def enumerateMol(mol, fragment):
"""
Enumerate a single molecule
:param mol:
:param fragment The fragmentation method, 'hac' or 'mw'. If not specified the whole molecules is passed to Dimorphite
:return:
"""
if fragment:
mol = mol_utils.fragment(mol, fragment)
inputmol = []
inputmol.append(mol)
protonated_mols = run_with_mol_list(inputmol)
return protonated_mols
def standardize(mol, neutralize, fragment):
"""
:param mol: The molecule to standardize
:param neutralize: Boolean for whether to neutralize the molecule
:param fragment: The approach for choosing the largest fragment. Either 'hac' or 'mw'. If not specified the whole
molecule is used.
:return: The standardized molecule
"""
mol = rdMolStandardize.Cleanup(mol)
#mol = lfc.choose(mol)
# We use our own largest fragment picker as the RDKit one behaves slightly differently
if fragment:
mol = mol_utils.fragment(mol, fragment)
if neutralize:
mol = uncharger.uncharge(mol)
return mol
def main():
### command line args defintions #########################################
parser = argparse.ArgumentParser(description='RDKit screen')
group = parser.add_mutually_exclusive_group()
group.add_argument(
'--qsmiles',
help=
'filename of query structures as smiles (incompatible with --sdf and --qjson args)'
)
group.add_argument(
'--qsdf',
help=
'filename of query structures as sdfile (incompatible with --smiles and --qjson args)'
)
group.add_argument(
'--qjson',
help=
'filename of query structures as MoleculeObject JSON (incompatible with --qsmiles and --qsdf args)'
)
parser.add_argument('--qsmilesTitleLine',
action='store_true',
help='the smiles file has a title line')
parser.add_argument('--qsmilesDelimiter',
default='\t',
help='delimiter for smiles file (default is tab)')
parser.add_argument(
'--qsmilesColumn',
type=int,
default=0,
help='column in smiles file with the smiles (default is first column)')
parser.add_argument(
'--qsmilesNameColumn',
type=int,
default=1,
help='column in smiles file with ID (default is second column)')
parser.add_argument(
'--qprop',
help=
'property name in query molecules to report. If not defined (or property is not present) '
+
'then name property is not written. JSON format uses the UUID as default'
)
parser.add_argument('--simmin',
type=float,
default=0.7,
help='similarity lower cutoff (1.0 means identical)')
parser.add_argument('--simmax',
type=float,
default=1.0,
help='similarity upper cutoff (1.0 means identical)')
parser.add_argument('-d',
'--descriptor',
type=str.lower,
choices=list(descriptors.keys()),
default='rdkit',
help='descriptor or fingerprint type (default rdkit)')
parser.add_argument('-m',
'--metric',
type=str.lower,
choices=list(metrics.keys()),
default='tanimoto',
help='similarity metric (default tanimoto)')
parser.add_argument(
'-f',
'--fragment',
choices=['hac', 'mw'],
help=
'Find single fragment if more than one (hac = biggest by heavy atom count, mw = biggest by mol weight )'
)
parser.add_argument('--hacmin', type=int, help='Min heavy atom count')
parser.add_argument('--hacmax', type=int, help='Max heavy atom count')
parser.add_argument('--mwmin', type=float, help='Min mol weight')
parser.add_argument('--mwmax', type=float, help='Max mol weight')
parameter_utils.add_default_io_args(parser)
parser.add_argument('--thin', action='store_true', help='Thin output mode')
parser.add_argument('-q',
'--quiet',
action='store_true',
help='Quiet mode')
args = parser.parse_args()
utils.log("Screen Args: ", args)
descriptor = descriptors[args.descriptor.lower()]
metric = metrics[args.metric.lower()]
propName = args.qprop
if args.qsmiles:
queryMolsupplier = rdkit_utils.default_open_input_smiles(
args.qsmiles,
delimiter=args.qsmilesDelimiter,
smilesColumn=args.qsmilesColumn,
nameColumn=args.qsmilesNameColumn,
titleLine=args.qsmilesTitleLine)
queryInput = None
elif args.qsdf:
queryInput, queryMolsupplier = rdkit_utils.default_open_input_sdf(
args.qsdf)
elif args.qjson:
queryInput, queryMolsupplier = rdkit_utils.default_open_input_json(
args.qjson, lazy=False)
if not propName:
propName = "uuid"
else:
raise ValueError('No query structure specified')
queryFps = {}
utils.log("Preparing query fingerprints")
count = 0
for q in queryMolsupplier:
count += 1
if q:
queryFps[q] = descriptor(q)
else:
utils.log("WARNING: Failed to parse Molecule", count)
if queryInput:
queryInput.close()
input, output, suppl, writer, output_base = rdkit_utils.default_open_input_output(
args.input, args.informat, args.output, 'screen_multi', args.outformat)
