %>%
+ ungroup() %>%
+ filter(strainID %in% c("HAMBI_0403", "HAMBI_1287", "HAMBI_1972", "HAMBI_1977", "HAMBI_2659")) %>%
+ # expanding this now to make later joining faster
+ expand_grid(prey_history = c("anc", "evo"),
+ predator_history = c("anc", "evo"),
+ fbin = unique(multiplicity_binned$fbin))
+```
+
+This prepares the tibble that we can use in the next step to calculate the G scores and $\Delta L$
+```{r}
+g_prepped <- left_join(gene_tab, multiplicity_binned) %>%
+ arrange(strainID, prey_history, predator_history, fbin, locus_tag) %>%
+ replace_na(list(n_i = 0, n_replicate = 0)) %>%
+ group_by(strainID, prey_history, predator_history, fbin) %>%
+ fill(Ntot, .direction = "updown") %>%
+ ungroup() %>%
+ group_by(strainID, prey_history, predator_history) %>%
+ filter(sum(Ntot, na.rm = TRUE) != 0) %>%
+ ungroup()
+```
+
+This function calculate $\Delta L$ but from a bootstrapping of the source dataframes to get sample means plus confidence intervals. We use `rsample` package hence the use of `analysis(df).` See the next section for implementation
+
+```{r}
+G_score_boot <- function(df, boot=TRUE){
+
+ if (boot) {
+ # convert the rsample splits to tibble
+ df <- rsample::analysis(df)
+ }
+ # compute the expected multiplicity value
+ m_exp <- unique(df$Ntot) / unique(df$n_genes)
+ # compute gene multiplicity for gene i
+ m_i <- df$n_i * unique(df$Lavg) / df$l_i
+ # compute the G score for each gene
+ g_score <- df$n_i * log(m_i / m_exp)
+ # net increase in log-likelihood compared to the null model of m_i / m_exp =1
+ # normalized to the total mutations
+ sum(g_score, na.rm=T) / unique(df$Ntot)
+}
+```
+
+We need the package `rsample` to do the boostrap analysis and we need `Hmisc` to get the bootstrapped confidence intervals
+
+```{r}
+#| eval: false
+#| echo: true
+library(rsample)
+library(Hmisc)
+
+# this is a bit slow so saving the results
+g_boot <- g_prepped %>%
+ tidyr::nest(.by = c(strainID, prey_history, predator_history, fbin)) %>%
+ dplyr::mutate(boots = purrr::map(data, function(df) rsample::bootstraps(df, times = 1000))) %>%
+ dplyr::mutate(g_scores = purrr::map(boots, \(boot) purrr::map_dbl(boot$splits, G_score_boot))) %>%
+ dplyr::summarize(meanboot = purrr::map(g_scores, ggplot2::mean_cl_boot),
+ .by = c(strainID, prey_history, predator_history, fbin)) %>%
+ tidyr::unnest(cols = c(meanboot))
+
+# Save this for later
+write_rds(g_boot, here::here(data, "g_score_bootstap.rds"))
+```
+
+Read in saved bootstrapped data
+```{r}
+# Read in previous results
+g_boot <- read_rds(here::here(data, "g_score_bootstap.rds"))
+```
+
+Plot bootstrapped $\Delta l$ range with $\Delta l$ from full dataset at different values of $f_{max}$
+
+::: {#fig-18}
+```{r}
+#| warning: false
+#| fig.width: 8
+#| fig.height: 7
+pg_boot <- ggplot(g_boot, aes(x = fbin, y = y, ymin = ymin, ymax = ymax)) +
+ geom_pointrange(aes(color = strainID),
+ position = position_dodge(width = 0.5)) +
+ facet_grid(prey_history ~ predator_history, labeller = label_both) +
+ labs(x = TeX("Binned maximum observed allele frequency ($f_{max}$) "), y = TeX("Extent of parallel evolution ($\\Delta l$)"), color = NULL, shape = NULL) +
+ scale_color_manual(values = hambi_colors) +
+ scale_x_discrete(guide = guide_axis(angle = 45)) +
+ theme_bw() +
+ theme(
+ legend.position = "bottom",
+ strip.placement = 'outside',
+ strip.background = element_blank(),
+ panel.grid = element_blank())
+
+pg_boot
+```
+
+:::
+
+## G score
+
+Also calculate the G score from [@tenaillon2016] and look at the relationship with $f_{max}$. The difference between the analysis above is that G-score is the non-summed and non-normalized form of $\Delta l$ and it is calculated for every gene uniquely instead of at different $f_{max}$ bins. Briefly,
+
+$$
+G_{i} = n_{i} \log \left ( \frac{m_{i}}{\overline{m}} \right )
+$$
+where the expected multiplicity $\overline{m}$ is
+
+$$
+\overline{m} = n_{tot}/n_{genes}
+$$
+
+and the observed multiplicity is
+
+$$
+m_{i} = n_{i} \cdot \frac{\overline{L}}{L_{i}}
+$$
+
+```{r}
+multiplicity <- mgvars_filt_mb_ns %>%
+ dplyr::summarize(n_i = n(),
+ n_replicate = n_distinct(replicate),
+ .by = c("locus_tag", "strainID", "prey_history", "predator_history")) %>%
+ dplyr::group_by(strainID, prey_history, predator_history) %>%
+ dplyr::mutate(Ntot = sum(n_i)) %>%
+ dplyr::ungroup()
+
+g_prepped_nobin <- left_join(dplyr::distinct(dplyr::select(gene_tab, -fbin)), multiplicity) %>%
+ arrange(strainID, prey_history, predator_history, locus_tag) %>%
+ replace_na(list(n_i = 0, n_replicate = 0)) %>%
+ group_by(strainID, prey_history, predator_history) %>%
+ fill(Ntot, .direction = "updown") %>%
+ ungroup() %>%
+ group_by(strainID, prey_history, predator_history) %>%
+ filter(sum(Ntot, na.rm = TRUE) != 0) %>%
+ ungroup()
+
+G_score_tidy <- function(df){
+ df %>%
+ # compute the expected multiplicity value
+ mutate(m_exp = Ntot / n_genes) %>%
+ # compute gene multiplicity for gene i
+ mutate(m_i = n_i * Lavg / l_i) %>%
+ # compute the G score for each gene
+ mutate(g_score = n_i * log(m_i / m_exp)) %>%
+ # net increase in log-likelihood compared to the null model of m_i / m_exp =1
+ # normalized to the total mutations
+ mutate(delta_l = sum(g_score, na.rm=T) / Ntot)
+}
+
+g_scores <- g_prepped_nobin %>%
+ group_by(strainID, prey_history, predator_history) %>%
+ G_score_tidy() %>%
+ drop_na() %>%
+ ungroup()
+
+g_scores_fmax <- left_join(g_scores, mgvars,
+ by = join_by(strainID, locus_tag, prey_history, predator_history)) %>%
+ group_by(locus_tag) %>%
+ filter(freq_alt_complete == max(freq_alt_complete, na.rm = TRUE)) %>%
+ ungroup()
+```
+
+::: {#fig-19}
+```{r}
+#| warning: false
+#| fig.width: 8
+#| fig.height: 7
+fig19 <- ggplot(g_scores_fmax, aes(x = freq_alt_complete, y = g_score)) +
+ geom_smooth(method = "lm", formula = 'y ~ x') +
+ geom_point(aes(color = strainID)) +
+ scale_color_manual(values = hambi_colors) +
+ facet_grid(prey_history ~ predator_history, labeller = label_both) +
+ labs(x = TeX("Maximum observed allele frequency ($f_{max}$) "), y = TeX("Extent of parallel evolution (Gene-wise $G_{i}$)"), color = NULL, shape = NULL) +
+ annotation_logticks(sides = "bl", color = "grey40") +
+ scale_x_log10() +
+ scale_y_log10() +
+ theme_bw() +
+ theme(
+ legend.position = "bottom",
+ strip.placement = 'outside',
+ strip.background = element_blank(),
+ panel.grid = element_blank())
+
+ggsave(here::here(figs, "G_score_fmax.svg"), fig19, width=7, height=6, units="in",
+ device="svg")
+
+fig19
+```
+
+:::
+
+Linear regression for different treatment combinations (over species). Overall, we find a clear positive relationship between the gene-wise extent of parallel evolution (G score) and the maximum observed allele frequency for that gene in 3/4 treatment combinations. We also find a positive relationship for the prey: evo | predator: anc treatment combination but the significance of this relationship is unclear (P value = 0.07).
+
+```{r}
+g_scores_fmax %>%
+ tidyr::nest(data = -c(prey_history, predator_history)) %>%
+ dplyr::mutate(model = map(data, \(df) lm(g_score ~ freq_alt_complete, na.rm = TRUE, data = df))) %>%
+ dplyr::mutate(tidy = map(model, broom::tidy)) %>%
+ unnest(tidy) %>%
+ dplyr::select(-data, -model) %>%
+ dplyr::filter(term != "(Intercept)") %>%
+ arrange(p.value)
+```
+
+Linear regression for all different species in different treatment combinations. By including every species we lose statistical power because some species have relatively few observations. However, for all 4 species in the prey: evo | predator: anc treatment combination we find a clear positive relationship between the gene-wise extent of parallel evolution (G score) and the maximum observed allele frequency for that gene. We also find this positive relationmship for HAMBI_2659 in the prey: evo | predator: evo treatment combination.
+
+```{r}
+g_scores_fmax %>%
+ tidyr::nest(data = -c(strainID, prey_history, predator_history)) %>%
+ dplyr::mutate(model = map(data, \(df) lm(g_score ~ freq_alt_complete, na.rm = TRUE, data = df))) %>%
+ dplyr::mutate(tidy = map(model, broom::tidy)) %>%
+ unnest(tidy) %>%
+ dplyr::select(-data, -model) %>%
+ dplyr::filter(term != "(Intercept)") %>%
+ arrange(p.value)
+```
diff --git a/R/metagenome/references.bib b/R/metagenome/references.bib
new file mode 100644
index 0000000..5944c81
--- /dev/null
+++ b/R/metagenome/references.bib
@@ -0,0 +1,88 @@
+
+@article{storey2003a,
+ title = {Statistical significance for genomewide studies},
+ author = {Storey, John D. and Tibshirani, Robert},
+ year = {2003},
+ month = {07},
+ date = {2003-07-25},
+ journal = {Proceedings of the National Academy of Sciences},
+ pages = {9440--9445},
+ volume = {100},
+ number = {16},
+ doi = {10.1073/pnas.1530509100},
+ url = {http://dx.doi.org/10.1073/pnas.1530509100},
+ langid = {en}
+}
+
+@article{tenaillon2016,
+ title = {Tempo and mode of genome evolution in a 50,000-generation experiment},
+ author = {Tenaillon, Olivier and Barrick, Jeffrey E. and Ribeck, Noah and Deatherage, Daniel E. and Blanchard, Jeffrey L. and Dasgupta, Aurko and Wu, Gabriel C. and Wielgoss, {Sébastien} and Cruveiller, {Stéphane} and {Médigue}, Claudine and Schneider, Dominique and Lenski, Richard E.},
+ year = {2016},
+ month = {08},
+ date = {2016-08},
+ journal = {Nature},
+ pages = {165--170},
+ volume = {536},
+ number = {7615},
+ doi = {10.1038/nature18959},
+ url = {http://dx.doi.org/10.1038/nature18959},
+ langid = {en}
+}
+
+@article{shoemaker2021,
+ title = {Molecular Evolutionary Dynamics of Energy Limited Microorganisms},
+ author = {Shoemaker, William R and Polezhaeva, Evgeniya and Givens, Kenzie B and Lennon, Jay T},
+ editor = {Battistuzzi, Fabia Ursula},
+ year = {2021},
+ month = {07},
+ date = {2021-07-13},
+ journal = {Molecular Biology and Evolution},
+ pages = {4532--4545},
+ volume = {38},
+ number = {10},
+ doi = {10.1093/molbev/msab195},
+ url = {http://dx.doi.org/10.1093/molbev/msab195},
+ langid = {en}
+}
+
+@article{good2017,
+ title = {The dynamics of molecular evolution over 60,000 generations},
+ author = {Good, Benjamin H. and McDonald, Michael J. and Barrick, Jeffrey E. and Lenski, Richard E. and Desai, Michael M.},
+ year = {2017},
+ month = {10},
+ date = {2017-10-18},
+ journal = {Nature},
+ pages = {45--50},
+ volume = {551},
+ number = {7678},
+ doi = {10.1038/nature24287},
+ url = {http://dx.doi.org/10.1038/nature24287},
+ langid = {en}
+}
+
+@article{durrant2020a,
+ title = {A Bioinformatic Analysis of Integrative Mobile Genetic Elements Highlights Their Role in Bacterial Adaptation},
+ author = {Durrant, Matthew G. and Li, Michelle M. and Siranosian, Benjamin A. and Montgomery, Stephen B. and Bhatt, Ami S.},
+ year = {2020},
+ month = {01},
+ date = {2020-01},
+ journal = {Cell Host & Microbe},
+ pages = {140--153.e9},
+ volume = {27},
+ number = {1},
+ doi = {10.1016/j.chom.2019.10.022},
+ url = {http://dx.doi.org/10.1016/j.chom.2019.10.022},
+ langid = {en}
+}
+
+@article{camargo2023,
+ title = {Identification of mobile genetic elements with geNomad},
+ author = {Camargo, Antonio Pedro and Roux, Simon and Schulz, Frederik and Babinski, Michal and Xu, Yan and Hu, Bin and Chain, Patrick S. G. and Nayfach, Stephen and Kyrpides, Nikos C.},
+ year = {2023},
+ month = {09},
+ date = {2023-09-21},
+ journal = {Nature Biotechnology},
+ doi = {10.1038/s41587-023-01953-y},
+ url = {http://dx.doi.org/10.1038/s41587-023-01953-y},
+ langid = {en}
+}
diff --git a/R/metagenome/utils_parallelism.R b/R/metagenome/utils_parallelism.R
new file mode 100644
index 0000000..326073d
--- /dev/null
+++ b/R/metagenome/utils_parallelism.R
@@ -0,0 +1,654 @@
+
+# Nucleotide parallelism --------------------------------------------------
+
+plot_nuc_survival <- function(df, ylim, ybreaks, ncol){
+ ggplot(df, aes(x=m, y=value, color=name, linetype = name)) +
+ geom_step(direction = "mid") +
+ scale_x_continuous(limits = c(0.5, 3.5),
+ breaks = c(1, 2, 3),
+ minor_breaks = NULL) +
+ scale_y_log10(breaks = {{ ybreaks }},
+ labels = scales::trans_format("log10", scales::math_format(10^.x))) +
+ labs(x=TeX("Nucleotide multiplicitiy, \\textit{m}"),
+ y=TeX("Total mutations $\\geq \\textit{m}$"),
+ color = NULL) +
+ scale_color_manual(values = c("blue", "red")) +
+ scale_linetype(guide = 'none') +
+ facet_wrap(~label, ncol = {{ ncol }}) +
+ annotation_logticks(sides = "l", color = "grey40") +
+ coord_cartesian(ylim = c(1, {{ ylim }})) +
+ theme_bw() +
+ theme(
+ strip.placement = 'outside',
+ strip.background = element_blank(),
+ panel.grid = element_blank(),
+ legend.position = "bottom"
+ )
+}
+
+# Gene parallelism --------------------------------------------------------
+
+# Prepare different data sources ------------------------------------------
+
+make_gene_length_df <- function(df, strainID){
+ # takes a combined dataframe for all HAMBI species total length of gene-wise
+ # length of nonsynonymous coding sites and returns a reduced dataframe for the
+ # focal species strainID
+ df %>%
+ # filter on strainID
+ filter(strainID == !!strainID) %>%
+ dplyr::select(locus_tag, l_i = ns_length) %>%
+ # remove NAs
+ drop_na() %>%
+ # convert to data.frame
+ column_to_rownames(var = "locus_tag") %>%
+ data.frame()
+}
+
+make_raw_mut_df <- function(df, gene_lengths) {
+ # formats and prepares a dataframe of observed pass gene hits per replicate
+ # and includes every CDS for the genome even if there were no hits to the gene
+ # (ie some rowSums are zero). Rownames are locus_tag and columns are the
+ # individual replicates where at least 1 mutation was detected
+ df %>%
+ # first check to remove any noncoding mutations
+ filter(locus_tag %in% rownames(gene_lengths)) %>%
+ summarize(n = n(),
+ .by = c("replicate", "locus_tag")) %>%
+ pivot_wider(
+ names_from = "replicate",
+ values_from = "n",
+ values_fill = 0
+ ) %>%
+ left_join(dplyr::select(
+ rownames_to_column(gene_lengths, var = "locus_tag"),
+ locus_tag
+ ),
+ .,
+ by = join_by(locus_tag)) %>%
+ mutate(across(-locus_tag, ~ replace_na(.x, 0))) %>%
+ column_to_rownames(var = "locus_tag") %>%
+ data.frame()
+}
+
+make_multiplicity_df <- function(df, gene_lengths) {
+ # calculate gene multiplicity for each gene
+ # m_i = n_i * mean(l_i)/l_i
+ df %>%
+ # first check to remove any noncoding mutations
+ filter(locus_tag %in% rownames(gene_lengths)) %>%
+ summarize(
+ n_i = n(),
+ n_replicate = n_distinct(replicate),
+ .by = c("locus_tag")
+ ) %>%
+ select(locus_tag, n_i, n_replicate) %>%
+ left_join(rownames_to_column(gene_lengths, var = "locus_tag"),
+ .,
+ by = join_by(locus_tag)) %>%
+ replace_na(list(n_i = 0, n_replicate = 0)) %>%
+ mutate(m_i = n_i * mean(l_i) / l_i) %>%
+ column_to_rownames(var = "locus_tag") %>%
+ data.frame()
+}
+
+prepare_all_input <- function(df, gene_table){
+ # prepares a list with the required dataframes and summary values to compute the
+ # gscore or to simulate the G score
+
+ strainID <- unique(df$strainID)
+ gene_lengths <- make_gene_length_df(gene_table, strainID)
+ raw_mut_df <- make_raw_mut_df(df, gene_lengths)
+ multiplicity_df <- make_multiplicity_df(df, gene_lengths)
+
+ return(lst(
+ strainID = strainID,
+ # number of replicates with at least on gene hit
+ rep_num = ncol(raw_mut_df),
+ # total gene hits in each replicate
+ mut_reps = colSums(raw_mut_df),
+ # total gene hits across all genes and replicates
+ mut_tot = sum(colSums(raw_mut_df)),
+ gene_lengths = gene_lengths,
+ multiplicity_df = multiplicity_df,
+ raw_mut_df = raw_mut_df
+ ))
+}
+
+# G scores ----------------------------------------------------------------
+
+G_score <- function(input_list){
+ # number of non-syn sites in gene i
+ l_i <- input_list$gene_lengths$l_i
+ # compute the sum of non-syn sites in genome
+ Ltot <- sum(l_i)
+ # compute the average non-syn sites per gene
+ Lavg <- mean(l_i)
+ # compute the number of genes in the genome
+ n_genes <- length(l_i)
+ # number of mutations in gene i across all replicates
+ n_i <- input_list$multiplicity_df$n_i
+ # total number of non-syn hits across genes and replicates
+ Ntot <- input_list$mut_tot
+ # compute the expected multiplicity value
+ m_exp <- Ntot / n_genes
+ # compute gene multiplicity for gene i
+ m_i <- n_i * Lavg / l_i
+ # compute the G score for each gene
+ g_score <- n_i * log(m_i / m_exp)
+ # net increase in log-likelihood compared to the null model of m_i / m_exp =1
+ # normalized to the total mutations
+ delta_l <- sum(g_score, na.rm=T) / Ntot
+ return(delta_l)
+}
+
+G_score_prep <- function(input_list){
+ # Special G score function preparing constants for use in G_score_resampled
+ # so we don't have to unnecessarily recalculate all the constants for each
+ # resample
+
+ # number of non-syn sites in gene i
+ l_i <- input_list$gene_lengths$l_i
+ # compute the sum of non-syn sites in genome
+ Ltot <- sum(l_i)
+ # compute the average non-syn sites per gene
+ Lavg <- mean(l_i)
+ # Null probability of non-syn hit on a gene based only on the
+ # number of nonsynonymous sites per gene relative to the total number of
+ # nonsynonymous sites in the genome (gene_probs = l_i/Ltot).
