Mercurial > repos > bgruening > openbabel_structure_distance_finder

annotate cheminfolib.py @ 0:c066b5accacf 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 6c84abdd07f292048bf2194073e2e938e94158c4"
author bgruening
date Wed, 25 Mar 2020 16:47:13 -0400
parents
children 4c9d6b47045c
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
1 #!/usr/bin/env python
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
2 """
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
3 Small library with cheminformatic functions based on openbabel and pgchem.
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
4 Copyright 2012, Bjoern Gruening and Xavier Lucas
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
5 """
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
6
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
7 import os, sys
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
8
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
9 try:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
10 from galaxy import eggs
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
11 eggs.require('psycopg2')
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
12 except:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
13 print('psycopg2 is not available. It is currently used in the pgchem wrappers, that are not shipped with default CTB')
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
14
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
15 try:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
16 import pybel
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
17 import openbabel
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
18 except:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
19 print('OpenBabel could not be found. A few functions are not available without OpenBabel.')
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
20
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
21 from multiprocessing import Pool
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
22 import glob, tempfile, re
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
23 import subprocess
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
24
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
25 def CountLines( path ):
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
26 out = subprocess.Popen(['wc', '-l', path],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
27 stdout=subprocess.PIPE,
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
28 stderr=subprocess.STDOUT
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
29 ).communicate()[0]
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
30 return int(out.partition(b' ')[0])
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
31
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
32 def grep(pattern, file_obj):
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
33 grepper = re.compile(pattern)
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
34 for line in file_obj:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
35 if grepper.search(line):
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
36 return True
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
37 return False
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
38
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
39 def check_filetype(filepath):
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
40 mol = False
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
41 possible_inchi = True
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
42 for line_counter, line in enumerate(open(filepath)):
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
43 if line_counter > 10000:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
44 break
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
45 if line.find('$$$$') != -1:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
46 return 'sdf'
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
47 elif line.find('@<TRIPOS>MOLECULE') != -1:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
48 return 'mol2'
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
49 elif line.find('ligand id') != -1:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
50 return 'drf'
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
51 elif possible_inchi and re.findall('^InChI=', line):
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
52 return 'inchi'
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
53 elif re.findall('^M\s+END', line):
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
54 mol = True
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
55 # first line is not an InChI, so it can't be an InChI file
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
56 possible_inchi = False
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
57
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
58 if mol:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
59 # END can occures before $$$$, so and SDF file will
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
60 # be recognised as mol, if you not using this hack'
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
61 return 'mol'
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
62 return 'smi'
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
63
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
64 def db_connect(args):
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
65 try:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
66 db_conn = psycopg2.connect("dbname=%s user=%s host=%s password=%s" % (args.dbname, args.dbuser, args.dbhost, args.dbpasswd));
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
67 return db_conn
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
68 except:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
69 sys.exit('Unable to connect to the db')
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
70
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
71 ColumnNames = {
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
72 'can_smiles' : 'Canonical SMILES',
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
73 'can' : 'Canonical SMILES',
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
74 'inchi' : 'InChI',
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
75 'inchi_key' : 'InChI key',
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
76 'inchi_key_first' : 'InChI key first',
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
77 'inchi_key_last' : 'InChI key last',
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
78 'molwt' : 'Molecular weight',
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
79 'hbd' : 'Hydrogen-bond donors',
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
80 'donors' : 'Hydrogen-bond donors',
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
81 'hba' : 'Hydrogen-bond acceptors',
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
82 'acceptors' : 'Hydrogen-bond acceptors',
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
83 'rotbonds' : 'Rotatable bonds',
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
84 'logp' : 'logP',
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
85 'psa' : 'Polar surface area',
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
86 'mr' : 'Molecular refractivity',
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
87 'atoms' : 'Number of heavy atoms',
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
88 'rings' : 'Number of rings',
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
89 'set_bits' : 'FP2 bits',
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
90 'id' : 'Internal identifier',
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
91 'tani' : 'Tanimoto coefficient',
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
92 'spectrophore' : 'Spectrophores(TM)',
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
93 'dist_spectrophore' : 'Spectrophores(TM) distance to target',
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
94 'synonym' : 'Entry id',
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
95 }
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
96
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
97 OBDescriptor = {
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
98 'atoms': ["atoms","Number of atoms"],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
99 'hatoms': ["hatoms","Number of heavy atoms"], # self defined tag hatoms in plugindefines.txt
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
100 'can_smiles' : ["cansmi","Canonical SMILES"],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
