view @ 12:2912ebf12ef2 draft

"planemo upload for repository commit 1fe240ef0064a1a4a66d9be1ccace53824280b75"
author bgruening
date Mon, 19 Oct 2020 14:46:16 +0000
parents 78640d0127ce
line wrap: on
line source

#!/usr/bin/env python
    Input: set of molecules with pre-calculated physico-chemical properties
    Output: set of molecules that pass all the filters
    Copyright 2012, Bjoern Gruening and Xavier Lucas

    TODO: AND/OR conditions?
import argparse
import json
import shlex
import subprocess
import sys

import cheminfolib
from openbabel import pybel

def parse_command_line():
    parser = argparse.ArgumentParser()
    parser.add_argument('-i', '--input', help='Input file name')
    parser.add_argument('-iformat', help='Input file format')
    parser.add_argument('-oformat', default='smi',
                        help='Output file format')
    parser.add_argument('-o', '--output', help='Output file name',
    parser.add_argument('--filters', help="Specify the filters to apply",
    parser.add_argument('--list_of_names', required=False,
                        help="A file with list of molecule names to extract. Every name is in one line.")
    return parser.parse_args()

def filter_precalculated_compounds(args, filters):
    outfile = pybel.Outputfile(args.oformat, args.output, overwrite=True)
    for mol in pybel.readfile('sdf', args.input):
        for key, elem in filters.items():
            # map the short description to the larger metadata names stored in the sdf file
            property = cheminfolib.ColumnNames.get(key, key)
            min = elem[0]
            max = elem[1]
            if float([property]) >= float(min) and float([property]) <= float(max):
                # leave the filter loop, because one filter constrained are not satisfied
            # if the filter loop terminates in a normal way (no break) all filter rules are satisfied, so save the compound

def filter_new_compounds(args, filters):

    if args.iformat == args.oformat:
        # use the -ocopy option from openbabel to speed up the filtering, additionally no conversion is carried out
        cmd = 'obabel -i%s %s -ocopy -O %s --filter' % (args.iformat, args.input, args.output)
        cmd = 'obabel -i%s %s -o%s -O %s --filter' % (args.iformat, args.input, args.oformat, args.output)
    filter_cmd = ''
    # OBDescriptor stores a mapping from our desc shortcut to the OB name [0] and a long description [1]
    for key, elem in filters.items():
        ob_descriptor_name = cheminfolib.OBDescriptor[key][0]
        min = elem[0]
        max = elem[1]
        filter_cmd += ' %s>=%s %s<=%s ' % (ob_descriptor_name, min, ob_descriptor_name, max)

    args = shlex.split('%s "%s"' % (cmd, filter_cmd))
    # print '%s "%s"' % (cmd, filter_cmd)
    # calling openbabel with subprocess and pipe potential errors occuring in openbabel to stdout
    child = subprocess.Popen(args, stdout=subprocess.PIPE, stderr=subprocess.PIPE)

    stdout, stderr = child.communicate()
    return_code = child.returncode

    if return_code:
        sys.stderr.write("Return error code %i from command:\n" % return_code)
        sys.stderr.write("%s\n" % cmd)

def filter_by_name(args):
    outfile = pybel.Outputfile(args.oformat, args.output, overwrite=True)
    for mol in pybel.readfile('sdf', args.input):
        for name in open(args.list_of_names):
            if mol.title.strip() == name.strip():

def __main__():
        Select compounds with certain properties from a small library
    args = parse_command_line()

    if args.filters == '__filter_by_name__':

    # Its a small trick to get the parameters in an easy way from the xml file.
    # To keep it readable in the xml file, many white-spaces are included in that string it needs to be removed.
    # Also the last loop creates a ',{' that is not an valid jason expression.
    filters = json.loads((args.filters).replace(' ', '').replace(',}', '}'))
    if args.iformat == 'sdf':
        # Check if the sdf file contains all of the required metadata to invoke the precalculation filtering
        mol = next(pybel.readfile('sdf', args.input))
        for key, elem in filters.items():
            property = cheminfolib.ColumnNames.get(key, key)
            if property not in
            # if the for loop finishes in a normal way, we should habe all properties at least in the first molecule
            # assume it is the same for all other molecules and start the precalculated filtering
            filter_precalculated_compounds(args, filters)
            return True
    filter_new_compounds(args, filters)

if __name__ == "__main__":