diff --git a/libcodonusage/__init__.py b/libcodonusage/__init__.py
index 7ea81e160a1c1a2e6eeeaf43aea9cdc034f7e73e..b2de77858389f3505a41be49b8a9dbd90543a154 100644
--- a/libcodonusage/__init__.py
+++ b/libcodonusage/__init__.py
@@ -1,6 +1,6 @@
__copyright__ = "Copyright (C) 2022 Blaise Li"
__licence__ = "GNU GPLv3"
-__version__ = "0.20"
+__version__ = "0.21"
from .libcodonusage import (
aa2colour,
aa_usage,
@@ -26,6 +26,7 @@ from .libcodonusage import (
save_counts_table,
sort_counts_by_aa,
star2stop,
+ to_long_form,
violin_usage,
violin_usage_vertical,
violin_usage_by_clusters,
diff --git a/libcodonusage/libcodonusage.py b/libcodonusage/libcodonusage.py
index fbf886ede2f425e22054723235ba0cf187eeef92..252ea02cbd8c3847c891d0eec08017e0fe91895c 100644
--- a/libcodonusage/libcodonusage.py
+++ b/libcodonusage/libcodonusage.py
@@ -16,7 +16,7 @@
"""Functions used in Jupyter notebooks."""
from itertools import combinations
import json
-from operator import itemgetter
+from operator import attrgetter, itemgetter
from pathlib import Path
# python3 -m pip install cytoolz
from cytoolz import concat, groupby, unique
@@ -705,8 +705,8 @@ def compare_clusterings(
*clust1_template* and *clust2_template*.
"""
crosstabs = {}
- for aa in aa2colour.keys():
- if aa == "M" or aa == "W":
+ for aa in aa2colour.keys(): # pylint: disable=C0103
+ if aa in {"M", "W"}:
continue
clust1_level = clust1_template.format(aa=aa)
clust2_level = clust2_template.format(aa=aa)
@@ -1081,7 +1081,7 @@ def boxplot_usage(usage_table, ylabel, whiskers="1.5 IQR"):
axis.set_ylabel(ylabel)
-def to_long_form(usage_table, ylabel):
+def to_long_form(usage_table, ylabel, others=None):
"""
Transform data in *usage_table* into "long form".
@@ -1096,17 +1096,49 @@ def to_long_form(usage_table, ylabel):
level=list(range(col_nb_levels))).to_frame().reset_index(
level=list(range(
row_nb_levels, row_nb_levels + col_nb_levels)))
+ if others is None:
+ others = []
if col_nb_levels == 1:
long_form.columns = ["aa", ylabel]
+ others = [other for other in others if other != "aa"]
elif col_nb_levels == 2:
long_form.columns = ["aa", "codon", ylabel]
+ others = [other for other in others if other not in {"aa", "codon"}]
else:
raise NotImplementedError(
"Don't know how to deal with column headers "
"with more than 2 levels ('aa' and 'codon')\n")
- return long_form.reset_index(level=[
+ long_form = long_form.reset_index(level=[
lvl_name for lvl_name in usage_table.index.names
if lvl_name.startswith("cluster")])
+ if not set(others).issubset(set(long_form.index.names)):
+ print(set(others))
+ print(set(long_form.index.names))
+ raise ValueError(
+ "All elements in *others* should be in the index "
+ "of *usage_table*")
+ return long_form.assign(**{
+ other: long_form.index.get_level_values(other) for other in others})
+
+
+def variable2order(variable):
+ """
+ Define the order of amino-acids or codons.
+ """
+ if variable == "aa":
+ return list(columns_by_aa.keys())
+ if variable == "codon":
+ return [codon for (_, codon) in concat(columns_by_aa.values())]
+ raise ValueError("variable can only be 'aa' or 'codon'.\n")
+
+
+def format_codon_labels(codons):
+ """
+ Add amino-acid information to codons meant to be used as tick labels.
+ """
+ return [
+ plt.Text(x, y, f"{codon} ({codon2aa[codon]})")
+ for (x, y, codon) in map(attrgetter("_x", "_y", "_text"), codons)]
def violin_usage(usage_table, variable, ylabel, hue="aa", axis=None):
@@ -1115,7 +1147,7 @@ def violin_usage(usage_table, variable, ylabel, hue="aa", axis=None):
*variable* should be either "codon" or "aa".
"""
- long_form = to_long_form(usage_table, ylabel)
+ long_form = to_long_form(usage_table, ylabel, others=[hue])
if axis is None:
_, axis = plt.subplots(figsize=(18, 6))
do_legend = True
@@ -1125,18 +1157,22 @@ def violin_usage(usage_table, variable, ylabel, hue="aa", axis=None):
palette = aa2colour
else:
palette = None
- if variable == "aa":
- order = [aa for aa in columns_by_aa.keys()]
- elif variable == "codon":
- order = [codon for (_, codon) in concat(columns_by_aa.values())]
- else:
- raise ValueError(f"variable can only be 'aa' or 'codon'.\n")
- sns.violinplot(x=variable, y=ylabel, order=order,
+ # if variable == "aa":
+ # order = list(columns_by_aa.keys())
+ # elif variable == "codon":
+ # order = [codon for (_, codon) in concat(columns_by_aa.values())]
+ # else:
+ # raise ValueError("variable can only be 'aa' or 'codon'.\n")
+ sns.violinplot(x=variable, y=ylabel, order=variable2order(variable),
hue=hue, palette=palette, dodge=False,
data=long_form, ax=axis, orient="v", scale="count")
if do_legend:
plt.legend(bbox_to_anchor=(1.01, 1), borderaxespad=0)
- axis.set_xticklabels(axis.get_xticklabels(), rotation=90)
+ if variable == "codon":
+ ticklabels = format_codon_labels(axis.get_xticklabels())
+ else:
+ ticklabels = axis.get_xticklabels()
+ axis.set_xticklabels(ticklabels, rotation=90)
return axis
@@ -1146,23 +1182,28 @@ def violin_usage_vertical(usage_table, variable, ylabel, hue="aa", axis=None):
*variable* should be either "codon" or "aa".
"""
- long_form = to_long_form(usage_table, ylabel)
+ long_form = to_long_form(usage_table, ylabel, others=[hue])
if axis is None:
_, axis = plt.subplots(figsize=(6, 44))
if hue == "aa":
palette = aa2colour
else:
palette = None
- if variable == "aa":
- order = [aa for aa in columns_by_aa.keys()]
- elif variable == "codon":
- order = [codon for (_, codon) in concat(columns_by_aa.values())]
- else:
- raise ValueError(f"variable can only be 'aa' or 'codon'.\n")
+ # if variable == "aa":
+ # order = list(columns_by_aa.keys())
+ # elif variable == "codon":
+ # order = [codon for (_, codon) in concat(columns_by_aa.values())]
+ # else:
+ # raise ValueError("variable can only be 'aa' or 'codon'.\n")
sns.violinplot(
- y=variable, x=ylabel, order=order,
+ y=variable, x=ylabel, order=variable2order(variable),
hue=hue, palette=palette, dodge=False,
data=long_form, ax=axis, orient="h", scale="count")
+ if variable == "codon":
+ ticklabels = format_codon_labels(axis.get_yticklabels())
+ else:
+ ticklabels = axis.get_xticklabels()
+ axis.set_yticklabels(ticklabels)
return axis