Mercurial > repos > mvdbeek > plot_correlation_matrix

annotate plot_corr.py @ 0:0cee38fb62af draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
author mvdbeek
date Fri, 27 Apr 2018 09:48:24 -0400
parents
children e9268619b503
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
1 import os
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
2 import click
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
3 import matplotlib
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
4 matplotlib.use('tkagg')
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
5
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
6 import pandas as pd # noqa: E402
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
7 import seaborn as sns # noqa: E402
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
8
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
9
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
10 def get_dataframe(files, labels, column, skiprows=1, sep='\t'):
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
11 d = {}
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
12 for file, label in zip(files, labels):
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
13 d[label] = pd.read_csv(file, usecols=[column], sep=sep, skiprows=skiprows, header=None, squeeze=True)
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
14 return pd.DataFrame.from_dict(d)
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
15
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
16
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
17 def plot_correlation(df, plot_path=None, method='pearson', correlation_matrix_path=None, figsize=(12, 12)):
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
18 corr = df.corr(method=method)
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
19 if correlation_matrix_path:
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
20 corr.to_csv(correlation_matrix_path, sep="\t")
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
21 if plot_path:
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
22 g = sns.clustermap(corr, annot=True, method="centroid", figsize=figsize, cbar_kws={'label': "%s correlation" % method})
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
23 g.fig.suptitle("Cluster based on %s correlation for all samples" % method)
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
24 g.savefig(plot_path, bbox_inches='tight')
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
25
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
26
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
27 @click.command()
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
28 @click.argument("files", type=click.Path(exists=True), nargs=-1, required=True)
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
29 @click.option("-c", "--column", help="Use this numeric column to calculate correlation across files", default=1, required=True)
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
30 @click.option("--labels", help="File containing a list of labels, one label per line. Must match number of files", type=click.Path(exists=True), required=False)
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
31 @click.option("--plot_path", help="Write correlation plot to this path", type=click.Path(exists=False), required=False)
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
32 @click.option("--correlation_matrix_path", help="Write correlation plot to this path", type=click.Path(exists=False), required=False)
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
33 @click.option("--method", help="Use this method for calculating the correlation", required=False, type=click.Choice(['pearson', 'spearman', 'kendall']))
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
34 @click.option("--skiprows", help="Skip this number of rows", required=False, default=0)
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
35 @click.option("--sep", help="Use this field separator when reading files", required=False, default="\t")
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
36 def main(files, column, labels=None, method="pearson", skiprows=1, plot_path=None, correlation_matrix_path=None, figsize=(12, 12), sep='\t'):
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
37 """Plot heatmap of pearson correlation and/or write matrix of pearson correlation values."""
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
38 if labels:
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
39 labels = [l.strip() for l in open(labels) if l.strip()]
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
40 assert len(labels) == len(files), "Got %d files for plotting, but %d labels. Label and file length must be equal" % (len(files), len(labels))
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
41 if not labels:
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
42 labels = [os.path.basename(f) for f in files]
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
43 if column != -1:
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
44 # Adjust for 0-based column selection
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
45 column -= 1
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
46 df = get_dataframe(files, labels, column=column, skiprows=skiprows, sep=sep)
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
47 plot_correlation(df, plot_path=plot_path, correlation_matrix_path=correlation_matrix_path, figsize=figsize, method=method)
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
48
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
49
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
50 if __name__ == '__main__':
0cee38fb62af planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/plot_correlation_matrix commit 8451e17775c0ffcd7fa63ef65081f442bef91496
mvdbeek
parents:
diff changeset
51 main()