Mercurial > repos > bgruening > sdf_to_tab

annotate rdkit_descriptors.py @ 0:06828e0cc8a7 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/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
author bgruening
date Wed, 16 Oct 2019 07:26:19 -0400
parents
children 351fbd750a6d
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
1 #!/usr/bin/env python
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
2
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
3 from rdkit.Chem import Descriptors
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
4 from rdkit import Chem
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
5 import sys, os, re
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
6 import argparse
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
7 import inspect
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
8
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
9 def get_supplier( infile, format = 'smiles' ):
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
10 """
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
11 Returns a generator over a SMILES or InChI file. Every element is of RDKit
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
12 molecule and has its original string as _Name property.
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
13 """
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
14 with open(infile) as handle:
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
15 for line in handle:
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
16 line = line.strip()
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
17 if format == 'smiles':
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
18 mol = Chem.MolFromSmiles( line, sanitize=True )
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
19 elif format == 'inchi':
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
20 mol = Chem.inchi.MolFromInchi( line, sanitize=True, removeHs=True, logLevel=None, treatWarningAsError=False )
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
21 if mol is None:
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
22 yield False
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
23 else:
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
24 mol.SetProp( '_Name', line.split('\t')[0] )
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
25 yield mol
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
26
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
27 def get_rdkit_descriptor_functions():
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
28 """
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
29 Returns all descriptor functions under the Chem.Descriptors Module as tuple of (name, function)
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
30 """
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
31 ret = [ (name, f) for name, f in inspect.getmembers( Descriptors ) if inspect.isfunction( f ) and not name.startswith( '_' ) ]
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
32 ret.sort()
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
33 return ret
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
34
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
35
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
36 def descriptors( mol, functions ):
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
37 """
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
38 Calculates the descriptors of a given molecule.
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
39 """
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
40 for name, function in functions:
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
41 yield (name, function( mol ))
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
42
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
43
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
44 if __name__ == "__main__":
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
45 parser = argparse.ArgumentParser()
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
46 parser.add_argument('-i', '--infile', required=True, help='Path to the input file.')
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
47 parser.add_argument("--iformat", help="Specify the input file format.")
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
48
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
49 parser.add_argument('-o', '--outfile', type=argparse.FileType('w+'),
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
50 default=sys.stdout, help="path to the result file, default it sdtout")
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
51
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
52 parser.add_argument("--header", dest="header", action="store_true",
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
53 default=False,
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
54 help="Write header line.")
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
55
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
56 args = parser.parse_args()
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
57
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
58 if args.iformat == 'sdf':
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
59 supplier = Chem.SDMolSupplier( args.infile )
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
60 elif args.iformat =='smi':
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
61 supplier = get_supplier( args.infile, format = 'smiles' )
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
62 elif args.iformat == 'inchi':
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
63 supplier = get_supplier( args.infile, format = 'inchi' )
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
64
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
65 functions = get_rdkit_descriptor_functions()
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
66
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
67 if args.header:
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
68 args.outfile.write( '%s\n' % '\t'.join( ['MoleculeID'] + [name for name, f in functions] ) )
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
69
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
70 for mol in supplier:
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
71 if not mol:
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
72 continue
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
73 descs = descriptors( mol, functions )
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
74 molecule_id = mol.GetProp("_Name")
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
75 args.outfile.write( "%s\n" % '\t'.join( [molecule_id]+ [str(round(res, 6)) for name, res in descs] ) )
06828e0cc8a7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff changeset
76