S3. Annotation of cell sets in AIT99 using TDT

[hkir-dev]

Epic:

Description

User: The same scientists from Use Case 2 #99 .  
Use Case: We have confirmed that AIT99 is set up properly on our computers, and now we want to start annotating the cell sets.
What we have:

• AIT99 loaded into TDT
• Additional context (e.g., data visualization tools for the data/metadata, PubMed, etc.)

What we need to know:

• I’ve just opened TDT and see a bunch of tables.  What do they mean? Documentation about tables and basic editing #104 Simplify table names #106
• I realize I forgot to include a column for neurotransmitters! How do I add it? - Add new user annotation column #108 for NT specifically - see comments below
• A while later I’m realizing that my “neurotransmitter” column has “GABA”, “GABAergic”, and “GABA-ergic” all meaning the same thing.  How do I ensure that this field has a controlled vocabulary? - Documentation on limiting column values #109 _ +see comments below_
• After reviewing the UMAPs more carefully, it turns out cl_1234 is an outlier. How do I remove this cluster? - clarification needed - see below
• Cl_2345 should be a Pvalb interneuron and not an Sst interneuron. How do I change this? Documentation about tables and basic editing #104
• I’ve been studying cl_456 carefully and have identified new marker genes, associated papers, and cell ontology terms, and have some unscripted notes I’d like to add.  How do I add this information and indicate that I’m the person making these changes? Documentation about tables and basic editing #104 - see also comment below

Tasks

• Documentation about tables and basic editing #104
• Add new user annotation column #108
• Simplify table names #106
• Documentation on limiting column values #109
• Tracking users making change #110

[dosumis]

I realize I forgot to include a column for neurotransmitters! How do I add it?

• Add new user annotation column #108
• Our recommendation for neurotransmitters is to record them as a labelset not a column. This is how ABC does it and how we agreed to represent previously We have ontology terms for neuron type by NT that can be associated with each. If we will change this, we need to discuss further.

A while later I’m realizing that my “neurotransmitter” column has “GABA”, “GABAergic”, and “GABA-ergic” all meaning the same thing. How do I ensure that this field has a controlled vocabulary?

#109 - The more general issue here is how to restrict the contents of a column to some enumeration. This is should be possible via adding a new 'datatype' & using this to specify type for column. However, this is a more advanced operation and might be better managed with help from out team.

After reviewing the UMAPs more carefully, it turns out cl_1234 is an outlier. How do I remove this cluster?

@jeremymiller - does this refer to deleting a cluster from the taxonomy or to removing it from under a parent cell set?

How do I add this information and indicate that I’m the person making these changes?

New feature under discussion. Some work needed on schema first. See #110

[jeremymiller]

After reviewing the UMAPs more carefully, it turns out cl_1234 is an outlier. How do I remove this cluster?
Does this refer to deleting a cluster from the taxonomy or to removing it from under a parent cell set?
I had originally meant it as deleting a row from the table, but after some more thinking and discussions, it probably makes sense to keep the row itself but to remove it from the tree (e.g., delete parent cell set and add a comment that it should be removed as an outlier in a future taxonomy version). If we remove it from the table it may cause an disconnect with the h5ad file with the data until we delete those cells from that data file.

[dosumis]

After reviewing the UMAPs more carefully, it turns out cl_1234 is an outlier. How do I remove this cluster?

Does this refer to deleting a cluster from the taxonomy or to removing it from under a parent cell set?

I had originally meant it as deleting a row from the table, but after some more thinking and discussions, it probably makes sense to keep the row itself but to remove it from the tree (e.g., delete parent cell set and add a comment that it should be removed as an outlier in a future taxonomy version). If we remove it from the table it may cause an disconnect with the h5ad file with the data until we delete those cells from that data file.

Thanks - that makes more sense to me. We should probably have a standard way to track these in CAS. We can make a ticket to support.