+ gene_probs <- l_i/Ltot
+ # number of genes in genome. In the simulation this only includes coding
+ # sequences and so we don't need to filter out any genes
+ n_genes <- length(l_i)
+ # total number of non-syn hits across genes and replicates
+ Ntot <- input_list$mut_tot
+ # compute the expected multiplicity value
+ m_exp <- Ntot / n_genes
+
+ lst(l_i=l_i, Ltot=Ltot, Lavg=Lavg, n_genes=n_genes,
+ Ntot=Ntot, m_exp=m_exp, gene_probs=gene_probs)
+}
+
+G_score_resampled <- function(df, input_list){
+ # remaining G score function that uses constants from G_score_prep while only
+ # calculating n_i dynamically
+
+ # number of mutations in gene i summed across all replicates
+ n_i <- rowSums(df)
+ # compute gene multiplicity for gene i
+ m_i = n_i * input_list$Lavg / input_list$l_i
+ # compute the G score for each gene
+ g_score = n_i * log(m_i / input_list$m_exp)
+ # net increase in log-likelihood compared to the null model of m_i / m_exp =1
+ # normalized to the total mutations
+ delta_l = sum(g_score, na.rm=T) / input_list$Ntot
+}
+
+# G score permutation -----------------------------------------------------
+
+mut_hit_sim <- function(mut_reps, gene_probs, resamples){
+ # randomly sample (n=resamples) the observed number of nonsynonymous mutations
+ # from each replicate (vector mut_reps) with probability based only on the
+ # number of nonsynonymous sites per gene relative to the total number of
+ # nonsynonymous sites in the genome (gene_probs = l_i/Ltot). Returns a list of
+ # length mut_reps where for each replicate in mut_reps there is a n=resamples
+ # column and n=gene_number row matrix
+ map(mut_reps, \(x) rmultinom(n = resamples, size = x, prob = gene_probs))
+}
+
+convert_nameless_mat2_df <- function(m, pop_num, gene_names){
+ # take output from mut_hit_sim function (list of sample x gene matrices) and
+ # reformat it so that it returns a dataframe with locus_tag for each gene as
+ # rownames and sampled replicates are columns
+ names(m) <- LETTERS[seq_along(1: pop_num)]
+ data.frame(m, row.names = gene_names)
+}
+
+gscore_p_value <- function(input_list, emp_g_score, samples){
+ G_score_prepped <- G_score_prep(input_list)
+ # randomly sample hits based on gene length and split to a convenient form
+ multinomhitsamp_split <- mapply(asplit,
+ MARGIN = 2,
+ x = mut_hit_sim(input_list$mut_reps,
+ G_score_prepped$gene_probs,
+ samples))
+ # for each random sample convert to a more useful dataframe form
+ g_reps <- seq_along(1:samples) %>%
+ map(\(x) convert_nameless_mat2_df(multinomhitsamp_split[x,],
+ input_list$rep_num,
+ rownames(input_list$gene_lengths))) %>%
+ # calculate the g_score for each sample
+ map_dbl(\(x) G_score_resampled(x, G_score_prepped))
+
+ return(length(g_reps[g_reps > emp_g_score]) / length(g_reps))
+}
+
+
+gscore_p_value2 <- function(input_list, emp_g_score, samples){
+ # also computes the observed distribution of hits per locus tag
+ # See Fig S3 from this paper: https://doi.org/10.1038/s41559-020-1128-3
+ G_score_prepped <- G_score_prep(input_list)
+ # randomly sample hits based on gene length and split to a convenient form
+ multinomhitsamp_split <- mapply(asplit,
+ MARGIN = 2,
+ x = mut_hit_sim(input_list$mut_reps,
+ G_score_prepped$gene_probs,
+ samples))
+ # for each random sample convert to a more useful dataframe form
+ multinomhitsamp_split_df <- seq_along(1:samples) %>%
+ map(\(x) convert_nameless_mat2_df(multinomhitsamp_split[x,],
+ input_list$rep_num,
+ rownames(input_list$gene_lengths)))
+ # get the observed distribution of "hits" i.e. non-syn mutations
+ observed_hit_distrubution <- input_list$raw_mut_df %>%
+ rownames_to_column(var = "locus_tag") %>%
+ pivot_longer(-locus_tag) %>%
+ filter(value > 0) %>%
+ summarize(n_reps = n_distinct(name), .by = c(locus_tag)) %>%
+ summarize(n = n(), .by = c(n_reps)) %>%
+ mutate(type = "Observed")
+
+ null_hit_distribution <- multinomhitsamp_split_df %>%
+ map_dfr(\(x) pivot_longer(rownames_to_column(x, var = "locus_tag"), -locus_tag), .id = "resample") %>%
+ filter(value > 0) %>%
+ summarize(n_reps = n_distinct(name), .by = c(resample, locus_tag)) %>%
+ summarize(n = n()/samples, .by = c(n_reps)) %>%
+ mutate(type = "Null")
+
+ # calculate the g_score for each sample
+ g_reps <- map_dbl(multinomhitsamp_split_df, \(x) G_score_resampled(x, G_score_prepped))
+
+ # calculate p values
+ g_score_pvalue <- length(g_reps[g_reps > emp_g_score]) / length(g_reps)
+
+ return(lst(g_score_pvalue = g_score_pvalue,
+ hit_distribution = bind_rows(observed_hit_distrubution, null_hit_distribution)))
+}
+
+# Calculating gene-wise significance --------------------------------------
+
+from_parallelism_statistics <- function(input_list) {
+ # number of non-syn sites in gene i
+ l_i <- input_list$gene_lengths$l_i
+ # compute the sum of non-syn sites in genome
+ Ltot <- sum(l_i)
+ # number of mutations in gene i across all replicates
+ n_i <- input_list$multiplicity_df$n_i
+ # total number of non-syn hits across genes and replicates
+ Ntot <- input_list$mut_tot
+ # compute the fraction of total non-syn sites occupied by each gene
+ ps <- l_i / Ltot
+ # compute the expected number of mutations in each gene based on gene length
+ n_exp <- Ntot * l_i / Ltot
+
+ lst(
+ "gene_names" = rownames(input_list$gene_lengths),
+ "Ls" = l_i,
+ "ns" = n_i,
+ "Ltot" = Ltot,
+ "Ntot" = Ntot,
+ "ps" = ps,
+ "n_exp" = n_exp
+ )
+}
+
+calculate_poisson_log_survival <- function(n_exp, ns, ethresh, filter = FALSE){
+ # survival function = 1-CDF
+ survivals <- (1 - ppois(ns - 0.1, n_exp))
+ # for the case if there are so many mutations in some gene that its survival is 0
+ if (filter) {
+ survivals[survivals==0] <- min(survivals[survivals>0])/1e10
+ }
+ logpvalues <- rep(0, each = length(survivals))
+ logpvalues[survivals > ethresh] = -log(survivals[survivals > ethresh])
+ logpvalues[survivals <= ethresh] = -(ns * log(ns / n_exp +
+ (ns == 0)) + ns - n_exp)[survivals <= ethresh]
+ return(logpvalues)
+}
+
+calculate_unnormalized_survival_from_vector <- function(xs, min_x=NULL, max_x=NULL, min_p=1e-10){
+ if (is.null(min_x)){
+ min_x <- min(xs)-1
+ }
+ if (is.null(max_x)){
+ max_x <- max(xs)+1
+ }
+ unique_xs <- unique(xs)
+ unique_xs <- sort(append(unique_xs, c(min_x, max_x)))
+ num_observations <- map_dbl(unique_xs, \(x) sum(xs >= x))
+
+ # So that we can plot CDF, SF on log scale
+ num_observations[0] <- num_observations[0] - min_p
+ num_observations[1] <- num_observations[1] - min_p
+ num_observations[-1] <- num_observations[-1] + min_p
+
+ return(list(unique_xs, num_observations))
+}
+
+map_pval_and_probs = function(observed_p, logpvalues, probabilities, ns_grid, nmin){
+ sum(map2_dbl(logpvalues, probabilities,
+ \(x, y) sum((x >= observed_p) * (ns_grid >= nmin) * y)))
+}
+
+format_survivalcurves_2plot <- function(observed_ps, survivals, observed_pvalue_survival,
+ pstar, num_significant){
+ df_step <- tibble(
+ obs_p = observed_ps,
+ Survival_Expected = survivals,
+ Survival_Observed = observed_pvalue_survival
+ ) %>%
+ pivot_longer(-obs_p) %>%
+ mutate(name = factor(
+ name,
+ levels = c("Survival_Observed", "Survival_Expected"),
+ labels = c("Survival Observed", "Survival Expected")
+ ))
+
+ df_seg <- tibble(
+ x = c(min(observed_ps), pstar),
+ y = c(num_significant, 0),
+ xend = c(pstar, pstar),
+ yend = c(num_significant, num_significant)
+ )
+
+ df_ann <- tibble(x = pstar, y = num_significant)
+
+ # return
+ lst(df_step = df_step,
+ df_seg = df_seg,
+ df_ann = df_ann
+ )
+}
+
+
+make_gene_table <- function(pooled_pvalues, strainID, obs_gene_multiplicity, raw_hits, pstar){
+ enframe(pooled_pvalues, name = "locus_tag", value = "neg_log10P") %>%
+ mutate(strainID = strainID, pstar = pstar) %>%
+ left_join(., rownames_to_column(obs_gene_multiplicity, var = "locus_tag"), join_by(locus_tag)) %>%
+ left_join(., rownames_to_column(raw_hits, var = "locus_tag"), join_by(locus_tag)) %>%
+ dplyr::rename(nonsyn_sites = l_i,
+ observed_hits_n_i = n_i,
+ gene_multiplicity_m_i = m_i)
+}
+
+NullGeneLogpSurvivalFunction <- function(mylist, nmin=3){
+
+ # Get formatted input
+ fps <- from_parallelism_statistics(mylist)
+
+ # Calculate log-P value for each gene
+ logpvalues_par <- calculate_poisson_log_survival(fps$n_exp, fps$ns,
+ ethresh=1e-30, filter = TRUE)
+ names(logpvalues_par) <- fps$gene_names
+
+ # Collect P values for genes with at least nmin (default = 3) mutations across
+ # all populations this limits the number of low P values driven primarily by
+ # gene length
+ pooled_pvalues <- sort(logpvalues_par[fps$ns >= nmin])
+
+ # Un-normalized poisson survival vector
+ tsrv <- calculate_unnormalized_survival_from_vector(pooled_pvalues)
+ observed_ps <- pluck(tsrv, 1)
+ observed_pvalue_survival <- pluck(tsrv, 2)
+
+ # get required inputs for the survival analysis to estimate significance
+ ns_grid <- rep(0:399)
+ alpha <- 0.05
+
+ logpvalues <- map(fps$n_exp, \(x) calculate_poisson_log_survival(x, ns = ns_grid, ethresh = 1e-20, filter = FALSE))
+ logprobabilities <- map(fps$n_exp, \(x) ns_grid * log(x) - lgamma(ns_grid + 1) - x)
+ probabilities <- map(logprobabilities, exp)
+ survivals <- map_dbl(observed_ps, \(x) map_pval_and_probs(x, logpvalues, probabilities, ns_grid, nmin))
+ threshold_idx <- which(survivals/observed_pvalue_survival <= alpha)[1]
+ pstar <- observed_ps[threshold_idx]
+ num_significant <- observed_pvalue_survival[threshold_idx]
+
+ # Return pertinent results
+ df2plot <- format_survivalcurves_2plot(observed_ps, survivals, observed_pvalue_survival,
+ pstar, num_significant)
+ gene_table <- make_gene_table(pooled_pvalues, mylist$strainID, mylist$multiplicity_df, mylist$raw_mut_df, pstar)
+
+ return(lst("df2plot" = df2plot,
+ "gene_table" = gene_table,
+ "observed_ps" = observed_ps,
+ "pooled_pvalues" = pooled_pvalues,
+ "observed_pvalue_survival" = observed_pvalue_survival,
+ "survivals" = survivals,
+ "pstar" = pstar,
+ "num_significant" = num_significant))
+
+}
+
+plotcurve <- function(input_list){
+ ggplot() +
+ geom_segment(
+ data = input_list$df_seg,
+ aes(
+ x = x,
+ y = y,
+ xend = xend,
+ yend = yend
+ ),
+ linetype = "dashed",
+ linewidth = 0.25
+ ) +
+ geom_step(data = input_list$df_step, aes(x = obs_p, y = value, color = name)) +
+ annotate("point", x = input_list$df_ann$x, y = input_list$df_ann$y) +
+ # annotate(
+ # "text",
+ # x = input_list$df_ann$x + 35,
+ # y = input_list$df_ann$y - 3,
+ # label = TeX("Critical P*\nthreshold = $1.6 \\times 10^{-4}$")
+ # ) +
+ labs(x = TeX("$-log_{10}P$"), y = "Number of genes") +
+ scale_color_manual(values = c("blue", "grey")) +
+ scale_y_log10(
+ breaks = c(1, 10, 100),
+ labels = scales::trans_format("log10", scales::math_format(10 ^.x))
+ ) +
+ annotation_logticks(sides = "l", color = "grey40") +
+ coord_cartesian(ylim = c(1, 1000)) +
+ theme_bw() +
+ theme(
+ strip.background = element_blank(),
+ panel.grid = element_blank(),
+ )
+}
+
+run_full_gene_parallelism_pipeline <- function(grouped, gene_lens, resamples){
+ # grouped is a tibble that has been grouped at the desired hierarchy level
+ # from the experiment. It should also contain a variable called "groupid" that
+ # is numeric variable unique to each group. The tibble gene_lens contains the
+ # number of non-synonymous sites found in each locus tag for all 23 HAMBI
+ # species in a variable called "ns_length." the variable "resamples is the
