Merge pull request #89 from jeromekelleher/arg-diagram

Arg diagram

Makefile

@@ -1,31 +1,37 @@
 DATA=data/arg.csv\
      data/mutations_perf.csv
 
-PERF_FIGURES=figures/mutations-perf.pdf\
+FIGURES=figures/mutations-perf.pdf\
 	figures/arg.pdf\
 	figures/sweeps-perf.pdf\
 	figures/gc-perf.pdf\
 	figures/dtwf-perf.pdf
 
-FIGURES=$(PERF_FIGURES)\
-	figures/mutated_tree.pdf\
-	figures/unmutated_tree.pdf\
-	figures/example_tree_sequence.pdf
+ILLUSTRATIONS=\
+	illustrations/mutated_tree.pdf\
+	illustrations/unmutated_tree.pdf\
+	illustrations/arg-ts.pdf\
+	illustrations/arg-ts-simplified.pdf\
+	illustrations/example_tree_sequence.pdf
 
-paper.pdf: paper.tex paper.bib ${DATA} ${FIGURES}
+paper.pdf: paper.tex paper.bib ${DATA} ${FIGURES} ${ILLUSTRATIONS}
 	pdflatex paper.tex
 	bibtex paper
 	pdflatex paper.tex
 	pdflatex paper.tex
 
-$(PERF_FIGURES):
+figures/%.pdf:
 	python3 evaluation/plot.py $*
 
-figures/mutated_tree.svg figures/unmutated_tree.svg : pretty_pictures.py
-	python3 $<
+illustrations/mutated_tree.svg illustrations/unmutated_tree.svg : pretty_pictures.py
+	python3 $< mutated-tree
+
+illustrations/arg-ts.svg: pretty_pictures.py
+	python3 $< arg-ts
 
 %.pdf : %.svg
-	inkscape $< --export-pdf=$@
+	chromium --headless --no-margins --print-to-pdf=$@ $<
+	# inkscape $< --export-pdf=$@
 
 %.pdf : %.ink.svg
 	inkscape $< --export-pdf=$@

illustrations/.gitignore

@@ -0,0 +1,2 @@
+*
+!.gitignore

paper.tex

@@ -13,6 +13,7 @@
 \usepackage{environ}%
 \usepackage[right]{lineno}%
 \usepackage{nameref}
+\usepackage{tikz}
 %\usepackage{showkeys}
 
 % local definitions
@@ -28,7 +29,8 @@
 \newcommand{\phastsim}[0]{\texttt{phastsim}}
 
 \newcommand{\jkcomment}[1]{\textcolor{red}{#1}}
-\newcommand{\plrcomment}[1]{\textcolor{purple}{#1}}
+% @plr - purple doesn't exist in my Latex
+\newcommand{\plrcomment}[1]{\textcolor{blue}{#1}}
 
 \begin{document}
 
@@ -191,7 +193,7 @@ \subsection*{Tree sequences}
 
 \begin{figure}
 \begin{center}
-\includegraphics[width=6cm]{figures/example_tree_sequence}
+\includegraphics[width=6cm]{illustrations/example_tree_sequence}
 \end{center}
 \caption{\label{fig-ts-example} Example tree sequence.
 An example tree sequence describing genealogies and genotypes
@@ -341,7 +343,7 @@ \subsection*{Genetic variation}
 % Mutations are not treated as first-class data --- the sequences
 % were really all people were after, classically.
 However, joint information about the mutations that have been simulated
--- e.g., where they have occurred on the trees -- 
+-- e.g., where they have occurred on the trees --
 is usually not retained or made available for subsequent analysis.
 
 The approach taken in \msprime\ is quite different.
@@ -421,8 +423,8 @@ \subsection*{Simulating mutations}
 virgoulay2021gspace}.
 
 \begin{figure}
-    \includegraphics{figures/unmutated_tree}
-    \includegraphics{figures/mutated_tree}
+    \includegraphics{illustrations/unmutated_tree}
+    \includegraphics{illustrations/mutated_tree}
     \caption{
         \label{fig-mutated-trees}
         Visualization of the separation between ancestry and mutation simulation:
@@ -858,12 +860,52 @@ \subsection*{Ancestral Recombination Graphs}
 have fully coalesced (like Hudson's algorithm).
 
