Mercurial > repos > bgruening > openbabel_remduplicates

annotate ob_spectrophore_search.py @ 14:b2569e22b40c 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/openbabel commit 327c29cc43f56d7067ab9fa51323ea31951db98b"
author bgruening
date Tue, 10 Nov 2020 20:30:47 +0000
parents 12aca74f07d7
children c5de6c19eb06
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
1 #!/usr/bin/env python
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
2 """
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
3 Input: tabular format file with one column storing the unique id for the compounds and any other with the Spectrophores(TM) descriptors.
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
4 Output: parse the target file using the same protocol used to generate the databases in our servers. Physico-chemical properties are computed and stored as metadata in the sdf output file.
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
5 Copyright 2012, Bjoern Gruening and Xavier Lucas
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
6 """
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
7 import argparse
13
12aca74f07d7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 1fe240ef0064a1a4a66d9be1ccace53824280b75"
bgruening
parents: 12
diff changeset
8
0
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
9 import numpy as np
12
50ca8845e7f5 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 944ea4bb8a9cd4244152a4a4fecd0485fabc2ad0"
bgruening
parents: 3
diff changeset
10 from openbabel import openbabel, pybel
50ca8845e7f5 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 944ea4bb8a9cd4244152a4a4fecd0485fabc2ad0"
bgruening
parents: 3
diff changeset
11 openbabel.obErrorLog.StopLogging()
13
12aca74f07d7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 1fe240ef0064a1a4a66d9be1ccace53824280b75"
bgruening
parents: 12
diff changeset
12 # TODO get rid of eval()
0
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
13
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
14 global spectrophore
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
15 spectrophore = pybel.ob.OBSpectrophore()
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
16
13
12aca74f07d7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 1fe240ef0064a1a4a66d9be1ccace53824280b75"
bgruening
parents: 12
diff changeset
17
0
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
18 def parse_command_line():
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
19 parser = argparse.ArgumentParser()
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
20 parser.add_argument('--target', required=True, help='target file name in sdf format with Spectrophores(TM) descriptors stored as meta-data')
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
21 parser.add_argument('--library', required=True, help='library of compounds with pre-computed physico-chemical properties, including Spectrophores(TM) in tabular format')
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
22 parser.add_argument('-c', '--column', required=True, type=int, help='#column containing the Spectrophores(TM) descriptors in the library file')
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
23 parser.add_argument('-o', '--output', required=True, help='output file name')
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
24 parser.add_argument('-n', '--normalization', default="ZeroMeanAndUnitStd", choices=['No', 'ZeroMean', 'UnitStd', 'ZeroMeanAndUnitStd'], help='Normalization method')
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
25 parser.add_argument('-a', '--accuracy', default="20", choices=['1', '2', '5', '10', '15', '20', '30', '36', '45', '60'], help='Accuracy expressed as angular stepsize')
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
26 parser.add_argument('-s', '--stereo', default="No", choices=['No', 'Unique', 'Mirror', 'All'], help='Stereospecificity of the cage')
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
27 parser.add_argument('-r', '--resolution', type=float, default="3.0", help='Resolution')
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
28 return parser.parse_args()
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
29
13
12aca74f07d7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 1fe240ef0064a1a4a66d9be1ccace53824280b75"
bgruening
parents: 12
diff changeset
30
0
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
31 def set_parameters(args):
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
32 if args.normalization == 'No':
13
12aca74f07d7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 1fe240ef0064a1a4a66d9be1ccace53824280b75"
bgruening
parents: 12
diff changeset
33 spectrophore.SetNormalization(spectrophore.NoNormalization)
0
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
34 else:
13
12aca74f07d7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 1fe240ef0064a1a4a66d9be1ccace53824280b75"
bgruening
parents: 12
diff changeset
35 spectrophore.SetNormalization(eval('spectrophore.NormalizationTowards' + args.normalization))
12aca74f07d7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 1fe240ef0064a1a4a66d9be1ccace53824280b75"
bgruening
parents: 12
diff changeset
36 spectrophore.SetAccuracy(eval('spectrophore.AngStepSize' + args.accuracy))
12aca74f07d7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 1fe240ef0064a1a4a66d9be1ccace53824280b75"
bgruening
parents: 12
diff changeset
37 spectrophore.SetStereo(eval('spectrophore.' + args.stereo + 'StereoSpecificProbes'))
12aca74f07d7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 1fe240ef0064a1a4a66d9be1ccace53824280b75"
bgruening
parents: 12
diff changeset
38 spectrophore.SetResolution(args.resolution)
0
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
39 return True
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
40
13
12aca74f07d7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 1fe240ef0064a1a4a66d9be1ccace53824280b75"
bgruening
parents: 12
diff changeset
41
0
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
42 def Compute_Spectrophores_distance(target_spectrophore, args):
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
43 outfile = open(args.output, 'w')
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
44 for mol in open(args.library, 'r'):
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
45 try:
13
12aca74f07d7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 1fe240ef0064a1a4a66d9be1ccace53824280b75"
bgruening
parents: 12
diff changeset
46 distance = ((np.asarray(target_spectrophore, dtype=float) - np.asarray(mol.split('\t')[args.column - 1].strip().split(', '), dtype=float))**2).sum()
0
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
47 except ValueError:
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
48 distance = 0
13
12aca74f07d7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 1fe240ef0064a1a4a66d9be1ccace53824280b75"
bgruening
parents: 12
diff changeset
49 outfile.write('%s\t%f\n' % (mol.strip(), distance))
0
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
50 outfile.close()
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
51
13
12aca74f07d7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 1fe240ef0064a1a4a66d9be1ccace53824280b75"
bgruening
parents: 12
diff changeset
52
0
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
53 def __main__():
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
54 """
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
55 Computation of Spectrophores(TM) distances to a target molecule.
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
56 """
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
57 args = parse_command_line()
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
58 # This sets up the parameters for the Spectrophore generation. Parameters are set to fit those of our standard parsing tool
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
59 set_parameters(args)
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
60
3
d9fa62b4fda3 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit c376ab981fafb25f0c04a2a313968fa2e55f5196
bgruening
parents: 0
diff changeset
61 mol = next(pybel.readfile('sdf', args.target))
0
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
62 target_spectrophore = mol.data["Spectrophores(TM)"].strip().split(', ')
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
63 # Compute the paired-distance between every molecule in the library and the target
13
12aca74f07d7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 1fe240ef0064a1a4a66d9be1ccace53824280b75"
bgruening
parents: 12
diff changeset
64 Compute_Spectrophores_distance(target_spectrophore, args)
12aca74f07d7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 1fe240ef0064a1a4a66d9be1ccace53824280b75"
bgruening
parents: 12
diff changeset
65
0
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
66
13
12aca74f07d7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 1fe240ef0064a1a4a66d9be1ccace53824280b75"
bgruening
parents: 12
diff changeset
67 if __name__ == "__main__":
0
75d6c2b7907a planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 01da22e4184a5a6f6a3dd4631a7b9c31d1b6d502
bgruening
parents:
diff changeset
68 __main__()