+ # number of resamplings you want to perform for the significance testing of
+ # the G score
+
+ # split into list of tibbles
+ grouped_split <- group_split(grouped)
+
+ # prepare ecah grouped input for gene parallelism analysis
+ grouped_split_prepped <- grouped_split %>%
+ map(\(x) prepare_all_input(x, gene_lens), .progress="inputprep")
+
+ # Calculate empirical G scores for non-syn mutations
+ emp_g_scores <- map(grouped_split_prepped, \(x) G_score(x), .progress="gscore")
+
+ # Calculate significance of empirical G scores
+ #g_score_pvals_lump <- map2_dbl(grouped_split_prepped, emp_g_scores, \(x, y) gscore_p_value(x, y, resamples))
+ g_score_pvals_lump2 <- map2(grouped_split_prepped, emp_g_scores, \(x, y) gscore_p_value2(x, y, resamples), .progress="gscore_pval")
+
+ # Estimate gene-wise p values and compare to null dependending only on total hits and gene length
+ nglpsf_out <- map(grouped_split_prepped, \(x) NullGeneLogpSurvivalFunction(x, nmin = 1), .progress="LogPSurvival")
+
+ # format G scores into a tibble
+ output_gscore <-
+ unnest(enframe(emp_g_scores, name = "groupid", value = "observed_g_score"), cols = observed_g_score) %>%
+ left_join(unnest(enframe(map(g_score_pvals_lump2, \(x) pluck(x, 1)), name = "groupid", value = "pvalue"), cols = "pvalue")) %>%
+ left_join(group_keys(grouped))
+
+ output_simulated_reps <- g_score_pvals_lump2 %>%
+ map(\(x) pluck(x, 2)) %>%
+ map_dfr(\(x) x, .id = "groupid") %>%
+ mutate(groupid = as.integer(groupid)) %>%
+ left_join(group_keys(grouped))
+
+ output_gene_table <- map_dfr(nglpsf_out, \(x) pluck(x, "gene_table"), .id = "groupid") %>%
+ mutate(groupid = as.integer(groupid)) %>%
+ left_join(group_keys(grouped))
+
+ getdfl2plot <- map(nglpsf_out, \(x) pluck(x, "df2plot"))
+ output_df2plot <- map_dfr(getdfl2plot, \(x) pluck(x, "df_step"), .id = "groupid") %>%
+ left_join(map_dfr(getdfl2plot, \(x) pluck(x, "df_seg"), .id = "groupid")) %>%
+ mutate(groupid = as.integer(groupid)) %>%
+ left_join(group_keys(grouped))
+
+ return(lst(output_gscore = output_gscore,
+ output_simulated_reps = output_simulated_reps,
+ output_gene_table = output_gene_table,
+ output_df2plot = output_df2plot))
+}
+
+plot_sims <- function(input_list, page){
+ input_list$output_simulated_reps %>%
+ mutate(label = paste(strainID, "prey:", prey_history, "\npredator:", predator_history)) %>%
+ mutate(n_reps = factor(n_reps),
+ labels = label) %>%
+ ggplot() +
+ geom_col(aes(x = n_reps, y = n, fill = type),
+ position = position_dodge2(preserve = "single")) +
+ labs(x = "Number of replicates with a hit in the same gene", y = "Number of genes with a hit") +
+ scale_y_sqrt() +
+ scale_fill_brewer(palette = "Set1") +
+ facet_wrap_paginate(vars(label), scales = "free_y",
+ labeller = labeller(label_wrap_gen()),
+ ncol = 4, nrow = 4, page = {{ page }}) +
+ theme_bw() +
+ theme(
+ strip.placement = 'outside',
+ strip.background = element_blank(),
+ panel.grid = element_blank(),
+ legend.position = "bottom"
+ )
+}
+
+# Functions for Jaccard index resampling ----------------------------------
+
+prep4jaccard_resample <- function(gene_table_filt, genetable, strainID, ...){
+ mygroup <- gene_table_filt %>%
+ filter(strainID == {{ strainID }}) %>%
+ group_by(...) %>%
+ mutate(groupid = cur_group_id())
+
+ mysplit <- mygroup %>%
+ group_split() %>%
+ map(\(x) pull(x, locus_tag))
+
+ genetable_filt <- genetable %>%
+ filter(strainID == {{ strainID }}) %>%
+ dplyr::select(locus_tag, ns_length) %>%
+ # remove NAs
+ drop_na() %>%
+ arrange(locus_tag)
+
+ locus_tag <- pull(genetable_filt, locus_tag)
+ l_i <- pull(genetable_filt, ns_length)
+ gene_probs <- l_i/sum(l_i)
+
+ return(
+ lst(mygroup, mysplit, locus_tag, gene_probs)
+ )
+}
+
+jaccard <- function(a, b) {
+ intersection <- length(intersect(a, b))
+ union <- length(a) + length(b) - intersection
+ return (intersection/union)
+}
+
+pairsamplejaccard <- function(x, y, locus_tag, gene_probs){
+ jaccard(
+ sample(x = locus_tag, size = length(x), prob = gene_probs, replace = FALSE),
+ sample(x = locus_tag, size = length(y), prob = gene_probs, replace = FALSE)
+ )
+}
+
+resample_jaccard <- function(nresamples, prepped4jaccard_resample) {
+
+ # defining inputs
+ input_gene_lists <- prepped4jaccard_resample$mysplit
+ genome_locus_tags <- prepped4jaccard_resample$locus_tag
+ genome_gene_probs <- prepped4jaccard_resample$gene_probs
+
+ # some housekeeping
+ myindxgroups <- prepped4jaccard_resample$mygroup %>%
+ ungroup() %>%
+ distinct(groupid, prey_history, predator_history) %>%
+ mutate(groupname = paste0("bact: ", prey_history, ", predator: ", predator_history)) %>%
+ dplyr::select(groupid, groupname)
+
+ # generate unique combinations
+ combos <- combn(1:length(input_gene_lists), 2)
+
+ # convert combo ids to variable names for pairs
+ myindxgroupnames <- as.data.frame(t(combos)) %>%
+ `colnames<-`(c("groupid1", "groupid2")) %>%
+ rownames_to_column(var = "treat_pair") %>%
+ mutate(treat_pair = as.integer(treat_pair)) %>%
+ left_join(rename(myindxgroups, groupid1 = groupid, groupname1 = groupname),
+ by = join_by(groupid1)) %>%
+ left_join(rename(myindxgroups, groupid2 = groupid, groupname2 = groupname),
+ by = join_by(groupid2))
+
+ # First inspect observed jaccard indexes for all pairs
+ jc_obs <-
+ map_dbl(asplit(combos, MARGIN = 2),
+ \(x) jaccard(input_gene_lists[[x[1]]],
+ input_gene_lists[[x[2]]])
+ )
+
+ # get only combos with a Jaccard index > 0 (i.e. any overlapping elements)
+ jc_obs_gt0 <- jc_obs[which(jc_obs > 0)]
+
+ if (length(jc_obs_gt0) > 1) {
+ ind2iterate <- asplit(combos[, which(jc_obs_gt0 > 0)], MARGIN = 2)
+ } else {
+ ind2iterate <- 1
+ }
+
+ # resampled jaccard index
+ jc_resampled <- replicate(nresamples,
+ map_dbl(
+ ind2iterate,
+ \(x) pairsamplejaccard(input_gene_lists[[x[1]]],
+ input_gene_lists[[x[2]]],
+ genome_locus_tags,
+ genome_gene_probs)
+ )
+ )
+
+ if (length(jc_obs_gt0) > 1) {
+ # Vector of p values for each combination with multiple comparisons
+ ps <- tibble(treat_pair = which(jc_obs > 0), p_values = rowSums(jc_resampled >= jc_obs_gt0) / rep(nresamples, length(jc_obs_gt0)))
+ } else {
+ # single p value for the single sample case
+ ps <- tibble(treat_pair = which(jc_obs > 0), p_values = length(jc_resampled[jc_resampled >= jc_obs_gt0]) / nresamples)
+ }
+ left_join(ps, myindxgroupnames)
+}
+
+# hypergeometric test -----------------------------------------------------
+
+enricher <- function(n_background, n_par, n_cog_background, n_cog_par, over=TRUE){
+ # n_background: total genes with COG annotation
+ # n_par: Set of target genes of interest. In this case the total number of COG
+ # annotated genes that have significant parallelism
+ # n_cog_background: set of genes from background annotated with the COG of interest
+ # n_cog_par: number of genes in a COG of interest from the total number of COG annotated
+ # genes w/ parallelism
+ if (over) {
+ phyper(n_cog_par-1, n_cog_background, n_background-n_cog_background, n_par, lower.tail= FALSE)
+ } else {
+ phyper(n_cog_par, n_cog_background, n_background-n_cog_background, n_par, lower.tail= TRUE)
+ }
+}
diff --git a/R/utils_generic.R b/R/utils_generic.R
new file mode 100644
index 0000000..7852fdb
--- /dev/null
+++ b/R/utils_generic.R
@@ -0,0 +1,107 @@
+tax <- tibble::tribble(
+ ~strainID, ~genus, ~species,
+ "HAMBI_0006", "Pseudomonas", "putida",
+ "HAMBI_0097", "Acinetobacter", "lwoffii",
+ "HAMBI_0105", "Agrobacterium", "tumefaciens",
+ "HAMBI_0262", "Brevundimonas", "bullata",
+ "HAMBI_0403", "Comamonas", "testosteroni",
+ "HAMBI_1279", "Hafnia", "alvei",
+ "HAMBI_1287", "Citrobacter", "koseri",
+ "HAMBI_1292", "Morganella", "morganii",
+ "HAMBI_1299", "Kluyvera", "intermedia",
+ "HAMBI_1842", "Sphingobium", "yanoikuyae",
+ "HAMBI_1896", "Sphingobacterium", "spiritivorum",
+ "HAMBI_1923", "Myroides", "odoratus",
+ "HAMBI_1972", "Aeromonas", "caviae",
+ "HAMBI_1977", "Pseudomonas", "chlororaphis",
+ "HAMBI_1988", "Chitinophaga", "sancti",
+ "HAMBI_2159", "Paraburkholderia", "caryophylli",
+ "HAMBI_2160", "Bordetella", "avium",
+ "HAMBI_2164", "Cupriavidus", "necator",
+ "HAMBI_2443", "Paracoccus", "denitrificans",
+ "HAMBI_2494", "Paraburkholderia", "kururiensis",
+ "HAMBI_2659", "Stenotrophomonas", "maltophilia",
+ "HAMBI_2792", "Moraxella", "canis",
+ "HAMBI_3031", "Niabella", "yanshanensis",
+ "HAMBI_3237", "Microvirga", "lotononidis"
+)
+
+rRNA_counts <- tibble::tribble(
+ ~strainID, ~copies,
+ "HAMBI_0006", 7L,
+ "HAMBI_0097", 7L,
+ "HAMBI_0105", 4L,
+ "HAMBI_0262", 3L,
+ "HAMBI_0403", 9L,
+ "HAMBI_1279", 7L,
+ "HAMBI_1287", 7L,
+ "HAMBI_1292", 7L,
+ "HAMBI_1299", 8L,
+ "HAMBI_1842", 4L,
+ "HAMBI_1896", 4L,
+ "HAMBI_1923", 6L,
+ "HAMBI_1972", 10L,
+ "HAMBI_1977", 5L,
+ "HAMBI_1988", 5L,
+ "HAMBI_2159", 4L,
+ "HAMBI_2160", 3L,
+ "HAMBI_2164", 5L,
+ "HAMBI_2443", 3L,
+ "HAMBI_2494", 4L,
+ "HAMBI_2659", 4L,
+ "HAMBI_2792", 4L,
+ "HAMBI_3031", 2L,
+ "HAMBI_3237", 6L
+ )
+
+hambi_colors <- c(`HAMBI_1287` = "#6DA3EE", `HAMBI_1977` = "#F1C82E", `HAMBI_0403` = "#FB8A5C",
+ `HAMBI_2659` = "#624090", `HAMBI_1972` = "#26818E", `HAMBI_1292` = "#32702C",
+ `HAMBI_1923` = "#FC2EDB", `HAMBI_1279` = "#FE1C35", `HAMBI_1299` = "#BEEF60",
+ `HAMBI_1896` = "#1C26FB", `HAMBI_0097` = "#F0DACB", `HAMBI_1988` = "#D71C76",
+ `HAMBI_0006` = "#870DAE", `HAMBI_2792` = "#0DFE32", `HAMBI_3031` = "#FC94D1",
+ `HAMBI_2160` = "#16EEA5", `HAMBI_0105` = "#00F5F7", `HAMBI_3237` = "#956616",
+ `HAMBI_2494` = "#DABBF3", `HAMBI_2164` = "#DF26FD", `HAMBI_0262` = "#8D324F",
+ `HAMBI_2443` = "#9DE0C5", `HAMBI_2159` = "#92950D", `HAMBI_1842` = "#F489FD"
+)
+
+# opposite of %in% fuction
+`%nin%` = Negate(`%in%`)
+
+# logit transform
+logit = function(x){
+ log(x/(1-x))
+}
+
+minnz = function(V) {
+ # Calculates the smallest value of the vector except for 0 (non-zero minumum)
+ # Argument: vector
+ C <- NULL # prepare
+ k <- length(V) # count to
+ for (i in 1:k) { # check all
+ if ((V[i] == 0) == FALSE) (C[i] <- V[i]) else (C[i] <- 9999919) # if V[i] is not 0, add it to C
+ }
+ m <- min(C) # minimum of V, not counting 0
+ if (max(V) == 1) (m <- 1) # fix for binary vectors (0,1)
+ if (m == 9999919) (warning("Error: Minimum calculation failed.")) # warning because of hard-coded replacement
+ return(m)
+ }
+
+quibble95 = function(x, q = c(0.025, 0.5, 0.975)) {
+ tibble(x = quantile(x, q), quantile = c("q2.5", "q50", "q97.5"))
+}
+
+`%nin%` <- Negate(`%in%`)
+
+getfiles <- function(dir, pattern){
+ list.files(dir, full.names = TRUE, pattern = pattern)
+}
+
+getnames <- function(fileslist, myregex){
+ set_names(fileslist,
+ str_extract(
+ fileslist,
+ regex(myregex)
+ ))
+}
+
+# "(?<=[/])([^/]+)(?=(_[:alpha:]+_[:alpha:]+\\.)[^.]+)|(?<=[/])([^/]+)(?=(_[:alpha:]+\\.)[^.]+)"
diff --git a/R/wgs/01_format_variants.qmd b/R/wgs/01_format_variants.qmd
new file mode 100644
index 0000000..cd78487
--- /dev/null
+++ b/R/wgs/01_format_variants.qmd
@@ -0,0 +1,168 @@
+---
+title: "Format WGS variants"
+author: "Shane Hogle"
+date: today
+abstract: "This notebook formats for downstream use the table and VCF output from mutect2 called on WGS population data in the evolved bacteria inoculum. These are species that had been experimentally evolved before the main match/mis-match experiment. The formatting removes some variants with spurious features and connects variants to functional annotations and gene identities."
