Mercurial > repos > bgruening > chemfp

annotate nxn_clustering.py @ 41:07e1a9d090e4 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit deb7ad38cefd13df312431b1138054a68381efdf"
author bgruening
date Fri, 18 Oct 2019 04:57:12 -0400
parents 28c487eb8399
children 0a1c281e9224
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
33
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
1 #!/usr/bin/env python
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
2 """
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
3 Modified version of code examples from the chemfp project.
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
4 http://code.google.com/p/chem-fingerprints/
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
5 Thanks to Andrew Dalke of Andrew Dalke Scientific!
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
6 """
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
7 import matplotlib
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
8 matplotlib.use('Agg')
39
28c487eb8399 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit e78367b77f2294891914151f642685644d43a5b7"
bgruening
parents: 38
diff changeset
9 from matplotlib import rcParams
28c487eb8399 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit e78367b77f2294891914151f642685644d43a5b7"
bgruening
parents: 38
diff changeset
10 rcParams.update({'figure.autolayout': True})
33
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
11 import argparse
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
12 import os
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
13 import chemfp
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
14 import scipy.cluster.hierarchy as hcluster
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
15 import pylab
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
16 import numpy
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
17
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
18 def distance_matrix(arena, tanimoto_threshold = 0.0):
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
19 n = len(arena)
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
20 # Start off a similarity matrix with 1.0s along the diagonal
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
21 try:
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
22 similarities = numpy.identity(n, "d")
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
23 except:
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
24 raise Exception('Input dataset is to large!')
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
25 chemfp.set_num_threads( args.processors )
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
26
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
27 ## Compute the full similarity matrix.
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
28 # The implementation computes the upper-triangle then copies
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
29 # the upper-triangle into lower-triangle. It does not include
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
30 # terms for the diagonal.
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
31 results = chemfp.search.threshold_tanimoto_search_symmetric(arena, threshold=tanimoto_threshold)
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
32
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
33 # Copy the results into the NumPy array.
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
34 for row_index, row in enumerate(results.iter_indices_and_scores()):
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
35 for target_index, target_score in row:
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
36 similarities[row_index, target_index] = target_score
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
37
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
38 # Return the distance matrix using the similarity matrix
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
39 return 1.0 - similarities
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
40
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
41
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
42 if __name__ == "__main__":
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
43 parser = argparse.ArgumentParser(description="""NxN clustering for fps files.
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
44 For more details please see the chemfp documentation:
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
45 https://chemfp.readthedocs.org
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
46 """)
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
47
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
48 parser.add_argument("-i", "--input", dest="input_path",
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
49 required=True,
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
50 help="Path to the input file.")
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
51
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
52 parser.add_argument("-c", "--cluster", dest="cluster_image",
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
53 help="Path to the output cluster image.")
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
54
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
55 parser.add_argument("-s", "--smatrix", dest="similarity_matrix",
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
56 help="Path to the similarity matrix output file.")
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
57
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
58 parser.add_argument("-t", "--threshold", dest="tanimoto_threshold",
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
59 type=float, default=0.0,
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
60 help="Tanimoto threshold [0.0]")
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
61
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
62 parser.add_argument("--oformat", default='png', help="Output format (png, svg)")
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
63
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
64 parser.add_argument('-p', '--processors', type=int,
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
65 default=4)
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
66
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
67 args = parser.parse_args()
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
68
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
69 targets = chemfp.open( args.input_path, format='fps' )
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
70 arena = chemfp.load_fingerprints( targets )
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
71 distances = distance_matrix( arena, args.tanimoto_threshold )
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
72
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
73 if args.similarity_matrix:
38
02e03ac072cf planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit ed9b6859de648aa5f7cde483732f5df20aaff90e
bgruening
parents: 33
diff changeset
74 numpy.savetxt(args.similarity_matrix, distances)
33
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
75
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
76 if args.cluster_image:
38
02e03ac072cf planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit ed9b6859de648aa5f7cde483732f5df20aaff90e
bgruening
parents: 33
diff changeset
77 linkage = hcluster.linkage(distances, method="single", metric="euclidean")
02e03ac072cf planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit ed9b6859de648aa5f7cde483732f5df20aaff90e
bgruening
parents: 33
diff changeset
78 hcluster.dendrogram(linkage, labels=arena.ids, leaf_rotation=90.)
02e03ac072cf planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit ed9b6859de648aa5f7cde483732f5df20aaff90e
bgruening
parents: 33
diff changeset
79 pylab.savefig(args.cluster_image, format=args.oformat)
33
73b8c87779ae planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/chemfp commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
80