Mercurial > repos > chemteam > biomd_extract_clusters

annotate get_clusters.py @ 1:078dfd7fb26d draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 79589d149a8ff2791d4f71d28b155011672db827"
author chemteam
date Fri, 11 Sep 2020 21:55:13 +0000
parents ad49025ba90d
children e0ecaf2d05fb
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
1 import argparse
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
2 import collections
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
3 import json
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
4
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
5 import numpy as np
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
6
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
7 from scipy.cluster.hierarchy import fcluster
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
8
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
9
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
10 def separate_clusters(Z_fpath, threshold, min_members, output):
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
11 Z = np.loadtxt(Z_fpath)
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
12 branch_assignments = fcluster(Z, threshold, criterion='distance')
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
13 cluster_dict = collections.defaultdict(list)
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
14 for n, val in enumerate(branch_assignments):
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
15 cluster_dict[branch_assignments[n]].append(n)
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
16 cluster_dict = {int(k): v for k, v in cluster_dict.items()
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
17 if len(v) >= min_members}
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
18 with open(output, 'w') as f:
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
19 json.dump(cluster_dict, f, indent=4, sort_keys=True)
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
20
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
21
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
22 def main():
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
23 parser = argparse.ArgumentParser()
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
24 parser.add_argument('--Z', required=True,
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
25 help='File for cluster linkage array.')
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
26 parser.add_argument('--threshold', type=int, required=True,
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
27 help='Distance cutoff.')
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
28 parser.add_argument('--min-members', type=int, required=True,
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
29 help='Minimum number of members of the cluster.')
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
30 parser.add_argument('--output', required=True,
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
31 help='Output file.')
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
32 args = parser.parse_args()
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
33
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
34 separate_clusters(args.Z, args.threshold,
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
35 args.min_members, args.output)
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
36
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
37
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
38 if __name__ == "__main__":
ad49025ba90d "planemo upload for repository https://github.com/galaxycomputationalchemistry/galaxy-tools-compchem/ commit 1b23e024af45cc0999d9142d07de6897d4189ec2"
chemteam
parents:
diff changeset
39 main()