+---
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+library(here)
+library(tidyverse)
+library(withr)
+library(fs)
+library(archive)
+source(here::here("R", "utils_generic.R"))
+
+# Set up some directories -------------------------------------------------
+data_raw <- here::here("_data_raw", "wgs")
+vcftar <- here::here(data_raw, "tables.tar.gz")
+
+# create temporary location to decompress
+tmpdir <- fs::file_temp()
+
+# make processed data directory if it doesn't exist
+data <- here::here("data", "wgs")
+fs::dir_create(data)
+
+# untar directory containing tables
+archive::archive_extract(
+ vcftar,
+ dir = tmpdir,
+ files = NULL,
+ options = character(),
+ strip_components = 0L
+)
+
+vcfdir <- here::here(tmpdir, "tables")
+
+# Read variant SnpEff annotations -----------------------------------------
+
+snpefffiles <- fs::dir_ls(
+ path = vcfdir,
+ all = FALSE,
+ recurse = TRUE,
+ type = "file",
+ glob = "*.snpeff.tsv",
+ regexp = NULL,
+ invert = FALSE,
+ fail = TRUE
+)
+
+snpeffslurped <- readr::read_tsv(
+ snpefffiles,
+ skip = 1,
+ col_names = c(
+ "chrom",
+ "pos",
+ "ref",
+ "alt",
+ "filter",
+ # Annotated using Sequence Ontology terms. Multiple effects can be concatenated using '&'.
+ "effect",
+ # A simple estimation of putative impact / deleteriousness : HIGH, MODERATE, LOW, or MODIFIER
+ "impact",
+ # gene id from prokka
+ "locus_tag",
+ # whether variant is coding or noncoding
+ "biotype",
+ # Variant using HGVS notation (DNA level)
+ "hgvs_c",
+ # Variant using HGVS notation (Protein level).
+ "hgvs_p",
+ # Position in coding bases (one based includes START and STOP codons).
+ "cds_pos",
+ # Total number of coding bases (one based includes START and STOP codons).
+ "cds_len",
+ # Position in translated product (one based includes START and STOP codons).
+ "aa_pos",
+ # Total length of translated product (one based includes START and STOP codons).
+ "aa_len"
+ ),
+ col_types = c("cdcccccccccdddd"),
+ id = "sample"
+)
+
+snpeffslurpedfmt <- snpeffslurped %>%
+ # this is weird, but joining on column like filter that contains characters like ';' doesnt work
+ dplyr::select(-filter) %>%
+ mutate(sample = str_extract(sample,
+ regex("(HAMBI_[:digit:]{4}-[:alpha:]-[:alnum:]{3})|(HAMBI_[:digit:]{4}-[:alpha:]-SH-WGS-[:digit:]{3})"))) %>%
+ separate(sample, into = c("strainID", "replicate", "sample"), sep="-", extra = "merge")
+
+# Read variant frequencies ------------------------------------------------
+
+varfiles <- fs::dir_ls(
+ path = vcfdir,
+ all = FALSE,
+ recurse = TRUE,
+ type = "file",
+ glob = "*.variants.tsv",
+ regexp = NULL,
+ invert = FALSE,
+ fail = TRUE
+)
+
+varslurped <- readr::read_tsv(
+ varfiles,
+ skip = 1,
+ col_names = c(
+ "chrom",
+ "pos",
+ "ref",
+ "alt",
+ "filter",
+ "type",
+ # Allelic depths for the ref and alt alleles in the order listed
+ "ad",
+ # Allele fractions of alternate alleles. Excludes filtering
+ "freq_alt_raw",
+ # Approximate read depth (reads are filtered if MQ=255 or with bad mates)
+ "depth_total",
+ # Count of fragments supporting each allele.
+ "frag_allele_depth",
+ # Genotype Quality
+ "genotype_qual"
+ ),
+ col_types = c("cdccccccccc"),
+ id = "sample"
+)
+
+varslurpedfmt <- varslurped %>%
+ mutate(sample = str_extract(sample,
+ regex("(HAMBI_[:digit:]{4}-[:alpha:]-[:alnum:]{3})|(HAMBI_[:digit:]{4}-[:alpha:]-SH-WGS-[:digit:]{3})"))) %>%
+ separate(sample, into = c("strainID", "replicate", "sample"), sep="-", extra = "merge") %>%
+ # taken only first two most abundant alleles
+ separate(ad, into = c("depth_ref", "depth_alt"),
+ sep=",", extra="drop") %>%
+ # format depths to numberic
+ mutate(depth_ref = as.numeric(depth_ref),
+ depth_alt = as.numeric(depth_alt)) %>%
+ select(-frag_allele_depth, -genotype_qual) %>%
+ mutate(freq_alt = depth_alt/(depth_ref + depth_alt),
+ freq_ref = 1 - freq_alt) %>%
+ # final format depths to numeric
+ mutate(freq_alt_raw = as.numeric(freq_alt_raw),
+ depth_total = as.numeric(depth_total))
+
+# Read premade annotations -----------------------------------------------------
+
+annotations <- read_rds(here::here("_data_raw", "shared", "annotations_codon_degeneracy.rds"))
+
+# Combine -----------------------------------------------------------------
+
+full <- left_join(varslurpedfmt, snpeffslurpedfmt,
+ by = join_by(strainID, replicate, sample, chrom, pos, ref, alt)) %>%
+ left_join(annotations,
+ by = join_by(strainID, chrom, locus_tag))
+# Write output ------------------------------------------------------------
+
+write_rds(full, here::here(data, "mutect_parsed_annotated.rds"))
+
+# Clean up ----------------------------------------------------------------
+
+# remove decompressed vcf and tables directory from temp location
+fs::dir_delete(vcfdir)
+```
diff --git a/R/wgs/02_parallelism.qmd b/R/wgs/02_parallelism.qmd
new file mode 100644
index 0000000..5d15c65
--- /dev/null
+++ b/R/wgs/02_parallelism.qmd
@@ -0,0 +1,824 @@
+---
+title: "WGS genetic parallelism analysis"
+author: "Shane Hogle"
+date: today
+link-citations: true
+bibliography: references.bib
+abstract: "This notebook reads in the formatted WGS variant frequencies and annotations. It then uses statistical tests to determine which mutations and which functional categories appeared across biological replicates within condition more often than would be expected by chance. It also looks for mutations/functions that arise within treatment categories more than expected by chance."
+---
+
+# Setup
+
+Libraries and global variables
+
+```{r}
+#| output: false
+library(tidyverse)
+library(here)
+library(fs)
+library(patchwork)
+library(ggforce)
+library(latex2exp)
+library(vctrs)
+
+source(here::here("R", "utils_generic.R"))
+source(here::here("R", "wgs", "utils_parallelism.R"))
+```
+
+Set up some directories
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+data_raw <- here::here("_data_raw", "wgs")
+shared <- here::here("_data_raw", "shared")
+data <- here::here("data", "wgs")
+figs <- here::here("figs", "wgs")
+
+# make processed data directory if it doesn't exist
+fs::dir_create(data)
+fs::dir_create(figs)
+```
+
+# Read data
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+mgvars <- read_rds(here::here(data, "mutect_parsed_annotated.rds"))
+degentab <- read_rds(here::here(shared, "annotations_codon_degeneracy.rds"))
+genome_len <- read_tsv(here::here(shared, "HAMBI_genome_len.tsv"))
+```
+
+[The mutect2 documentation](https://gatk.broadinstitute.org/hc/en-us/articles/360050722212-FAQ-for-Mutect2) says that any variant without "PASS" is false positive and should be removed. We will follow this convention here and only focus on PASSing mutations.
+
+```{r}
+#| output: false
+# note that 1287 got an extra replicate of replicate "A" sequenced. It was called replicate "B" in the
+# files and I will combine it with the other replicate "A"
+tofilter <- mgvars %>%
+ filter(strainID == "HAMBI_1287" & replicate %in% c("A","B"))
+
+# this is to make sure we filter out duplicates by randomly taking one
+# from each group BUT to make sure to select duplicates that have a PASS
+tocombine <- tofilter %>%
+ group_by(chrom, pos, ref, alt) %>%
+ mutate(W = n()) %>%
+ mutate(W1 = case_when(W > 1 & filter == "PASS" ~ 1,
+ W > 1 & filter != "PASS" ~ 0,
+ W == 1 ~ 1)) %>%
+ mutate(W2 = if_else(W > 1 & sum(W1) == 0, 1, 0)) %>%
+ mutate(W3 = W2 + W1) %>%
+ relocate(W, W1, W2, W3) %>%
+ filter(W3 == 1) %>%
+ slice(1) %>%
+ dplyr::select(-W, -W1, -W2, -W3) %>%
+ mutate(replicate = "A")
+
+mgvars_filt <- anti_join(mgvars, tofilter) %>%
+ bind_rows(tocombine) %>%
+ mutate(replicate = if_else(replicate == "B", "C", replicate)) %>%
+ # we are filtering by requiring a depth of at least 10 reads
+ # and that the variant passed the mutect2 internal filters
+ filter(filter == "PASS") %>%
+ group_by(chrom, pos, ref, alt, replicate) %>%
+ filter(freq_alt == max(freq_alt)) %>%
+ ungroup()
+```
+
+# Parallelism at nucleotide level
+
+Looking at all mutations (inside and outside of coding sequences, synonymous and nonsynonymous)
+
+```{r}
+mgvars_filt %>%
+ group_by(strainID, chrom, pos, ref, alt) %>%
+ mutate(mi=n_distinct(replicate)) %>%
+ group_by(strainID) %>%
+ count(mi) %>%
+ group_by(mi) %>%
+ mutate(tot = sum(n)) %>%
+ arrange(desc(mi))
+```
+
+We identified 22 species with detectable mutations (basically everything, except for 1299 for which the sequencing failed). Interestingly, (as shown above) there are many species with a huge amount of parallel mutations across all three replicates. For example there are 1497 sites with the exact same mutation in all three sequenced replicates. Most of these are not fixed mutations (but some of them are), but what I realized when digging deeper is that most of these extremely parallel mutations reside on predicted plasmid sequences, in mobile genetic elements (like transposases), and in predicted prophages. Certainly something interesting is going on in these regions of these species, but I don't think that short read sequencing is really sufficient to tell us what is going on. Instead short reads seem to give lots of short indels in very small windows of the genome producing a lot of noise. So I think it is best to try and remove these kinds of mutations before proceeding with the analysis.
+
+We used the tools MGEfinder v1.0.3 with database v1.0.2 [@durrant2020a] and geNomad v1.7.4 [@camargo2023] to identify mobile elements, plasmids and integrated viruses (prophages) in the genomes and excluded these regions from subsequent analyses.
+
+```{r}
+#| output: false
+#| warning: false
+#| error: false
+mgefinder <- read_tsv(here::here(shared, "MGE_finder_HAMBI_combined.tsv"))
+genomad <- read_tsv(here::here(shared, "genomad_HAMBI_combined.tsv")) %>%
+ # exclude very large plasmids over 1Mbp from the filtering
+ filter((end - start) < 1000000)
+```
+
+```{r}
+mgvars_filt_mb <- mgvars_filt %>%
+ left_join(mgefinder, by = join_by(strainID, chrom), relationship = "many-to-many") %>%
+ filter(!(pos >= start.y & pos <= end.y)) %>%
+ dplyr::select(-name, -start.y, -end.y, -prediction_tool) %>%
+ distinct() %>%
+ left_join(genomad, by = join_by(strainID, chrom), relationship = "many-to-many") %>%
+ filter(!(pos >= start & pos <= end)) %>%
+ dplyr::select(-name, -start, -end, -prediction_tool) %>%
+ distinct() %>%
+ rename(start = start.x, end = end.x)
+```
+
+```{r}
+mgvars_filt_mb %>%
+ # remove the annotations for saving
+ select(-c(feature_type:last_col())) %>%
+ write_tsv(here::here(data, "wgsvars_filt_mb_nonsyn.tsv"))
+```
+
+
+```{r}
+mgvars_filt_mb %>%
+ group_by(strainID, chrom, pos, ref, alt) %>%
+ mutate(mi=n_distinct(replicate)) %>%
+ group_by(strainID) %>%
+ count(mi) %>%
+ group_by(mi) %>%
+ mutate(tot = sum(n)) %>%
+ arrange(desc(mi))
+```
+
+OK this seems a lot more reasonable here... Now there are much fewer mutations that are parallel across the replicates.
+
+To put these observations in context, we can compare them to a simple null model in which mutations are uniformly distributed across the sites in the genome. We can then check how many sites with 1, 2, and 3-fold multiplicity that we would expect by chance. We define nucleotide parallelism as the number of mutations occurring at the same site in the genome in independent populations. For each site, we define the multiplicity, $m_{i}$, as the number of populations (i.e., replicates) with a mutation detected at that site, so that the multiplicity in this experiment can range from one to three.
+
+The expected number of mutations with $m_{i} \geq m$ in a sample of total mutations size $n_{tot}$ is:
+
+$$ S(m) \approx \sum_{n \geq m} \frac{n}{n_{tot}} \cdot L_{tot} \cdot \frac{ \left( \frac{n_{tot}}{L_{tot}} \right) ^n}{n!} e^{-n_{tot}/L_{tot}} $$
+
+where $L_{tot}$ is the total number of bases in the genome.
+
+## Nucleotide parallelism by treatment
+
+Here we will do the estimation separately for each species $\times$ treatment combination.
+
+```{r}
+nuc_survival <- mgvars_filt_mb %>%
+ dplyr::select(strainID, pos, ref, alt, replicate) %>%
+ # multiplicity = number of replicate populations each unique mutation is
+ # observed in group by the mutation position, with alternative allele to the
+ # grouping
+ summarize(m = n_distinct(replicate), .by = c(strainID, pos, ref, alt)) %>%
+ # now calculate the total number of mutations across all replicates so we need
+ # to ungroup by mutation position/alt allele but because we still want to
+ # determine this value by treatment category we keep the treatment category
+ # grouping. However, this should be changed if you want to for example average
+ # over all the treatment conditions on a species basis
+ group_by(strainID) %>%
+ count(m, name = "m_count") %>%
+ mutate(n = m * m_count,
+ Ntot = sum(n),
+ perc = n / Ntot * 100) %>%
+ left_join(genome_len, by = join_by(strainID)) %>%
+ arrange(cur_group_id(), desc(m)) %>%
+ # dpois() tells the probability mass at a given number of counts. Here we
+ # want to get the probability of observing n mutations with multiplicity
+ # = mi (i.e. the counts of mi in the observed data). We assume that
+ # mutations independently occur on the genome of size Ltot at a rate of
+ # lambda = Ntot/Ltot and that generally the events are rare. Thus this
+ # situation can be modeled by the Poisson distribution. We can get the
+ # binned number of mutations per level of multiplicity m by multiplying
+ # the probability by the length of the genome and the binned mutations
+ # divided by the total number of mutations.
+ mutate(m_count_expected = cumsum((m_count / Ntot) *
+ total_len *
+ dpois(m, lambda = Ntot / total_len))) %>%
+ dplyr::select(-num_contigs) %>%
+ relocate(m, n, Ntot, perc, m_count, m_count_expected) %>%
+ ungroup()
+
+# setup for plotting
+nuc_survival_plot <- nuc_survival %>%
+ group_by(strainID) %>%
+ # when there is only one multiplicity observed for a mutation filter such
+ # that the multiplicty of that mutation must be greater than one.
