From 648948253f9f14ca0626c3cf5e9001dbc7dbc39a Mon Sep 17 00:00:00 2001
From: Blaise Li <blaise.li__git@nsup.org>
Date: Fri, 29 Apr 2022 17:08:47 +0200
Subject: [PATCH] Option to exclude columns for PCA.
---
libcodonusage/libcodonusage.py | 9 ++++++++-
1 file changed, 8 insertions(+), 1 deletion(-)
diff --git a/libcodonusage/libcodonusage.py b/libcodonusage/libcodonusage.py
index 70da58f..a95f880 100644
--- a/libcodonusage/libcodonusage.py
+++ b/libcodonusage/libcodonusage.py
@@ -667,7 +667,9 @@ methionine (M) and tryptophan (W).
all_nan_cols)
-def codon_usage_pca(usage_data, figs_dir=None, hue="chrom"):
+def codon_usage_pca(
+ usage_data,
+ figs_dir=None, hue="chrom", exclude_cols=None):
"""
Perform Principal Component Analysis on *usage_data*.
@@ -689,10 +691,15 @@ def codon_usage_pca(usage_data, figs_dir=None, hue="chrom"):
in the first four principal components (0 vs. 1 and 2 vs. 3)
as well as graphics representing the influence of each data column
on the first four principal components.
+
+ If *exclude_cols* is not None, the columns with the names contained
+ in the iterable *exclude_cols* will not be included in the PCA analysis.
"""
if figs_dir is not None:
figs_dir = Path(figs_dir)
figs_dir.mkdir(parents=True, exist_ok=True)
+ if exclude_cols is not None:
+ usage_data = usage_data.drop(columns=exclude_cols)
pca = PCA().fit(usage_data)
transformed_data = pd.DataFrame(
pca.transform(usage_data),
--
GitLab