101 'can_smilesNS' : ["cansmiNS","Canonical SMILES without isotopes or stereo"],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
102 #["abonds","Number of aromatic bonds"],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
103 #["bonds","Number of bonds"],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
104 #["dbonds","Number of double bonds"],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
105 #["formula","Chemical formula"],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
106 'hba': ["HBA1","Number of Hydrogen Bond Acceptors 1 (JoelLib)"],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
107 'hba2': ["HBA2","Number of Hydrogen Bond Acceptors 2 (JoelLib)"],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
108 'hbd': ["HBD","Number of Hydrogen Bond Donors (JoelLib)"],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
109 'inchi': ["InChI","IUPAC InChI identifier"],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
110 'inchi_key': ["InChIKey","InChIKey"],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
111 #["L5","Lipinski Rule of Five"],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
112 'logp': ["logP","octanol/water partition coefficient"],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
113 'mr': ["MR","molar refractivity"],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
114 'molwt': ["MW","Molecular Weight filter"],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
115 #["nF","Number of Fluorine Atoms"],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
116 #["s","SMARTS filter"],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
117 #["sbonds","Number of single bonds"],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
118 #["smarts","SMARTS filter"],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
119 #["tbonds","Number of triple bonds"],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
120 #["title","For comparing a molecule's title"],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
121 'psa': ["TPSA","topological polar surface area"],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
122 'rotbonds' : ['ROTATABLE_BOND', 'rotatable bonds'],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
123 }
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
124
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
125
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
126 def print_output(args, rows):
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
127 if args.oformat == 'table':
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
128 outfile = open(args.output, 'w')
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
129 requested_fields = (filter(lambda x: x not in ["[", "]", "'"], args.fetch)).split(', ')
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
130 if args.header:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
131 outfile.write( 'Identifier\t' + '\t'.join( [ColumnNames[key] for key in requested_fields] ) + '\n' )
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
132 for row in rows:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
133 outfile.write( row['synonym'] + '\t' + '\t'.join( [str(row[key]) for key in requested_fields] ) + '\n' )
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
134
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
135 elif args.oformat in ['sdf', 'mol2']:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
136 outfile = pybel.Outputfile(args.oformat, args.output, overwrite=True)
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
137 for row in rows:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
138 try:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
139 mol = pybel.readstring('sdf', row['mol'])
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
140 if args.oformat == 'sdf':
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
141 keys = filter(lambda x: x not in ["[", "]", "'"], args.fetch).split(', ')
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
142 mol.data.update( { ColumnNames['synonym'] : row['synonym'] } )
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
143 if 'inchi_key' in keys:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
144 keys = (', '.join(keys).replace( "inchi_key", "inchi_key_first, inchi_key_last" )).split(', ')
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
145 [ mol.data.update( { ColumnNames[key] : row[key] } ) for key in keys if key]
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
146 outfile.write(mol)
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
147 except:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
148 pass
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
149 else:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
150 outfile = open(args.output, 'w')
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
151 outfile.write( '\n'.join( [ '%s\t%s' % (row[args.oformat], row['synonym'] ) for row in rows ] ) )
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
152 outfile.close()
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
153
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
154 def pybel_stop_logging():
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
155 openbabel.obErrorLog.StopLogging()
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
156
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
157 def get_properties_ext(mol):
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
158
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
159 HBD = pybel.Smarts("[!#6;!H0]")
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
160 HBA = pybel.Smarts("[$([$([#8,#16]);!$(*=N~O);" +
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
161 "!$(*~N=O);X1,X2]),$([#7;v3;" +
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
162 "!$([nH]);!$(*(-a)-a)])]"
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
163 )
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
164 calc_desc_dict = mol.calcdesc()
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
165
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
166 try:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
167 logp = calc_desc_dict['logP']
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
168 except:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
169 logp = calc_desc_dict['LogP']
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
170
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
171 return {"molwt": mol.molwt,
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
172 "logp": logp,
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
173 "donors": len(HBD.findall(mol)),
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
174 "acceptors": len(HBA.findall(mol)),
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
175 "psa": calc_desc_dict['TPSA'],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
176 "mr": calc_desc_dict['MR'],
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
177 "rotbonds": mol.OBMol.NumRotors(),
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
178 "can": mol.write("can").split()[0].strip(), ### tthis one works fine for both zinc and chembl (no ZINC code added after can descriptor string)
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
179 "inchi": mol.write("inchi").strip(),
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
180 "inchi_key": get_inchikey(mol).strip(),
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
181 "rings": len(mol.sssr),
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
182 "atoms": mol.OBMol.NumHvyAtoms(),
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
183 "spectrophore" : OBspectrophore(mol),
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
184 }
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
185
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
186 def get_inchikey(mol):
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
187 conv = openbabel.OBConversion()
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
188 conv.SetInAndOutFormats("mol", "inchi")
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
189 conv.SetOptions("K", conv.OUTOPTIONS)
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
190 inchikey = conv.WriteString( mol.OBMol )
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
191 return inchikey
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
192
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
193 def OBspectrophore(mol):
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
194 spectrophore = pybel.ob.OBSpectrophore()
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
195 # Parameters: rotation angle = 20, normalization for mean and sd, accuracy = 3.0 A and non-stereospecific cages.