+ # Otherwise include all remaining mutations (m > 0)
+ filter(case_when(n() == 1 ~ m > 1,
+ TRUE ~ m > 0)) %>%
+ pivot_longer(cols = c("m_count", "m_count_expected")) %>%
+ mutate(label = paste(strainID)) %>%
+ # and make final plot
+ plot_nuc_survival(., 5000, c(1, 10, 100, 1000, 10000), 4)
+```
+
+::: {#fig-01}
+```{r}
+#| fig.width: 8
+#| fig.height: 6
+#| echo: false
+#| warning: false
+nuc_survival_plot
+```
+
+Distribution of nucleotide multiplicity $(m_{i})$ - i.e., the number of replicates with with the same genomic mutation including site and alternative allele - for each species with multiplicities $\gt 1$. The observed data is shown in blue while the null expectation is shown in red. In all cases the observed $m_{i}$ exceeded the null expectation meaning that we must reject the simple null model of uniform mutation distribution in favor of the alternative model where mutations are nonrandom and cluster across replicate populations.
+:::
+
+## Conclusion
+
+Nucleotide parallelism results are presented in Figure @fig-01. For these species this very simple null model mostly predicts that we should expect fewer than two parallel mutations (same position, same alternative allele) across the three replicate populations. For 12 evolved species the observed data show an excess of nucleotide parallelism relative to this simple null expectation. In particular HAMBI_0006, HAMBI_1287, and HAMBI_2494 all have at least 1 mutation that is identical across all three replicate evolved populations. However, multi-hit sites are only a very small fraction of the total observed mutations across all species (\~ 3% across species with multi-hit sites). Thus, we will continue our analysis but looking at the gene level.
+
+```{r}
+#| warning: false
+nuc_survival %>%
+ mutate(Ntot_f = sum(n)) %>%
+ summarize(perc = sum(n)/Ntot_f*100, .by = "m") %>%
+ distinct() %>%
+ mutate(cum_perc = cumsum(perc))
+```
+
+# Parallelism at gene level
+
+If selection pressures and mutation rates did not vary between genes, the number of mutations in each gene should be proportional to the target gene size. We assume that gene length is a primary driver of the target size. We then define a multiplicity for each gene according to:
+
+$$m_{i} = n_{i} \cdot \frac{\overline{L}}{L_{i}}$$ {#eq-01}
+
+where $n_{i}$ is the number of nonsynonymous mutations (i.e., amino acid changing) in $gene_{i}$ across all replicate populations per species, $L_{i}$ is the total number of nonsynonymous sites (i.e., excluding 4-fold degenerate sites) in $gene_{i}$, and $\overline{L}$ is the average value of $L_{i}$ across all genes in the genome. This definition ensures that under the null hypothesis, all genes have the same expected multiplicity $\overline{m} = n_{tot}/n_{genes}$.
+
+We next define a null model for gene multiplicity that assumes mutations are assigned to genes with probability:
+
+$$p_{i} \propto L_{i} r_{i}$$ {#eq-02}
+
+for some set of enrichment factors $r_{i}$. In the alternate model, the maximum likelihood estimator of the enrichment factors $r_{i}$ is the ratio of observed to expected gene multiplicities, $r_{i} = m_{i}/\overline{m}$. The net increase of the log-likelihood relative to the null model ($r_{i} = 1$) is then given by:
+
+$$\Delta \ell = \sum_{i} n_{i} \log \left ( \frac{m_{i}}{\overline{m}} \right )$$ {#eq-03}
+
+This likelihood ratio estimator is equivalent to the total G-score introduced in earlier work [@tenaillon2016]. Here we assess statistical significance using permutation tests following Shoemaker *et al.* [@shoemaker2021]. For permutations tests (n = 10,000) we randomly resample mutations to the observed $n_{tot}$ set size from each species using the multinomial distribution, where the probability of sampling a non-synonymous mutation at $gene_{i}$ was given by $p_{i} = {L_{i}} / {L_{tot}}$ where $L_{i}$ is the total number of nonsynonymous sites in $gene_{i}$ and $L_{tot}$ is the total number of nonsynonymous sites in the genome.
+
+We next identify specific genes enriched for mutations following the criteria of Good *et. al* . [@good2017]. We calculate a \textit{p} value for each gene using the equation:
+
+$$P_{i} = \sum_{n \geq n_{i}} \frac{ \left ( \frac{n_{tot} L_{i}}{ \overline{L} N_{genes}} \right )^{n}}{n!} e^{- \frac{n_{tot} L_{i} }{\overline{L} N_{genes}}}$$ {#eq-04}
+
+Here we only consider genes with $n_{i} \geq 3$ to exclude low \textit{p} values driven primarily by gene length. Under the null hypothesis, the expected number of genes with $P_{i} \geq p$ can be found using a Poisson survival curve given by:
+
+$$\overline{N}(P) \approx \sum_{i=1}^{N_{genes}} \sum_{n=3}^{\infty} \theta (P - P_{i}(n, L_{i}) \cdot \frac{ \left( \frac{n_{tot} L_{i}}{\overline{L} N_{genes}} \right)^{n}}{n!} e^{-\left( \frac{n_{tot} L_{i}}{\overline{L} N_{genes}} \right)}$$ {#eq-05}
+
+We can compare this expected number to the observed number of genes $N(P)$ using a critical \textit{p} value ($P^{\ast}$) such that
+
+$$\frac{\overline{N}(P^{\ast})}{N(P^{\ast})} \leq \alpha$$ {#eq-06}
+
+for a given FDR $\alpha$ value of 0.05. For this value we then define the set of significantly enriched genes as:
+
+$$I = \left\{ i : P_{i} \leq P^{\ast} \left(\alpha \right) \right\}$$ {#eq-07}
+
+## Prepare input data for non-syn mutations only
+
+```{r}
+mgvars_filt_ns <- mgvars_filt_mb %>%
+ # exclude intragenic, intergenic, and synonymous mutations. Also exclude
+ # fusion functions because these are weird and also rare. Excluding the
+ # modifier category also ensures that we filter out any tRNAs with mutations
+ filter(!str_detect(effect, "intergenic|intragenic|synonymous|fusion")) %>%
+ filter(!str_detect(impact, "MODIFIER")) %>%
+ dplyr::select(strainID, replicate, chrom:alt, locus_tag) %>%
+ relocate(strainID, locus_tag)
+```
+
+## G score
+
+Warning! this takes some time
+
+```{r}
+#| eval: false
+#| echo: true
+muts_by_sp <- mgvars_filt_ns %>%
+ group_by(strainID) %>%
+ distinct() %>%
+ mutate(groupid = cur_group_id()) %>%
+ group_by(groupid, strainID)
+
+# doing 10000 permutations
+result_sp <- run_full_gene_parallelism_pipeline(muts_by_sp, degentab, 10000)
+
+# Save this for later
+write_rds(result_sp, here::here(data, "parallelism_results_sp_nomobile.rds"))
+```
+
+```{r}
+# Read in previous results
+result_sp <- read_rds(here::here(data, "parallelism_results_sp_nomobile.rds"))
+```
+
+### Simulating the number of replicates with a parallel gene hit
+
+As another sanity check, we took the resamplings from the null distribution and calculated the number of replicates with a parallel gene hit (i.e. the same gene hit in $\geq 2$ experimental replicates) and compared that to the observed distribution. The null distribution reflected the process of sampling a mutation under a multinomial distribution with probability proportional to the total number of nonsynonymous sites $L_{i}$ in $gene_{i}$. We observed that in 9/16 cases the null distribution produced fewer parallel gene hits than observed in the data from 2 experimental replicates. In all 16 cases the null distribution produced fewer parallel gene hits than observed in the data from 3 experimental replicates. Thus, for some suspicious genes with very high mutational density (see below) we required these genes to be present in all three replicates to include them in our final significant gene lists.
+
+::: {#fig-03a}
+```{r}
+#| fig.width: 8
+#| fig.height: 7
+#| echo: false
+#| warning: false
+plot_sims(result_sp, 1)
+```
+
+Number of genes (y axis) with nonsynonymous mutation hits in $N_{replicates} = \{n : 1 \leq n \leq 3\}$ (x axis). Blue bars depict the distribution observed in the data, while red bars depict the null distribution from resampling mutations to the observed $n_{tot}$ set size from each species using the multinomial distribution, where the probability of sampling a non-synonymous mutation at $gene_{i}$ was dependent only on the number of nonsynonymous sites in the gene.
+:::
+
+::: {#fig-03b}
+```{r}
+#| fig.width: 8
+#| fig.height: 7
+#| echo: false
+#| warning: false
+plot_sims(result_sp, 2)
+```
+
+Same as in @fig-03a but for different treatment combinations.
+:::
+
+### Visualize the survival curves for each treatment combination
+
+```{r}
+survcurv <- result_sp$output_df2plot %>%
+ filter(!is.na(xend)) %>%
+ mutate(label = strainID) %>%
+ ggplot() +
+ geom_segment(
+ aes(
+ x = x,
+ y = y,
+ xend = xend,
+ yend = yend
+ ),
+ linetype = "dashed",
+ linewidth = 0.25, color = "red"
+ ) +
+ geom_step( aes(x = obs_p, y = value, color = name)) +
+ facet_wrap(~ label, ncol = 4) +
+ labs(x = TeX("$-log_{10}P$"), y = "Number of genes") +
+ scale_color_manual(values = c("blue", "grey50")) +
+ scale_y_log10(
+ breaks = c(1, 10, 100),
+ labels = scales::trans_format("log10", scales::math_format(10 ^.x))
+ ) +
+ annotation_logticks(sides = "l", color = "grey40") +
+ coord_cartesian(ylim = c(1, 200)) +
+ theme_bw() +
+ theme(
+ strip.background = element_blank(),
+ panel.grid = element_blank(),
+ legend.position = "bottom"
+ )
+```
+
+::: {#fig-04}
+```{r}
+#| fig.width: 8
+#| fig.height: 8
+#| echo: false
+#| warning: false
+survcurv
+```
+
+The observed $(N)$ (blue line) and expected number $(\overline{N})$ (grey line) of genes with $P_{i} \leq P$, as a function of $P$. The red dashed line denotes the genome-wide significance threshold $P^{*}$ for $\alpha = 0.05$ defined in Equation @eq-06.
+:::
+
+### Inspecting individual genes
+
+```{r}
+result_sp$output_gscore %>%
+ filter(pvalue <= 0.05)
+```
+
+Also some genes tend to have a large number of consecutive mutations which is probably due to incorrect read mapping and not real mutations so we will try and filter these out. We'll do this by filtering on genes with a mutational density $d_{g} \gt 0.15$, where the mutational density in gene $g$ was defined as
+
+$$d_{g} = \frac{|A|_{g}}{b_{g} - a_{g}}$$
+
+for all genome positions $x$ on the interval $\{x | a_{g} \leq x \leq b_{g}\}$ where $b_{g}$ is the greatest position and $a_{g}$ is the smallest position of observed mutations in gene $g$ and $|A|_{g}$ is the cardinality of the mutation set.
+
+```{r}
+high_density_genes <- mgvars_filt_ns %>%
+ group_by(strainID, locus_tag, replicate) %>%
+ mutate(mutwindox = pos - lag(pos),
+ mut_num = n(),
+ mut_dens = n()/(max(pos) - min(pos))) %>%
+ ungroup() %>%
+ filter(mut_dens != Inf) %>%
+ filter(mut_dens > 0.15) %>%
+ distinct(locus_tag) %>%
+ pull()
+```
+
+```{r}
+output_gene_table_filt <- result_sp$output_gene_table %>%
+ # only genes exceeding the critical pstar value from the survival curves above
+ filter(neg_log10P >= pstar) %>%
+ # only include genes with at least 3 hits in the same gene or found in more than one replicate.
+ filter(observed_hits_n_i >= 3 | n_replicate > 1) %>%
+ # filters out the problematic genes above unless the number of replicates detected is 2 or more
+ filter(case_when(n_replicate < 2 ~ !(locus_tag %in% high_density_genes),
+ TRUE ~ n_replicate > 1)) %>%
+ left_join(distinct(select(mgvars_filt, locus_tag, product, cds_length:gene)),
+ by = join_by(locus_tag))
+
+output_gene_table_filt
+```
+
+# Convergent/divergent evolution
+
+So now we have a list of enriched genes for each species. We can see that, for example, some gene functions pop up multiple times (like transcription factors and sensory kinases). Are these different functions popping up across species more that we would expect by chance? We'll try and test this using a simulation based approach
+
+```{r}
+#| eval: false
+#| echo: true
+# Read in results of eggnogmapper which maps a COG category to every gene it can in each genome
+eggnograw <- read_tsv(here::here(shared, "HAMBI_all.emapper.annotations.xz"))
+```
+
+```{r}
+cog_description <- tibble::tribble(
+ ~COG_category_single, ~COG_category_long,
+ "J", "J - Translation, ribosomal structure and biogenesis",
+ "A", "A - RNA processing and modification",
+ "K", "K – Transcription",
+ "L", "L - Replication, recombination and repair",
+ "B", "B - Chromatin structure and dynamics",
+ "D", "D - Cell cycle control, cell division, chromosome partitioning",
+ "Y", "Y - Nuclear structure",
+ "V", "V - Defense mechanisms",
+ "T", "T - Signal transduction mechanisms",
+ "M", "M - Cell wall/membrane/envelope biogenesis",
+ "N", "N - Cell motility",
+ "Z", "Z – Cytoskeleton",
+ "W", "W - Extracellular structures",
+ "U", "U - Intracellular trafficking, secretion, and vesicular transport",
+ "O", "O - Posttranslational modification, protein turnover, chaperones",
+ "X", "X - Mobilome: prophages, transposons",
+ "C", "C - Energy production and conversion",
+ "G", "G - Carbohydrate transport and metabolism",
+ "E", "E - Amino acid transport and metabolism",
+ "F", "F - Nucleotide transport and metabolism",
+ "H", "H - Coenzyme transport and metabolism",
+ "I", "I - Lipid transport and metabolism",
+ "P", "P - Inorganic ion transport and metabolism",
+ "Q", "Q - Secondary metabolites biosynthesis, transport and catabolism",
+ "R", "R - General function prediction only",
+ "S", "S - Function unknown"
+ )
+```
+
+```{r}
+#| eval: false
+#| echo: true
+# formatting the eggnog dataframe for downstream use
+enrichedspprefix <- unique(paste0("H", str_extract(output_gene_table_filt$locus_tag, "\\d+")))
+enrichedsp <- unique(output_gene_table_filt$strainID)
+names(enrichedsp) <- enrichedspprefix
+
+# use a seed for reproducibility because we are randomly sampling the split COGs
+withr::with_seed(6783543, eggnog <- eggnograw %>%
+ filter(str_detect(query, paste(enrichedspprefix, collapse="|"))) %>%
+ mutate(COG_category = if_else(COG_category == "-", NA_character_, COG_category)) %>%
+ # If multiple cog categories, split them into a character vector then randomly sample
+ # one of them
+ mutate(COG_category1 = strsplit(COG_category, split = "")) %>%
+ rowwise() %>%
+ mutate(COG_category_single = sample(COG_category1, 1)) %>%
+ ungroup() %>%
+ mutate(strmatchvar = str_extract(query, "^[:alnum:]*")) %>%
+ mutate(strainID = str_replace_all(strmatchvar, enrichedsp)) %>%
+ dplyr::select(-strmatchvar, -COG_category1) %>%
+ rename(locus_tag = query) %>%
+ left_join(cog_description, by = join_by(COG_category_single)) %>%
+ relocate(strainID, locus_tag, COG_category_single, COG_category_long))
+
+write_tsv(eggnog, here::here(shared, "HAMBI_all_eggnog_formatted.tsv.xz"))
+```
+
+```{r}
+#| output: false
+# read back the formatted eggnog data
+eggnog <- read_tsv(here::here(shared, "HAMBI_all_eggnog_formatted.tsv.xz"))
+```
+
+save a copy of significantly parallel genes as a table
+
+```{r}
+left_join(output_gene_table_filt, eggnog,
+ by = join_by(locus_tag, strainID)) %>%
+ rename(prokka_annotation = product) %>%
+ write_tsv(here::here(data, "enriched_parallel_genes.tsv"))
+```
+
+
+## All species
+
+Calculate the probability of each gene (ns_sites/total_ns_sites) and link to COG categories
+
+```{r}
+eggnog_nslen <- left_join(eggnog, degentab,
+ by = join_by(strainID, locus_tag)) %>%
+ dplyr::select(strainID, locus_tag, COG_category_single, ns_length) %>%
+ filter(!is.na(COG_category_single)) %>%
+ group_by(strainID) %>%
+ mutate(p = ns_length/sum(ns_length),
+ cog_genes = n()) %>%
+ ungroup()
+```
+
+### Jensen Shannon Divergence
+
+This function draws n genes from each species with a probability proportional to gene length, where n = the number of significant parallel genes observed from the data. We repeat this simulation 10 000 times and below we check if 1) the Jensen Shannon divergence between the distribution of the COG categories drawn randomly and the background (i.e., the distribution of COG categories in the whole genome) is greater than the Jensen Shannon divergence between the background and the observed distribution of COG categories in significantly parallel genes. The idea is that if some COG categories are enriched/depleted in the parallel genes, then the JSD divergence between the observed and the background should be larger than the JSD between the random sampled genes (sampling probability based only on gene length) and the background. We calculate an empirical P value by "counting" how many times the simulated JSD exceeds the observed.
