Mercurial > repos > bgruening > sdf_to_tab
annotate rdkit_descriptors.py @ 4:67b9a5b22bf0 draft
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 944ea4bb8a9cd4244152a4a4fecd0485fabc2ad0"
author | bgruening |
---|---|
date | Tue, 28 Jul 2020 12:05:30 +0000 |
parents | d58dbfb01a29 |
children | 48c536413a2f |
rev | line source |
---|---|
0
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
1 #!/usr/bin/env python |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
2 |
d58dbfb01a29
"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 |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
4 from rdkit import Chem |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
5 import sys, os, re |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
6 import argparse |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
7 import inspect |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
8 |
d58dbfb01a29
"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' ): |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
10 """ |
d58dbfb01a29
"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 |
d58dbfb01a29
"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. |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
13 """ |
d58dbfb01a29
"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: |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
15 for line in handle: |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
16 line = line.strip() |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
17 if format == 'smiles': |
d58dbfb01a29
"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 ) |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
19 elif format == 'inchi': |
d58dbfb01a29
"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 ) |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
21 if mol is None: |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
22 yield False |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
23 else: |
d58dbfb01a29
"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] ) |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
25 yield mol |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
26 |
d58dbfb01a29
"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(): |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
28 """ |
d58dbfb01a29
"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) |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
30 """ |
d58dbfb01a29
"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( '_' ) ] |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
32 ret.sort() |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
33 return ret |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
34 |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
35 |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
36 def descriptors( mol, functions ): |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
37 """ |
d58dbfb01a29
"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. |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
39 """ |
d58dbfb01a29
"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: |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
41 yield (name, function( mol )) |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
42 |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
43 |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
44 if __name__ == "__main__": |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
45 parser = argparse.ArgumentParser() |
d58dbfb01a29
"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.') |
d58dbfb01a29
"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.") |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
48 |
d58dbfb01a29
"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+'), |
d58dbfb01a29
"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") |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
51 |
d58dbfb01a29
"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", |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
53 default=False, |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
54 help="Write header line.") |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
55 |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
56 args = parser.parse_args() |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
57 |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
58 if args.iformat == 'sdf': |
d58dbfb01a29
"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 ) |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
60 elif args.iformat =='smi': |
d58dbfb01a29
"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' ) |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
62 elif args.iformat == 'inchi': |
d58dbfb01a29
"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' ) |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
64 |
d58dbfb01a29
"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() |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
66 |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
67 if args.header: |
d58dbfb01a29
"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] ) ) |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
69 |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
70 for mol in supplier: |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
71 if not mol: |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
72 continue |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
73 descs = descriptors( mol, functions ) |
d58dbfb01a29
"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") |
d58dbfb01a29
"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] ) ) |
d58dbfb01a29
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/rdkit commit 714e984db6ba1198cacf4dcf325320a5889fa02c"
bgruening
parents:
diff
changeset
|
76 |