Figure size tweaks, violins by chromosome.

libcodonusage/__init__.py

@@ -30,6 +30,7 @@ from .libcodonusage import (
     violin_usage,
     violin_usage_vertical,
     violin_usage_by_clusters,
+    violin_usage_by_clusters_splitby,
     violin_with_thresh,
     write_cluster_lists,
     )

libcodonusage/libcodonusage.py

@@ -1154,9 +1154,17 @@ def violin_usage(
 
     *variable* should be either "codon" or "aa".
     """
+    if hue in {"aa", "codon"}:
+        dodge = False
+        nb_violins = 1
+    else:
+        dodge = True
+        # dodge = True implies there will be one violin
+        # per possible value of *hue*, side by side
+        nb_violins = len(usage_table.index.get_level_values(hue).unique())
     long_form = to_long_form(usage_table, ylabel, others=[hue])
     if axis is None:
-        _, axis = plt.subplots(figsize=(18, 6))
+        _, axis = plt.subplots(figsize=(18 * nb_violins, 6))
         do_legend = True
     else:
         do_legend = False
@@ -1164,10 +1172,6 @@ def violin_usage(
         palette = aa2colour
     else:
         palette = None
-    if hue in {"aa", "codon"}:
-        dodge = False
-    else:
-        dodge = True
     kwargs = {
         "x": variable, "y": ylabel, "order": variable2order(variable),
         "hue": hue, "palette": palette, "dodge": dodge,
@@ -1196,17 +1200,21 @@ def violin_usage_vertical(
 
     *variable* should be either "codon" or "aa".
     """
+    if hue in {"aa", "codon"}:
+        dodge = False
+        nb_violins = 1
+    else:
+        dodge = True
+        # dodge = True implies there will be one violin
+        # per possible value of *hue*, side by side
+        nb_violins = len(usage_table.index.get_level_values(hue).unique())
     long_form = to_long_form(usage_table, ylabel, others=[hue])
     if axis is None:
-        _, axis = plt.subplots(figsize=(6, 44))
+        _, axis = plt.subplots(figsize=(6, 44 * nb_violins))
     if hue == "aa":
         palette = aa2colour
     else:
         palette = None
-    if hue in {"aa", "codon"}:
-        dodge = False
-    else:
-        dodge = True
     kwargs = {
         "y": variable, "x": ylabel,
         "order": variable2order(variable),
@@ -1235,7 +1243,7 @@ def violin_usage_by_clusters(usage_with_clusters, aa,  # pylint: disable=C0103
     Plot a series of violin plots for each cluster of genes.
 
     The clusters are defined in usage table *usage_with_clusters*
-    based on usage biases for the codons codong for amino-acid *aa*.
+    based on usage biases for the codons coding for amino-acid *aa*.
 
     *usage_with_clusters* should have clustering indications as
     an index level named according to *cluster_level_template*
@@ -1243,14 +1251,25 @@ def violin_usage_by_clusters(usage_with_clusters, aa,  # pylint: disable=C0103
     """
     clusters = usage_with_clusters.groupby(
         level=cluster_level_template.format(aa=aa))
+    # To adjust figure size
+    if "hue" in violin_kwargs and violin_kwargs["hue"] not in {"aa", "codon"}:
+        # dodge will be set to True in later plotting call.
+        # This implies that there will be one violin
+        # per possible value of *hue*, side by side
+        nb_violins = len(usage_with_clusters.index.get_level_values(
+            violin_kwargs["hue"]).unique())
+    else:
+        nb_violins = 1
     if vertical:
-        _, axes = plt.subplots(
+        fig, axes = plt.subplots(
             ncols=clusters.ngroups,
-            figsize=(6 * clusters.ngroups, 44))
+            constrained_layout=True,
+            figsize=(6 * clusters.ngroups, 44 * nb_violins))
     else:
-        _, axes = plt.subplots(
+        fig, axes = plt.subplots(
             nrows=clusters.ngroups,
-            figsize=(18, 6 * clusters.ngroups))
+            constrained_layout=True,
+            figsize=(18 * nb_violins, 6 * clusters.ngroups))
     for ((cluster, usage_table), axis) in zip(clusters, axes):
         kwargs = {"axis": axis}
         kwargs.update(violin_kwargs)
@@ -1264,6 +1283,38 @@ def violin_usage_by_clusters(usage_with_clusters, aa,  # pylint: disable=C0103
                 usage_table, "codon",
                 ylabel_template.format(aa=aa, cluster=cluster),
                 **kwargs)
+    return fig
+
+
+def violin_usage_by_clusters_splitby(
+        usage_with_clusters, aa,  # pylint: disable=C0103
+        ylabel_template,
+        cluster_level_template="cluster_{aa}",
+        vertical=False, by_lvl="chrom",
+        **violin_kwargs):
+    """
+    Plot a series of violin plots for each cluster of genes,
+    splitting the figures according to groups defined by the
+    content of the index level *by_lvl* of *usage_with_clusters*.
+
+    The clusters are defined in usage table *usage_with_clusters*
+    based on usage biases for the codons coding for amino-acid *aa*.
+
+    *usage_with_clusters* should have clustering indications as
+    an index level named according to *cluster_level_template*
+    (default "cluster_{aa}"), where {aa} is to be replaced by *aa*.
+    """
+    classes = sorted(set(
+        usage_with_clusters.index.get_level_values(by_lvl)))
+    idx_level = usage_with_clusters.index.names.index(by_lvl)
+    for clss in classes:
+        fig = violin_usage_by_clusters(
+            usage_with_clusters.loc[
+                (*(slice(None) for _ in range(idx_level)), clss),],
+            aa, ylabel_template + f" for {clss}",
+            cluster_level_template=cluster_level_template,
+            vertical=vertical, **violin_kwargs)
+        fig.suptitle(f"Violin plots for {clss}")
 
 
 # Based on