+
+```{r}
+#| eval: false
+#| echo: true
+gene_draw <- function(){
+ output_gene_table_filt %>%
+ group_by(strainID) %>%
+ summarize(nsiggenes = n()) %>%
+ left_join(eggnog_nslen,
+ by = join_by(strainID)) %>%
+ nest(data = -c(strainID)) %>%
+ mutate(locus_tag = map(
+ data,
+ \(x) sample(
+ x = x$locus_tag,
+ size = unique(x$nsiggenes),
+ prob = x$p,
+ replace = FALSE
+ )
+ )) %>%
+ dplyr::select(-data) %>%
+ unnest(cols = locus_tag)
+}
+
+# take 10 000 random draws of the genes. Use a seed for reproducibility
+withr::with_seed(12367,
+ genes_simulated <- map(1:10000, ~gene_draw(), .progress=TRUE) %>%
+ list_rbind(names_to = "id"))
+
+# save the output so we don't need to run again
+write_rds(genes_simulated, here::here(data, "genes_simulated_nomobile.rds"))
+```
+
+Some data formatting
+
+```{r}
+# read data back
+genes_simulated <- read_rds(here::here(data, "genes_simulated_nomobile.rds"))
+
+# map the locus_tags to COG categories for the simulated
+# gene draws
+COGs_distribution_simulated <- genes_simulated %>%
+ left_join(eggnog_nslen, by = join_by(strainID, locus_tag)) %>%
+ group_by(id) %>%
+ count(COG_category_single, name = "nsim") %>%
+ ungroup()
+
+# get the discrete distribution of COGs in the significant parallel genes
+COGs_distribution_observed <- left_join(output_gene_table_filt, eggnog,
+ by = join_by(locus_tag, strainID)) %>%
+ count(COG_category_single, name = "nobs")
+
+# get the discrete distribution of COGs across the whole genome of all species
+# this is the null case (i.e. the background that we would expect if there was no
+# enrichment)
+COGs_distribution_nullexpectation <- eggnog %>%
+ count(COG_category_single, name = "nnull")
+```
+
+Now we calculate the Jensen Shannon divergence between the background and the simulated data to estimate an empirical P value
+
+```{r}
+# function to calculate the Jensen Shannon divergene of a pair of
+# distributions
+JSD_pair <- function(p, q){
+ m <- 0.5 * (p + q)
+ JS <- 0.5 * (sum(p * log(p/m)) + sum(q * log(q/m)))
+ return(JS)
+}
+
+# Calculate the Jensen Shannon divergence between the COG distribution observed in
+# enriched genes and the COG distribution in the genomic background
+JSD_observed <- left_join(COGs_distribution_observed, COGs_distribution_nullexpectation,
+ by = join_by(COG_category_single)) %>%
+ mutate(p = nobs/sum(nobs),
+ q = nnull/sum(nnull)) %>%
+ summarize(JSD = JSD_pair(p, q)) %>%
+ pull()
+
+# now calculate JSD between the simulated draws and the backgound distribution
+JSD_simulated <- left_join(COGs_distribution_simulated,
+ COGs_distribution_nullexpectation,
+ by = join_by(COG_category_single)) %>%
+ # drops from calculation categories that aren't in both the observed data
+ # and the resampled data. Need to do this because JSD can't handle zero values
+ drop_na() %>%
+ group_by(id) %>%
+ mutate(p = nsim/sum(nsim),
+ q = nnull/sum(nnull)) %>%
+ summarize(JSD = JSD_pair(p, q)) %>%
+ pull(JSD)
+
+# proportion of JSD measures between simulated and the background that are greater
+# than the JSD between the observed and the background
+global_p <- length(JSD_simulated[JSD_simulated >= JSD_observed])/length(JSD_simulated)
+
+global_p
+```
+
+### Enrichment of individual COG categories
+
+OK so the overall distribution of COG categories in the parallel genes is not significantly different than in the rest of the genome - ~40% of the time the JSD between the COG distribution in a random sample and the background is greater than the observed COG distribution and the background.
+
+#### Hypergeometric Test
+
+We can test enrichment in individual COG categories in individual species using the hypergeometric test
+```{r}
+library(qvalue)
+
+par2test <- left_join(output_gene_table_filt, eggnog,
+ by = join_by(locus_tag, strainID)) %>%
+ filter(!is.na(COG_category)) %>%
+ group_by(strainID) %>%
+ count(COG_category_single, name = "n_cog_par") %>%
+ mutate(n_par = sum(n_cog_par))
+
+back2test <- eggnog %>%
+ filter(!is.na(COG_category)) %>%
+ group_by(strainID) %>%
+ count(COG_category_single, name = "n_cog_background") %>%
+ mutate(n_background = sum(n_cog_background))
+
+phypres <- left_join(par2test, back2test, by = join_by(strainID, COG_category_single)) %>%
+ mutate(p_enrich = enricher(n_background, n_par, n_cog_background, n_cog_par, over=TRUE),
+ p_deplete = enricher(n_background, n_par, n_cog_background, n_cog_par, over=FALSE)) %>%
+ mutate(p_enrich_adj = p.adjust(p_enrich, method = "fdr"),
+ p_deplete_adj = p.adjust(p_deplete, method = "fdr")) %>%
+ arrange(p_enrich_adj)
+
+qenrich <- qvalue_truncp(phypres$p_enrich)
+qdeplete <- qvalue_truncp(phypres$p_deplete)
+
+phypres$q_enrich <- qenrich$qvalues
+phypres$q_deplete <- qdeplete$qvalues
+
+phypres %>%
+ relocate(p_enrich, q_enrich, p_enrich_adj) %>%
+ arrange(p_enrich)
+```
+
+Here COG categories "D - Cell cycle control, cell division, chromosome partitioning", "M - Cell wall/membrane/envelope biogenesis", and "P - Inorganic ion transport and metabolism" are enriched with $p_{adj} \leq 0.05$.
+
+
+#### Nonparametric simulation
+
+We can also look at different COG categories by simply counting the number of random samplings that have a COG count greater/less than that the observed COG count. Since we resampled the same number of genes as in the observed we can compare the two directly. Below we do this integrated over all species in the community to see if there was any pathway enriched across all species.
+
+```{r}
+left_join(COGs_distribution_simulated, COGs_distribution_observed,
+ by = join_by(COG_category_single)) %>%
+ drop_na() %>%
+ mutate(gt = if_else(nsim >= nobs, 1, 0)) %>%
+ group_by(COG_category_single) %>%
+ summarize(p = sum(gt)/n()) %>%
+ arrange(p) %>%
+ mutate(p_adj = p.adjust(p, method = "fdr"))
+```
+
+Only about 1% of the re-samples had a higher number of K COG genes compared to the observed number of K COG genes. No other COG category meets the standard alpha = 0.05 threshold. However, after correcting for multiple testing the 5% alpha threshold is no longer met.
+
+This is just a sanity check that the randomly sampled COGs reflect the background distribution - i.e., a large fraction of the resamples are larger (and smaller) than the null. There is no systematic difference, and we observe the different COG categories in the resampels at basically the same rate as we observe the COG categories in the background.
+
+```{r}
+left_join(COGs_distribution_simulated, COGs_distribution_nullexpectation,
+ by = join_by(COG_category_single)) %>%
+ group_by(id) %>%
+ mutate(fsim = nsim/sum(nsim), fnull = nnull/sum(nnull)) %>%
+ drop_na() %>%
+ mutate(gt = if_else(fsim >= fnull, 1, 0)) %>%
+ group_by(COG_category_single) %>%
+ summarize(p = sum(gt)/n()) %>%
+ arrange(p)
+```
+
+### COG Distribution Plot
+
+```{r}
+cog_pmut <- left_join(output_gene_table_filt, eggnog,
+ by = join_by(locus_tag, strainID)) %>%
+ count(COG_category_single) %>%
+ mutate(f = n/sum(n),
+ n = n*1000,
+ type = "parallel_mutations")
+
+cog_everything <- eggnog %>%
+ count(COG_category_single) %>%
+ mutate(f = n/sum(n),
+ type = "all_genes")
+
+pcogs <- bind_rows(cog_pmut, cog_everything) %>%
+ drop_na() %>%
+ ggplot() +
+ geom_bar(aes(x = fct_reorder(COG_category_single, f, .desc = F), y=n, fill = type), stat="identity",
+ position = position_dodge( preserve = "total")) +
+ scale_fill_brewer(palette = "Set1") +
+ labs(x = "COG Category", y = "Fraction genes in COG category", fill = "Subset") +
+ scale_y_continuous(
+ "N all genes",
+ sec.axis = sec_axis(~ . * 1/1000, name = "N parallel genes")
+ ) +
+ theme_bw() +
+ theme(
+ panel.grid = element_blank(),
+ panel.background = element_blank(),
+ legend.position = "bottom"
+ )
+```
+
+::: {#fig-05}
+```{r}
+#| fig.width: 8
+#| fig.height: 6
+#| echo: false
+pcogs
+```
+
+Number of genes in each Cluster of Orthologous Groups (COG) category for genes with a COG annotation across all 23 HAMBI population genomes (red bars, left y-axis) and across only genes exhibiting significant genomic parallelism (blue bars, right y-axis).
+:::
+
+## Species present in the metagenomes
+
+Now that I've done the analysis on the metagenomes we know that we can only detect HAMBI_2659, HAMBI_1972, HAMBI_1977, and HAMBI_1287 with high enough coverate to detect sequence variants. From the hypergeometric test above we know that COG categories D, M, and P are enriched in HAMBI_1972, HAMBI_1977, amd HAMBI_2659 respsectively. Here we will make a plot focusing only on those species
+
+```{r}
+# species detected in the metagenomes
+mgsp <- c("HAMBI_2659", "HAMBI_1972", "HAMBI_1977", "HAMBI_1287", "HAMBI_2443")
+
+cog_pmut_mgsp <- left_join(output_gene_table_filt, eggnog,
+ by = join_by(locus_tag, strainID)) %>%
+ filter(strainID %in% mgsp) %>%
+ group_by(strainID) %>%
+ count(COG_category_single, name = "tot") %>%
+ # scaling factor to make ploting on 2 axes possible
+ mutate(n = tot*250,
+ f = n/sum(n),
+ type = "parallel_mutations")
+
+cog_everything_mgsp <- eggnog %>%
+ filter(strainID %in% mgsp) %>%
+ group_by(strainID) %>%
+ count(COG_category_single) %>%
+ mutate(f = n/sum(n),
+ type = "all_genes")
+
+pcogs_mgsp <- bind_rows(cog_pmut_mgsp, cog_everything_mgsp) %>%
+ filter(!is.na(COG_category_single)) %>%
+ ggplot() +
+ geom_bar(aes(x = fct_reorder(COG_category_single, n, .desc = F), y=n, fill = type), stat="identity",
+ position = position_dodge( preserve = "total")) +
+ scale_fill_brewer(palette = "Set1") +
+ geom_point(data = tibble(x = c("D", "M", "M", "P"),
+ y = c(700, 700, 700, 700),
+ strainID = c("HAMBI_1972", "HAMBI_1977", "HAMBI_2443", "HAMBI_2659")),
+ aes(x = x, y = y), shape = 8) +
+ facet_wrap(~ strainID, nrow = 2) +
+ labs(x = "COG Category", y = "Fraction genes in COG category", fill = "Subset") +
+ scale_y_continuous(
+ "N all genes",
+ sec.axis = sec_axis(~ . * 1/250, name = "N parallel genes")
+ ) +
+ theme_bw() +
+ theme(
+ panel.grid = element_blank(),
+ strip.background = element_blank(),
+ panel.background = element_blank(),
+ legend.position = "bottom"
+ )
+```
+
+::: {#fig-06}
+```{r}
+#| fig.width: 8
+#| fig.height: 6
+#| echo: false
+pcogs_mgsp
+```
+
+Fraction of genes in each Cluster of Orthologous Groups (COG) category for genes with a COG annotation (red bars, left y-axis) and only genes exhibiting significant genomic parallelism (blue bars, right y-axis) for the four most abundant HAMBI genomes. COG categories "D - Cell cycle control, cell division, chromosome partitioning" for HAMBI_1972, "M - Cell wall/membrane/envelope biogenesis" for HAMBI_1977, and "P - Inorganic ion transport and metabolism" for HAMBI_2659 are significantly enriched (denoted with *) in the parallel mutations compared to the rest of the genome for these species.