 \begin{figure}
+
 \begin{center}
-\includegraphics[width=7cm]{figures/arg}
+    \begin{tikzpicture}[scale=0.45]
+
+    \draw (-2,0) -- (-2,3) -- (0, 3);
+    \draw (1,0) -- (1,1);
+    \draw (0,1) -- (2,1) -- (2,2) -- (4,2) -- (4,0);
+    \draw (3,2) -- (3,5);
+    \draw (0,1) -- (0,3);
+    \draw (-1, 3) -- (-1, 5) -- (3,5);
+
+    \draw[fill] (-2,0) circle [radius=0.05];
+    \draw[fill] (-1,3) circle [radius=0.05];
+    \draw[fill] (1,0) circle [radius=0.05];
+    \draw[fill] (4,0) circle [radius=0.05];
+    \draw[fill] (1,1) circle [radius=0.05];
+    \draw[fill] (3,2) circle [radius=0.05];
+    \draw[fill] (1,5) circle [radius=0.05];
+
+    \node [below] at (-2,0) {0};
+    \node [below] at (1,0) {1};
+    \node [below] at (4,0) {2};
+    \node [above] at (1,1) {3};
+    \node [below right] at (1,1) {$x = 0.3$};
+    \node [above right] at (3,2) {4};
+    \node [above left] at (-1,3) {5};
+    \node [above] at (1,5) {6};
+
+    \end{tikzpicture}
+% \begin{center}
+% $\equiv$
+% \end{center}
+% \end{column}
+% \begin{column}{5cm}
+\includegraphics[width=3.8cm]{illustrations/arg-ts.pdf}
+\includegraphics[width=3.8cm]{illustrations/arg-ts-simplified.pdf}
+\includegraphics[width=3.8cm]{figures/arg.pdf}
+
 \end{center}
-\caption{\label{fig-arg} A figure illustrating the relative size of
-the full ARG representation compared the minimal tree sequence.
-\jkcomment{This figure is quite preliminary and we need to improve it.}}
+\caption{\label{fig-arg} (A) An simple ARG in which a recombination
+occurs at position 0.3; (B) the equivalent topology depicted as a tree
+sequence, including the recombination node; (C) the same tree sequence
+topology ``simplified'' down to its minimal tree sequence representation;
+(D) the fraction of nodes in a tree sequence that are concerned with
+extra ARG information in a tree sequences returned using \msprime's
+\texttt{record\_full\_arg} mode}
 \end{figure}
 
 However, this original definition of the ARG in terms of a stochastic process

pretty_pictures.py

@@ -1,40 +1,99 @@
-#!/usr/bin/env python
-
+import click
+import tskit
 import msprime
 
 
-## Adding mutations
-# 
-# The goal here is to make a figure with (a) a tree and (b) some mutations added to it.
-
-
-ts = msprime.sim_ancestry(
+@click.command()
+def mutated_tree():
+    """
+    Make a figure with (a) a tree and (b) some mutations added to it.
+    """
+    ts = msprime.sim_ancestry(
         50,
         population_size=1e4,
         recombination_rate=0.5e-8,
         sequence_length=1000,
-        random_seed=14
-)
+        random_seed=14,
+    )
+
+    model = msprime.F84(kappa=2)
+    mts = msprime.sim_mutations(ts, rate=5e-8, model=model, random_seed=5)
 
-model = msprime.F84(kappa=2)
-mts = msprime.sim_mutations(
-        ts,
-        rate=5e-8,
-        model=model,
-        random_seed=5
-)
 
+    def do_svg(ts, **kwargs):
 
-def do_svg(ts, **kwargs):
-    ts.draw_svg(
-            size=(300,200),
+        # Sizes are probably all wrong here
+        style = """\
+            @media print {
+              @page { margin: 0; size: 3in 2in}
+              body { margin: 1.6cm; }
+            }
+        """
+        ts.draw_svg(
+            size=(300, 200),
             node_labels={},
             mutation_labels={m.id: m.derived_state for m in ts.mutations()},
             symbol_size=5,
             force_root_branch=True,
+            style=style,
             **kwargs
-    )
+        )
+
+
+    do_svg(ts, path="illustrations/unmutated_tree.svg")
+    do_svg(mts, path="illustrations/mutated_tree.svg")
+
+
+@click.command()
+def arg_ts():
+    tables = tskit.TableCollection(1.0)
+    tables.nodes.add_row(flags=tskit.NODE_IS_SAMPLE, time=0)
+    tables.nodes.add_row(flags=tskit.NODE_IS_SAMPLE, time=0)
+    tables.nodes.add_row(flags=tskit.NODE_IS_SAMPLE, time=0)
+    tables.nodes.add_row(flags=msprime.NODE_IS_RE_EVENT, time=0.5)
+    # tables.nodes.add_row(flags=msprime.NODE_IS_RE_EVENT, time=0.5)
+    tables.nodes.add_row(flags=0, time=1.0)
+    tables.nodes.add_row(flags=0, time=1.5)
+    tables.nodes.add_row(flags=0, time=2.0)
+
+    tables.edges.add_row(0, 1, 5, 0)
+    tables.edges.add_row(0, 1, 3, 1)
+    tables.edges.add_row(0.0, 0.3, 5, 3)
+    tables.edges.add_row(0.3, 1, 4, 3)
+    tables.edges.add_row(0, 1, 4, 2)
+    tables.edges.add_row(0, 1, 6, 5)
+    tables.edges.add_row(0, 1, 6, 4)
+
+    tables.sort()
+    print(tables)
+    ts = tables.tree_sequence()
+
+    # https://stackoverflow.com/questions/46077392/additional-options-in-chrome-headless-print-to-pdf
+    # Sizes are probably all wrong here
+    style = """\
+        @media print {
+          @page { margin: 0; size: 4in 3in}
+          body { margin: 1.6cm; }
+        }
+    """
+    svgfile = "illustrations/arg-ts.svg"
+    ts.draw_svg(svgfile, size=(400, 300), style=style)
+
+    print(ts.draw_text())
+
+    ts = ts.simplify()
+    svgfile = "illustrations/arg-ts-simplified.svg"
+    ts.draw_svg(svgfile, size=(400, 300), style=style)
+    print(ts.draw_text())
+
+
+@click.group()
+def cli():
+    pass
+
 
-do_svg(ts, path="figures/unmutated_tree.svg")
-do_svg(mts, path="figures/mutated_tree.svg")
+cli.add_command(mutated_tree)
+cli.add_command(arg_ts)
 
+if __name__ == "__main__":
+    cli()