From cf468c0f0edbdedbe96630d653d6306ad5bf009a Mon Sep 17 00:00:00 2001
From: Blaise Li <blaise.li__git@nsup.org>
Date: Fri, 4 Mar 2022 14:13:18 +0100
Subject: [PATCH] Added function to group counts columns by aa.
---
libcodonusage/__init__.py | 3 ++-
libcodonusage/libcodonusage.py | 30 ++++++++++++++++++++++++++++++
2 files changed, 32 insertions(+), 1 deletion(-)
diff --git a/libcodonusage/__init__.py b/libcodonusage/__init__.py
index 12d8398..6f7ce6d 100644
--- a/libcodonusage/__init__.py
+++ b/libcodonusage/__init__.py
@@ -1,6 +1,6 @@
__copyright__ = "Copyright (C) 2022 Blaise Li"
__licence__ = "GNU GPLv3"
-__version__ = 0.4
+__version__ = 0.5
from .libcodonusage import (
aa2colour,
codon2aa,
@@ -11,6 +11,7 @@ from .libcodonusage import (
make_aa_codon_columns,
make_counts_only,
render_md,
+ sort_counts_by_aa,
violin_usage,
violin_usage_vertical,
violin_usage_by_clusters,
diff --git a/libcodonusage/libcodonusage.py b/libcodonusage/libcodonusage.py
index 04128d8..9bbda1f 100644
--- a/libcodonusage/libcodonusage.py
+++ b/libcodonusage/libcodonusage.py
@@ -72,6 +72,9 @@ columns_by_aa = groupby(itemgetter(0), codon_headers)
######################################
# Associating colours to amino-acids #
######################################
+# We load amino-acid colouring information to use in graphical representations.
+# We'll use colour schemes provided by the biotite package:
+# https://www.biotite-python.org/examples/gallery/sequence/color_schemes_protein.html
# We look directly at the json file
# instead of using the "native" biotite mechanisms
with Path(bgraphs.colorschemes._scheme_dir).joinpath(
@@ -269,9 +272,36 @@ def make_aa_codon_columns(counts_table):
counts_table.columns = pd.MultiIndex.from_tuples(
((codon2aa[codon], codon) for codon in counts_table.columns),
names=("aa", "codon"))
+ render_md(
+ "We associated amino-acid information to codons, "
+ "as an extra level in the table columns.")
return counts_table
+def sort_counts_by_aa(counts_table):
+ """
+ Sort columns so that codons are grouped by corresponding amino-acid.
+ """
+ aacodons_order = list(aa2colour.keys())
+ # We need to include both aas and codons in the list
+ # because it will be used for both the aa and the codon levels
+ # when sorting the (aa, codon) MultiIndex
+ aacodons_order.extend(codon2aa.keys())
+
+ def aa_sortkey(col):
+ """
+ Key function to use when sorting a MultiIndex consisting in
+ (aa, codon) pairs.
+ """
+ return col.map(aacodons_order.index)
+ sorted_counts = counts_table.sort_index(
+ axis=1, level=(0, 1), ascending=(True, True), key=aa_sortkey)
+ render_md(
+ "The columns were sorted in order to group codons"
+ " by associated amino-acid.")
+ return sorted_counts
+
+
def load_bias_table(table_path, nb_cluster_series=2):
"""
Load a table containing by-amino-acid codon usage biases.
--
GitLab