+:::
+
+
+```{r}
+ggsave(here::here(figs, "COG_enrich_wgs.svg"), pcogs_mgsp, width=7, height=5,
+ units="in", device="svg")
+```
+
+| strainID | Abundant in metagenomes | COG category | q value |
+|------------|-------------------------|-------------------------------------------------------------------|---------|
+| HAMBI_1972 | yes | D - Cell cycle control, cell division, chromosome partitioning | 0.006 |
+| HAMBI_2443 | no | M - Cell wall/membrane/envelope biogenesis | 0.013 |
+| HAMBI_1977 | yes | M - Cell wall/membrane/envelope biogenesis | 0.015 |
+| HAMBI_2659 | yes | P - Inorganic ion transport and metabolism | 0.031 |
diff --git a/R/wgs/references.bib b/R/wgs/references.bib
new file mode 100644
index 0000000..5944c81
--- /dev/null
+++ b/R/wgs/references.bib
@@ -0,0 +1,88 @@
+
+@article{storey2003a,
+ title = {Statistical significance for genomewide studies},
+ author = {Storey, John D. and Tibshirani, Robert},
+ year = {2003},
+ month = {07},
+ date = {2003-07-25},
+ journal = {Proceedings of the National Academy of Sciences},
+ pages = {9440--9445},
+ volume = {100},
+ number = {16},
+ doi = {10.1073/pnas.1530509100},
+ url = {http://dx.doi.org/10.1073/pnas.1530509100},
+ langid = {en}
+}
+
+@article{tenaillon2016,
+ title = {Tempo and mode of genome evolution in a 50,000-generation experiment},
+ author = {Tenaillon, Olivier and Barrick, Jeffrey E. and Ribeck, Noah and Deatherage, Daniel E. and Blanchard, Jeffrey L. and Dasgupta, Aurko and Wu, Gabriel C. and Wielgoss, {Sébastien} and Cruveiller, {Stéphane} and {Médigue}, Claudine and Schneider, Dominique and Lenski, Richard E.},
+ year = {2016},
+ month = {08},
+ date = {2016-08},
+ journal = {Nature},
+ pages = {165--170},
+ volume = {536},
+ number = {7615},
+ doi = {10.1038/nature18959},
+ url = {http://dx.doi.org/10.1038/nature18959},
+ langid = {en}
+}
+
+@article{shoemaker2021,
+ title = {Molecular Evolutionary Dynamics of Energy Limited Microorganisms},
+ author = {Shoemaker, William R and Polezhaeva, Evgeniya and Givens, Kenzie B and Lennon, Jay T},
+ editor = {Battistuzzi, Fabia Ursula},
+ year = {2021},
+ month = {07},
+ date = {2021-07-13},
+ journal = {Molecular Biology and Evolution},
+ pages = {4532--4545},
+ volume = {38},
+ number = {10},
+ doi = {10.1093/molbev/msab195},
+ url = {http://dx.doi.org/10.1093/molbev/msab195},
+ langid = {en}
+}
+
+@article{good2017,
+ title = {The dynamics of molecular evolution over 60,000 generations},
+ author = {Good, Benjamin H. and McDonald, Michael J. and Barrick, Jeffrey E. and Lenski, Richard E. and Desai, Michael M.},
+ year = {2017},
+ month = {10},
+ date = {2017-10-18},
+ journal = {Nature},
+ pages = {45--50},
+ volume = {551},
+ number = {7678},
+ doi = {10.1038/nature24287},
+ url = {http://dx.doi.org/10.1038/nature24287},
+ langid = {en}
+}
+
+@article{durrant2020a,
+ title = {A Bioinformatic Analysis of Integrative Mobile Genetic Elements Highlights Their Role in Bacterial Adaptation},
+ author = {Durrant, Matthew G. and Li, Michelle M. and Siranosian, Benjamin A. and Montgomery, Stephen B. and Bhatt, Ami S.},
+ year = {2020},
+ month = {01},
+ date = {2020-01},
+ journal = {Cell Host & Microbe},
+ pages = {140--153.e9},
+ volume = {27},
+ number = {1},
+ doi = {10.1016/j.chom.2019.10.022},
+ url = {http://dx.doi.org/10.1016/j.chom.2019.10.022},
+ langid = {en}
+}
+
+@article{camargo2023,
+ title = {Identification of mobile genetic elements with geNomad},
+ author = {Camargo, Antonio Pedro and Roux, Simon and Schulz, Frederik and Babinski, Michal and Xu, Yan and Hu, Bin and Chain, Patrick S. G. and Nayfach, Stephen and Kyrpides, Nikos C.},
+ year = {2023},
+ month = {09},
+ date = {2023-09-21},
+ journal = {Nature Biotechnology},
+ doi = {10.1038/s41587-023-01953-y},
+ url = {http://dx.doi.org/10.1038/s41587-023-01953-y},
+ langid = {en}
+}
diff --git a/R/wgs/utils_parallelism.R b/R/wgs/utils_parallelism.R
new file mode 100644
index 0000000..6eef1d8
--- /dev/null
+++ b/R/wgs/utils_parallelism.R
@@ -0,0 +1,655 @@
+
+# Nucleotide parallelism --------------------------------------------------
+
+plot_nuc_survival <- function(df, ylim, ybreaks, ncol){
+ ggplot(df, aes(x=m, y=value, color=name, linetype = name)) +
+ geom_step(direction = "mid") +
+ scale_x_continuous(limits = c(0.5, 3.5),
+ breaks = c(1, 2, 3),
+ minor_breaks = NULL) +
+ scale_y_log10(breaks = {{ ybreaks }},
+ labels = scales::trans_format("log10", scales::math_format(10^.x))) +
+ labs(x=TeX("Nucleotide multiplicitiy, \\textit{m}"),
+ y=TeX("Total mutations $\\geq \\textit{m}$"),
+ color = NULL) +
+ scale_color_manual(values = c("blue", "red")) +
+ scale_linetype(guide = 'none') +
+ facet_wrap(~label, ncol = {{ ncol }}) +
+ annotation_logticks(sides = "l", color = "grey40") +
+ coord_cartesian(ylim = c(1, {{ ylim }})) +
+ theme_bw() +
+ theme(
+ strip.placement = 'outside',
+ strip.background = element_blank(),
+ panel.grid = element_blank(),
+ legend.position = "bottom"
+ )
+}
+
+# Gene parallelism --------------------------------------------------------
+
+# Prepare different data sources ------------------------------------------
+
+make_gene_length_df <- function(df, strainID){
+ # takes a combined dataframe for all HAMBI species total length of gene-wise
+ # length of nonsynonymous coding sites and returns a reduced dataframe for the
+ # focal species strainID
+ df %>%
+ # filter on strainID
+ filter(strainID == !!strainID) %>%
+ dplyr::select(locus_tag, l_i = ns_length) %>%
+ # remove NAs
+ drop_na() %>%
+ # convert to data.frame
+ column_to_rownames(var = "locus_tag") %>%
+ data.frame()
+}
+
+make_raw_mut_df <- function(df, gene_lengths) {
+ # formats and prepares a dataframe of observed pass gene hits per replicate
+ # and includes every CDS for the genome even if there were no hits to the gene
+ # (ie some rowSums are zero). Rownames are locus_tag and columns are the
+ # individual replicates where at least 1 mutation was detected
+ df %>%
+ # first check to remove any noncoding mutations
+ filter(locus_tag %in% rownames(gene_lengths)) %>%
+ summarize(n = n(),
+ .by = c("replicate", "locus_tag")) %>%
+ pivot_wider(
+ names_from = "replicate",
+ values_from = "n",
+ values_fill = 0
+ ) %>%
+ left_join(dplyr::select(
+ rownames_to_column(gene_lengths, var = "locus_tag"),
+ locus_tag
+ ),
+ .,
+ by = join_by(locus_tag)) %>%
+ mutate(across(-locus_tag, ~ replace_na(.x, 0))) %>%
+ column_to_rownames(var = "locus_tag") %>%
+ data.frame()
+}
+
+make_multiplicity_df <- function(df, gene_lengths) {
+ # calculate gene multiplicity for each gene
+ # m_i = n_i * mean(l_i)/l_i
+ df %>%
+ # first check to remove any noncoding mutations
+ filter(locus_tag %in% rownames(gene_lengths)) %>%
+ summarize(
+ n_i = n(),
+ n_replicate = n_distinct(replicate),
+ .by = c("locus_tag")
+ ) %>%
+ select(locus_tag, n_i, n_replicate) %>%
+ left_join(rownames_to_column(gene_lengths, var = "locus_tag"),
+ .,
+ by = join_by(locus_tag)) %>%
+ replace_na(list(n_i = 0, n_replicate = 0)) %>%
+ mutate(m_i = n_i * mean(l_i) / l_i) %>%
+ column_to_rownames(var = "locus_tag") %>%
+ data.frame()
+}
+
+prepare_all_input <- function(df, gene_table){
+ # prepares a list with the required dataframes and summary values to compute the
+ # gscore or to simulate the G score
+
+ strainID <- unique(df$strainID)
+ gene_lengths <- make_gene_length_df(gene_table, strainID)
+ raw_mut_df <- make_raw_mut_df(df, gene_lengths)
+ multiplicity_df <- make_multiplicity_df(df, gene_lengths)
+
+ return(lst(
+ strainID = strainID,
+ # number of replicates with at least on gene hit
+ rep_num = ncol(raw_mut_df),
+ # total gene hits in each replicate
+ mut_reps = colSums(raw_mut_df),
+ # total gene hits across all genes and replicates
+ mut_tot = sum(colSums(raw_mut_df)),
+ gene_lengths = gene_lengths,
+ multiplicity_df = multiplicity_df,
+ raw_mut_df = raw_mut_df
+ ))
+}
+
+# G scores ----------------------------------------------------------------
+
+G_score <- function(input_list){
+ # number of non-syn sites in gene i
+ l_i <- input_list$gene_lengths$l_i
+ # compute the sum of non-syn sites in genome
+ Ltot <- sum(l_i)
+ # compute the average non-syn sites per gene
+ Lavg <- mean(l_i)
+ # compute the number of genes in the genome
+ n_genes <- length(l_i)
+ # number of mutations in gene i across all replicates
+ n_i <- input_list$multiplicity_df$n_i
+ # total number of non-syn hits across genes and replicates
+ Ntot <- input_list$mut_tot
+ # compute the expected multiplicity value
+ m_exp <- Ntot / n_genes
+ # compute gene multiplicity for gene i
+ m_i <- n_i * Lavg / l_i
+ # compute the G score for each gene
+ g_score <- n_i * log(m_i / m_exp)
+ # net increase in log-likelihood compared to the null model of m_i / m_exp =1
+ # normalized to the total mutations
+ delta_l <- sum(g_score, na.rm=T) / Ntot
+ return(delta_l)
+}
+
+G_score_prep <- function(input_list){
+ # Special G score function preparing constants for use in G_score_resampled
+ # so we don't have to unnecessarily recalculate all the constants for each
+ # resample
+
+ # number of non-syn sites in gene i
+ l_i <- input_list$gene_lengths$l_i
+ # compute the sum of non-syn sites in genome
+ Ltot <- sum(l_i)
+ # compute the average non-syn sites per gene
+ Lavg <- mean(l_i)
+ # Null probability of non-syn hit on a gene based only on the
+ # number of nonsynonymous sites per gene relative to the total number of
+ # nonsynonymous sites in the genome (gene_probs = l_i/Ltot).
+ gene_probs <- l_i/Ltot
+ # number of genes in genome. In the simulation this only includes coding
+ # sequences and so we don't need to filter out any genes
+ n_genes <- length(l_i)
+ # total number of non-syn hits across genes and replicates
+ Ntot <- input_list$mut_tot
+ # compute the expected multiplicity value
+ m_exp <- Ntot / n_genes
+
+ lst(l_i=l_i, Ltot=Ltot, Lavg=Lavg, n_genes=n_genes,
+ Ntot=Ntot, m_exp=m_exp, gene_probs=gene_probs)
+}
+
+G_score_resampled <- function(df, input_list){
+ # remaining G score function that uses constants from G_score_prep while only
+ # calculating n_i dynamically
+
+ # number of mutations in gene i summed across all replicates
+ n_i <- rowSums(df)
+ # compute gene multiplicity for gene i
+ m_i = n_i * input_list$Lavg / input_list$l_i
+ # compute the G score for each gene
+ g_score = n_i * log(m_i / input_list$m_exp)
+ # net increase in log-likelihood compared to the null model of m_i / m_exp =1
+ # normalized to the total mutations
+ delta_l = sum(g_score, na.rm=T) / input_list$Ntot
+}
+
+# G score permutation -----------------------------------------------------
+
+mut_hit_sim <- function(mut_reps, gene_probs, resamples){
+ # randomly sample (n=resamples) the observed number of nonsynonymous mutations
+ # from each replicate (vector mut_reps) with probability based only on the
+ # number of nonsynonymous sites per gene relative to the total number of
+ # nonsynonymous sites in the genome (gene_probs = l_i/Ltot). Returns a list of
+ # length mut_reps where for each replicate in mut_reps there is a n=resamples
+ # column and n=gene_number row matrix
+ map(mut_reps, \(x) rmultinom(n = resamples, size = x, prob = gene_probs))
+}
+
+convert_nameless_mat2_df <- function(m, pop_num, gene_names){
+ # take output from mut_hit_sim function (list of sample x gene matrices) and
+ # reformat it so that it returns a dataframe with locus_tag for each gene as
+ # rownames and sampled replicates are columns
+ names(m) <- LETTERS[seq_along(1: pop_num)]
+ data.frame(m, row.names = gene_names)
+}
+
+gscore_p_value <- function(input_list, emp_g_score, samples){
+ G_score_prepped <- G_score_prep(input_list)
+ # randomly sample hits based on gene length and split to a convenient form
+ multinomhitsamp_split <- mapply(asplit,
+ MARGIN = 2,
+ x = mut_hit_sim(input_list$mut_reps,
+ G_score_prepped$gene_probs,
+ samples))
+ # for each random sample convert to a more useful dataframe form
+ g_reps <- seq_along(1:samples) %>%
+ map(\(x) convert_nameless_mat2_df(multinomhitsamp_split[x,],
+ input_list$rep_num,
+ rownames(input_list$gene_lengths))) %>%
+ # calculate the g_score for each sample
+ map_dbl(\(x) G_score_resampled(x, G_score_prepped))
+
+ return(length(g_reps[g_reps > emp_g_score]) / length(g_reps))
+}
+
+
+gscore_p_value2 <- function(input_list, emp_g_score, samples){
+ # also computes the observed distribution of hits per locus tag
+ # See Fig S3 from this paper: https://doi.org/10.1038/s41559-020-1128-3
+ G_score_prepped <- G_score_prep(input_list)
+ # randomly sample hits based on gene length and split to a convenient form
+ multinomhitsamp_split <- mapply(asplit,
+ MARGIN = 2,
+ x = mut_hit_sim(input_list$mut_reps,
+ G_score_prepped$gene_probs,
+ samples))
+ # for each random sample convert to a more useful dataframe form
+ multinomhitsamp_split_df <- seq_along(1:samples) %>%
+ map(\(x) convert_nameless_mat2_df(multinomhitsamp_split[x,],
+ input_list$rep_num,
+ rownames(input_list$gene_lengths)))
+ # get the observed distribution of "hits" i.e. non-syn mutations
+ observed_hit_distrubution <- input_list$raw_mut_df %>%
+ rownames_to_column(var = "locus_tag") %>%
+ pivot_longer(-locus_tag) %>%
+ filter(value > 0) %>%
+ summarize(n_reps = n_distinct(name), .by = c(locus_tag)) %>%
+ summarize(n = n(), .by = c(n_reps)) %>%
+ mutate(type = "Observed")
+
+ null_hit_distribution <- multinomhitsamp_split_df %>%
+ map_dfr(\(x) pivot_longer(rownames_to_column(x, var = "locus_tag"), -locus_tag), .id = "resample") %>%
+ filter(value > 0) %>%
+ summarize(n_reps = n_distinct(name), .by = c(resample, locus_tag)) %>%
+ summarize(n = n()/samples, .by = c(n_reps)) %>%
+ mutate(type = "Null")
+
+ # calculate the g_score for each sample
+ g_reps <- map_dbl(multinomhitsamp_split_df, \(x) G_score_resampled(x, G_score_prepped))
+
+ # calculate p values
+ g_score_pvalue <- length(g_reps[g_reps > emp_g_score]) / length(g_reps)
+
+ return(lst(g_score_pvalue = g_score_pvalue,
+ hit_distribution = bind_rows(observed_hit_distrubution, null_hit_distribution)))
+}
+
+# Calculating gene-wise significance --------------------------------------
+
+from_parallelism_statistics <- function(input_list) {
+ # number of non-syn sites in gene i
+ l_i <- input_list$gene_lengths$l_i
+ # compute the sum of non-syn sites in genome
+ Ltot <- sum(l_i)
+ # number of mutations in gene i across all replicates
+ n_i <- input_list$multiplicity_df$n_i
+ # total number of non-syn hits across genes and replicates
+ Ntot <- input_list$mut_tot
+ # compute the fraction of total non-syn sites occupied by each gene
+ ps <- l_i / Ltot
+ # compute the expected number of mutations in each gene based on gene length
+ n_exp <- Ntot * l_i / Ltot
+
+ lst(