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
196 spectrophore.SetNormalization( spectrophore.NormalizationTowardsZeroMeanAndUnitStd )
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
197 return ', '.join( [ "%.3f" % value for value in spectrophore.GetSpectrophore( mol.OBMol ) ] )
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
198
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
199 def squared_euclidean_distance(a, b):
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
200 try:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
201 return ((np.asarray( a ) - np.asarray( b ))**2).sum()
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
202 except ValueError:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
203 return 0
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
204
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
205 def split_library( lib_path, lib_format = 'sdf', package_size = None ):
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
206 """
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
207 Split a library of compounds. Usage: split_library( lib_path, lib_format, package_size )
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
208 IT currently ONLY WORKS FOR SD-Files
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
209 """
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
210 pack = 1
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
211 mol_counter = 0
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
212
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
213 outfile = open('/%s/%s_pack_%i.%s' % ( '/'.join(lib_path.split('/')[:-1]), lib_path.split('/')[-1].split('.')[0], pack, 'sdf'), 'w' )
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
214
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
215 for line in open(lib_path, 'r'):
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
216 outfile.write( line )
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
217 if line.strip() == '$$$$':
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
218 mol_counter += 1
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
219 if mol_counter % package_size == 0:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
220 outfile.close()
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
221 pack += 1
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
222 outfile = open('/%s/%s_pack_%i.%s' % ( '/'.join(lib_path.split('/')[:-1]), lib_path.split('/')[-1].split('.')[0], pack, 'sdf'), 'w' )
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
223 if mol_counter*10 % package_size == 0:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
224 print('%i molecules parsed, starting pack nr. %i' % ( mol_counter, pack - 1 ))
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
225 outfile.close()
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
226
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
227 return True
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
228
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
229 def split_smi_library( smiles_file, structures_in_one_file ):
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
230 """
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
231 Split a file with SMILES to several files for multiprocessing usage.
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
232 Usage: split_smi_library( smiles_file, 10 )
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
233 """
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
234 output_files = []
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
235 tfile = tempfile.NamedTemporaryFile(delete=False)
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
236
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
237 smiles_handle = open(smiles_file, 'r')
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
238 for count, line in enumerate( smiles_handle ):
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
239 if count % structures_in_one_file == 0 and count != 0:
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
240 tfile.close()
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
241 output_files.append(tfile.name)
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
242 tfile = tempfile.NamedTemporaryFile(delete=False)
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
243 tfile.write(line)
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
244 tfile.close()
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
245 output_files.append(tfile.name)
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
246 smiles_handle.close()
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
247 return output_files
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
248
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
249
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
250 def mp_run(input_path, regex, PROCESSES, function_to_call ):
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
251 paths = []
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
252 [ paths.append(compound_file) for compound_file in glob.glob(str(input_path) + str(regex)) ]
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
253 paths.sort()
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
254
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
255 pool = Pool(processes=PROCESSES)
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
256 print('Process initialized with', PROCESSES, 'processors')
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
257 result = pool.map_async(function_to_call, paths)
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
258 result.get()
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
259
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
260 return paths
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
261
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
262 if __name__ == '__main__':
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
263 print(check_filetype(sys.argv[1]))
c066b5accacf "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/chemicaltoolbox/openbabel commit 6c84abdd07f292048bf2194073e2e938e94158c4"
bgruening
parents:
diff changeset
264