+ "gene_names" = rownames(input_list$gene_lengths),
+ "Ls" = l_i,
+ "ns" = n_i,
+ "Ltot" = Ltot,
+ "Ntot" = Ntot,
+ "ps" = ps,
+ "n_exp" = n_exp
+ )
+}
+
+calculate_poisson_log_survival <- function(n_exp, ns, ethresh, filter = FALSE){
+ # survival function = 1-CDF
+ survivals <- (1 - ppois(ns - 0.1, n_exp))
+ # for the case if there are so many mutations in some gene that its survival is 0
+ if (filter) {
+ survivals[survivals==0] <- min(survivals[survivals>0])/1e10
+ }
+ logpvalues <- rep(0, each = length(survivals))
+ logpvalues[survivals > ethresh] = -log(survivals[survivals > ethresh])
+ logpvalues[survivals <= ethresh] = -(ns * log(ns / n_exp +
+ (ns == 0)) + ns - n_exp)[survivals <= ethresh]
+ return(logpvalues)
+}
+
+calculate_unnormalized_survival_from_vector <- function(xs, min_x=NULL, max_x=NULL, min_p=1e-10){
+ if (is.null(min_x)){
+ min_x <- min(xs)-1
+ }
+ if (is.null(max_x)){
+ max_x <- max(xs)+1
+ }
+ unique_xs <- unique(xs)
+ unique_xs <- sort(append(unique_xs, c(min_x, max_x)))
+ num_observations <- map_dbl(unique_xs, \(x) sum(xs >= x))
+
+ # So that we can plot CDF, SF on log scale
+ num_observations[0] <- num_observations[0] - min_p
+ num_observations[1] <- num_observations[1] - min_p
+ num_observations[-1] <- num_observations[-1] + min_p
+
+ return(list(unique_xs, num_observations))
+}
+
+map_pval_and_probs = function(observed_p, logpvalues, probabilities, ns_grid, nmin){
+ sum(map2_dbl(logpvalues, probabilities,
+ \(x, y) sum((x >= observed_p) * (ns_grid >= nmin) * y)))
+}
+
+format_survivalcurves_2plot <- function(observed_ps, survivals, observed_pvalue_survival,
+ pstar, num_significant){
+ df_step <- tibble(
+ obs_p = observed_ps,
+ Survival_Expected = survivals,
+ Survival_Observed = observed_pvalue_survival
+ ) %>%
+ pivot_longer(-obs_p) %>%
+ mutate(name = factor(
+ name,
+ levels = c("Survival_Observed", "Survival_Expected"),
+ labels = c("Survival Observed", "Survival Expected")
+ ))
+
+ df_seg <- tibble(
+ x = c(min(observed_ps), pstar),
+ y = c(num_significant, 0),
+ xend = c(pstar, pstar),
+ yend = c(num_significant, num_significant)
+ )
+
+ df_ann <- tibble(x = pstar, y = num_significant)
+
+ # return
+ lst(df_step = df_step,
+ df_seg = df_seg,
+ df_ann = df_ann
+ )
+}
+
+
+make_gene_table <- function(pooled_pvalues, strainID, obs_gene_multiplicity, raw_hits, pstar){
+ enframe(pooled_pvalues, name = "locus_tag", value = "neg_log10P") %>%
+ mutate(strainID = strainID, pstar = pstar) %>%
+ left_join(., rownames_to_column(obs_gene_multiplicity, var = "locus_tag"), join_by(locus_tag)) %>%
+ left_join(., rownames_to_column(raw_hits, var = "locus_tag"), join_by(locus_tag)) %>%
+ dplyr::rename(nonsyn_sites = l_i,
+ observed_hits_n_i = n_i,
+ gene_multiplicity_m_i = m_i)
+}
+
+NullGeneLogpSurvivalFunction <- function(mylist, nmin=3){
+
+ # Get formatted input
+ fps <- from_parallelism_statistics(mylist)
+
+ # Calculate log-P value for each gene
+ logpvalues_par <- calculate_poisson_log_survival(fps$n_exp, fps$ns,
+ ethresh=1e-30, filter = TRUE)
+ names(logpvalues_par) <- fps$gene_names
+
+ # Collect P values for genes with at least nmin (default = 3) mutations across
+ # all populations this limits the number of low P values driven primarily by
+ # gene length
+ pooled_pvalues <- sort(logpvalues_par[fps$ns >= nmin])
+
+ # Un-normalized poisson survival vector
+ tsrv <- calculate_unnormalized_survival_from_vector(pooled_pvalues)
+ observed_ps <- pluck(tsrv, 1)
+ observed_pvalue_survival <- pluck(tsrv, 2)
+
+ # get required inputs for the survival analysis to estimate significance
+ ns_grid <- rep(0:399)
+ alpha <- 0.05
+
+ logpvalues <- map(fps$n_exp, \(x) calculate_poisson_log_survival(x, ns = ns_grid, ethresh = 1e-20, filter = FALSE))
+ logprobabilities <- map(fps$n_exp, \(x) ns_grid * log(x) - lgamma(ns_grid + 1) - x)
+ probabilities <- map(logprobabilities, exp)
+ survivals <- map_dbl(observed_ps, \(x) map_pval_and_probs(x, logpvalues, probabilities, ns_grid, nmin))
+ threshold_idx <- which(survivals/observed_pvalue_survival <= alpha)[1]
+ pstar <- observed_ps[threshold_idx]
+ num_significant <- observed_pvalue_survival[threshold_idx]
+
+ # Return pertinent results
+ df2plot <- format_survivalcurves_2plot(observed_ps, survivals, observed_pvalue_survival,
+ pstar, num_significant)
+ gene_table <- make_gene_table(pooled_pvalues, mylist$strainID, mylist$multiplicity_df, mylist$raw_mut_df, pstar)
+
+ return(lst("df2plot" = df2plot,
+ "gene_table" = gene_table,
+ "observed_ps" = observed_ps,
+ "pooled_pvalues" = pooled_pvalues,
+ "observed_pvalue_survival" = observed_pvalue_survival,
+ "survivals" = survivals,
+ "pstar" = pstar,
+ "num_significant" = num_significant))
+
+}
+
+plotcurve <- function(input_list){
+ ggplot() +
+ geom_segment(
+ data = input_list$df_seg,
+ aes(
+ x = x,
+ y = y,
+ xend = xend,
+ yend = yend
+ ),
+ linetype = "dashed",
+ linewidth = 0.25
+ ) +
+ geom_step(data = input_list$df_step, aes(x = obs_p, y = value, color = name)) +
+ annotate("point", x = input_list$df_ann$x, y = input_list$df_ann$y) +
+ # annotate(
+ # "text",
+ # x = input_list$df_ann$x + 35,
+ # y = input_list$df_ann$y - 3,
+ # label = TeX("Critical P*\nthreshold = $1.6 \\times 10^{-4}$")
+ # ) +
+ labs(x = TeX("$-log_{10}P$"), y = "Number of genes") +
+ scale_color_manual(values = c("blue", "grey")) +
+ scale_y_log10(
+ breaks = c(1, 10, 100),
+ labels = scales::trans_format("log10", scales::math_format(10 ^.x))
+ ) +
+ annotation_logticks(sides = "l", color = "grey40") +
+ coord_cartesian(ylim = c(1, 1000)) +
+ theme_bw() +
+ theme(
+ strip.background = element_blank(),
+ panel.grid = element_blank(),
+ )
+}
+
+run_full_gene_parallelism_pipeline <- function(grouped, gene_lens, resamples){
+ # grouped is a tibble that has been grouped at the desired hierarchy level
+ # from the experiment. It should also contain a variable called "groupid" that
+ # is numeric variable unique to each group. The tibble gene_lens contains the
+ # number of non-synonymous sites found in each locus tag for all 23 HAMBI
+ # species in a variable called "ns_length." the variable "resamples is the
+ # number of resamplings you want to perform for the significance testing of
+ # the G score
+
+ # split into list of tibbles
+ grouped_split <- group_split(grouped)
+
+ # prepare ecah grouped input for gene parallelism analysis
+ grouped_split_prepped <- grouped_split %>%
+ map(\(x) prepare_all_input(x, gene_lens), .progress="inputprep")
+
+ # Calculate empirical G scores for non-syn mutations
+ emp_g_scores <- map(grouped_split_prepped, \(x) G_score(x), .progress="gscore")
+
+ # Calculate significance of empirical G scores
+ #g_score_pvals_lump <- map2_dbl(grouped_split_prepped, emp_g_scores, \(x, y) gscore_p_value(x, y, resamples))
+ g_score_pvals_lump2 <- map2(grouped_split_prepped, emp_g_scores, \(x, y) gscore_p_value2(x, y, resamples), .progress="gscore_pval")
+
+ # Estimate gene-wise p values and compare to null dependending only on total hits and gene length
+ nglpsf_out <- map(grouped_split_prepped, \(x) NullGeneLogpSurvivalFunction(x, nmin = 1), .progress="LogPSurvival")
+
+ # format G scores into a tibble
+ output_gscore <-
+ unnest(enframe(emp_g_scores, name = "groupid", value = "observed_g_score"), cols = observed_g_score) %>%
+ left_join(unnest(enframe(map(g_score_pvals_lump2, \(x) pluck(x, 1)), name = "groupid", value = "pvalue"), cols = "pvalue")) %>%
+ left_join(group_keys(grouped))
+
+ output_simulated_reps <- g_score_pvals_lump2 %>%
+ map(\(x) pluck(x, 2)) %>%
+ map_dfr(\(x) x, .id = "groupid") %>%
+ mutate(groupid = as.integer(groupid)) %>%
+ left_join(group_keys(grouped))
+
+ output_gene_table <- map_dfr(nglpsf_out, \(x) pluck(x, "gene_table"), .id = "groupid") %>%
+ mutate(groupid = as.integer(groupid)) %>%
+ left_join(group_keys(grouped))
+
+ getdfl2plot <- map(nglpsf_out, \(x) pluck(x, "df2plot"))
+ output_df2plot <- map_dfr(getdfl2plot, \(x) pluck(x, "df_step"), .id = "groupid") %>%
+ left_join(map_dfr(getdfl2plot, \(x) pluck(x, "df_seg"), .id = "groupid")) %>%
+ mutate(groupid = as.integer(groupid)) %>%
+ left_join(group_keys(grouped))
+
+ return(lst(output_gscore = output_gscore,
+ output_simulated_reps = output_simulated_reps,
+ output_gene_table = output_gene_table,
+ output_df2plot = output_df2plot))
+}
+
+plot_sims <- function(input_list, page){
+ input_list$output_simulated_reps %>%
+ mutate(label = paste(strainID)) %>%
+ mutate(n_reps = factor(n_reps),
+ labels = label) %>%
+ ggplot() +
+ geom_col(aes(x = n_reps, y = n, fill = type),
+ position = position_dodge2(preserve = "single")) +
+ labs(x = "Number of replicates with a hit in the same gene", y = "Number of genes with a hit") +
+ scale_y_sqrt() +
+ scale_fill_brewer(palette = "Set1") +
+ facet_wrap_paginate(vars(label), scales = "free_y",
+ labeller = labeller(label_wrap_gen()),
+ ncol = 4, nrow = 3, page = {{ page }}) +
+ theme_bw() +
+ theme(
+ strip.placement = 'outside',
+ strip.background = element_blank(),
+ panel.grid = element_blank(),
+ legend.position = "bottom"
+ )
+}
+
+# Functions for Jaccard index resampling ----------------------------------
+
+prep4jaccard_resample <- function(gene_table_filt, genetable, strainID, ...){
+ mygroup <- gene_table_filt %>%
+ filter(strainID == {{ strainID }}) %>%
+ group_by(...) %>%
+ mutate(groupid = cur_group_id())
+
+ mysplit <- mygroup %>%
+ group_split() %>%
+ map(\(x) pull(x, locus_tag))
+
+ genetable_filt <- genetable %>%
+ filter(strainID == {{ strainID }}) %>%
+ dplyr::select(locus_tag, ns_length) %>%
+ # remove NAs
+ drop_na() %>%
+ arrange(locus_tag)
+
+ locus_tag <- pull(genetable_filt, locus_tag)
+ l_i <- pull(genetable_filt, ns_length)
+ gene_probs <- l_i/sum(l_i)
+
+ return(
+ lst(mygroup, mysplit, locus_tag, gene_probs)
+ )
+}
+
+jaccard <- function(a, b) {
+ intersection <- length(intersect(a, b))
+ union <- length(a) + length(b) - intersection
+ return (intersection/union)
+}
+
+pairsamplejaccard <- function(x, y, locus_tag, gene_probs){
+ jaccard(
+ sample(x = locus_tag, size = length(x), prob = gene_probs, replace = FALSE),
+ sample(x = locus_tag, size = length(y), prob = gene_probs, replace = FALSE)
+ )
+}
+
+resample_jaccard <- function(nresamples, prepped4jaccard_resample) {
+
+ # defining inputs
+ input_gene_lists <- prepped4jaccard_resample$mysplit
+ genome_locus_tags <- prepped4jaccard_resample$locus_tag
+ genome_gene_probs <- prepped4jaccard_resample$gene_probs
+
+ # some housekeeping
+ myindxgroups <- prepped4jaccard_resample$mygroup %>%
+ ungroup() %>%
+ distinct(groupid, bacteria_anc_evo, c_elegans, strep_conc_ugml) %>%
+ mutate(groupname = paste0("bact: ", bacteria_anc_evo, ", wrm: ", c_elegans, ", strep: ", strep_conc_ugml)) %>%
+ dplyr::select(groupid, groupname)
+
+ # generate unique combinations
+ combos <- combn(1:length(input_gene_lists), 2)
+
+ # convert combo ids to variable names for pairs
+ myindxgroupnames <- as.data.frame(t(combos)) %>%
+ `colnames<-`(c("groupid1", "groupid2")) %>%
+ rownames_to_column(var = "treat_pair") %>%
+ mutate(treat_pair = as.integer(treat_pair)) %>%
+ left_join(rename(myindxgroups, groupid1 = groupid, groupname1 = groupname),
+ by = join_by(groupid1)) %>%
+ left_join(rename(myindxgroups, groupid2 = groupid, groupname2 = groupname),
+ by = join_by(groupid2))
+
+ # First inspect observed jaccard indexes for all pairs
+ jc_obs <-
+ map_dbl(asplit(combos, MARGIN = 2),
+ \(x) jaccard(input_gene_lists[[x[1]]],
+ input_gene_lists[[x[2]]])
+ )
+
+ # get only combos with a Jaccard index > 0 (i.e. any overlapping elements)
+ jc_obs_gt0 <- jc_obs[which(jc_obs > 0)]
+
+ if (length(jc_obs_gt0) > 1) {
+ ind2iterate <- asplit(combos[, which(jc_obs_gt0 > 0)], MARGIN = 2)
+ } else {
+ ind2iterate <- 1
+ }
+
+ # resampled jaccard index
+ jc_resampled <- replicate(nresamples,
+ map_dbl(
+ ind2iterate,
+ \(x) pairsamplejaccard(input_gene_lists[[x[1]]],
+ input_gene_lists[[x[2]]],
+ genome_locus_tags,
+ genome_gene_probs)
+ )
+ )
+
+ if (length(jc_obs_gt0) > 1) {
+ # Vector of p values for each combination with multiple comparisons
+ ps <- tibble(treat_pair = which(jc_obs > 0), p_values = rowSums(jc_resampled >= jc_obs_gt0) / rep(nresamples, length(jc_obs_gt0)))
+ } else {
+ # single p value for the single sample case
+ ps <- tibble(treat_pair = which(jc_obs > 0), p_values = length(jc_resampled[jc_resampled >= jc_obs_gt0]) / nresamples)
+ }
+ left_join(ps, myindxgroupnames)
+}
+
+
+# hypergeometric test -----------------------------------------------------
+
+enricher <- function(n_background, n_par, n_cog_background, n_cog_par, over=TRUE){
+ # n_background: total genes with COG annotation
+ # n_par: Set of target genes of interest. In this case the total number of COG
+ # annotated genes that have significant parallelism
+ # n_cog_background: set of genes from background annotated with the COG of interest
+ # n_cog_par: number of genes in a COG of interest from the total number of COG annotated
+ # genes w/ parallelism
+ if (over) {
+ phyper(n_cog_par-1, n_cog_background, n_background-n_cog_background, n_par, lower.tail= FALSE)
+ } else {
+ phyper(n_cog_par, n_cog_background, n_background-n_cog_background, n_par, lower.tail= TRUE)
+ }
+}
diff --git a/README.md b/README.md
new file mode 100644
index 0000000..8791bc0
--- /dev/null
+++ b/README.md
@@ -0,0 +1,45 @@
+# hambiDoubleEvo
+
+[Click here to view rendered notebooks of the analysis.](https://slhogle.github.io/hambiDoubleEvo/)
+
+## Manuscript:
+
+### Published record
+
+**Temporal changes in the role of species sorting and evolution determine community dynamics.** Hoffman J‡, Hogle SL‡, Hiltunen T◇, Becks L◇. *Ecol. Lett.* (2024/5) [doi:]()
+
+‡ equal contribution, ◇ Corresponding author and equal contribution
+
+### Preprint
+
+**Temporal changes in the role of species sorting and evolution determine community dynamics.** Hoffman J‡, Hogle SL‡, Hiltunen T◇, Becks L◇. (2024) [doi: 10.21203/rs.3.rs-4647074/v1](https://doi.org/10.1101/2023.10.31.565024)
+
+‡ equal contribution, ◇ Corresponding author and equal contribution
+
+## Data and Code
+
+Data and code here is provided under GPL3. Feel free to use or remix as you see fit.
+
+[](https://zenodo.org/badge/latestdoi/411568623)
+
+### Project structure
+
+1. `/R` contains R scripts
+2. `/data` contains data that has been processed in some way for later use
+3. `/_data_raw` contains unprocessed data scraped from compute cluster
+4. `/figs` contains figures generated from R scripts
+
+## Availability
+
+The rendered project site is available at . The website has been produced using [Quarto notebooks](https://quarto.org/).
+
+This GitHub repository () hosts the code and data for this project. The rendered webpage can be fully recreated using the code at