# OK, all looks good so we can hope that things will run OK.
# But before we start lets write the metadata so that the results can be handled.
#if args.meta:
# t = open(output_base + '_types.txt', 'w')
# t.write(field_Similarity + '=integer\n')
# t.flush()
# t.close()
i = 0
count = 0
for mol in suppl:
i += 1
if mol is None: continue
if args.fragment:
mol = mol_utils.fragment(mol, args.fragment, quiet=args.quiet)
if not filter.filter(mol,
minHac=args.hacmin,
maxHac=args.hacmax,
minMw=args.mwmin,
maxMw=args.mwmax,
quiet=args.quiet):
continue
targetFp = descriptor(mol)
idx = 0
hits = 0
bestScore = 0
bestName = None
for queryMol in queryFps:
idx += 1
sim = metric(queryFps[queryMol], targetFp)
if propName:
name = str(queryMol.GetProp(propName))
else:
name = None
if sim >= args.simmin and sim <= args.simmax:
hits += 1
if not args.quiet:
utils.log(i, idx, sim)
if sim > bestScore:
bestScore = sim
bestIdx = idx
if name:
bestName = name
if name:
mol.SetDoubleProp(field_Similarity + "_" + name, sim)
else:
mol.SetDoubleProp(
field_Similarity + "_" + str(idx) + "_Score", sim)
if hits > 0:
count += 1
mol.SetDoubleProp(field_Similarity + "_BestScore", bestScore)
if bestName:
mol.SetProp(field_Similarity + "_BestName", bestName)
else:
mol.SetIntProp(field_Similarity + "_BestIndex", bestIdx)
mol.SetIntProp(field_Similarity + "_Count", hits)
writer.write(mol)
utils.log("Found", count, "similar molecules")
writer.flush()
writer.close()
input.close()
output.close()
if args.meta:
utils.write_metrics(output_base, {
'__InputCount__': i,
'__OutputCount__': count,
'RDKitScreen': count
})
return count
def main():
### command line args defintions #########################################
parser = argparse.ArgumentParser(description='RDKit screen')
group = parser.add_mutually_exclusive_group()
group.add_argument(
'--qsmiles',
help='query structure as smiles (incompatible with -qmolfile arg)')
group.add_argument(
'--qmolfile',
help=
'query structure as filename in molfile format (incompatible with -qsmiles arg)'
)
parser.add_argument('--simmin',
type=float,
default=0.7,
help='similarity lower cutoff (1.0 means identical)')
parser.add_argument('--simmax',
type=float,
default=1.0,
help='similarity upper cutoff (1.0 means identical)')
parser.add_argument('-d',
'--descriptor',
type=str.lower,
choices=list(descriptors.keys()),
default='rdkit',
help='descriptor or fingerprint type (default rdkit)')
parser.add_argument('-m',
'--metric',
type=str.lower,
choices=list(metrics.keys()),
default='tanimoto',
help='similarity metric (default tanimoto)')
parser.add_argument(
'-f',
'--fragment',
choices=['hac', 'mw'],
help=
'Find single fragment if more than one (hac = biggest by heavy atom count, mw = biggest by mol weight )'
)
parser.add_argument('--hacmin', type=int, help='Min heavy atom count')
parser.add_argument('--hacmax', type=int, help='Max heavy atom count')
parser.add_argument('--mwmin', type=float, help='Min mol weight')
parser.add_argument('--mwmax', type=float, help='Max mol weight')
parameter_utils.add_default_io_args(parser)
parser.add_argument('--thin', action='store_true', help='Thin output mode')
parser.add_argument('-q',
'--quiet',
action='store_true',
help='Quiet mode')
args = parser.parse_args()
utils.log("Screen Args: ", args)
descriptor = descriptors[args.descriptor.lower()]
metric = metrics[args.metric.lower()]
if args.qsmiles:
query_rdkitmol = Chem.MolFromSmiles(args.qsmiles)
elif args.qmolfile:
query_rdkitmol = Chem.MolFromMolFile(args.qmolfile)
else:
raise ValueError('No query structure specified')
query_fp = descriptor(query_rdkitmol)
input, output, suppl, writer, output_base = rdkit_utils.default_open_input_output(
args.input,
args.informat,
args.output,
'screen',
args.outformat,
thinOutput=args.thin)
i = 0
count = 0
for mol in suppl:
i += 1
if mol is None: continue
if args.fragment:
mol = mol_utils.fragment(mol, args.fragment, quiet=args.quiet)
if not filter.filter(mol,
minHac=args.hacmin,
maxHac=args.hacmax,
minMw=args.mwmin,
maxMw=args.mwmax,
quiet=args.quiet):
continue
target_fp = descriptor(mol)
sim = metric(query_fp, target_fp)
if sim >= args.simmin and sim <= args.simmax:
count += 1
if not args.quiet:
utils.log(i, sim)
mol.SetDoubleProp(field_Similarity, sim)
writer.write(mol)
utils.log("Found", count, "similar molecules")
writer.flush()
writer.close()
input.close()
output.close()
if args.meta:
utils.write_metrics(output_base, {
'__InputCount__': i,
'__OutputCount__': count,
'RDKitScreen': i
})
def main():
### command line args defintions #########################################
parser = argparse.ArgumentParser(description='RDKit filter')
parser.add_argument(
'-f',
'--fragment',
choices=['hac', 'mw'],
help=
'Find single fragment if more than one (hac = biggest by heavy atom count, mw = biggest by mol weight)'
)
parser.add_argument('--hacmin', type=int, help='Min heavy atom count')
parser.add_argument('--hacmax', type=int, help='Max heavy atom count')
parser.add_argument('--mwmin', type=float, help='Min mol weight')
parser.add_argument('--mwmax', type=float, help='Max mol weight')
parser.add_argument('--rotbmin',
type=float,
help='Min rotatable bond count')
parser.add_argument('--rotbmax',
type=float,
help='Max rotatable bond count')
parser.add_argument('--logpmin', type=float, help='Min logP')
parser.add_argument('--logpmax', type=float, help='Max logP')
parser.add_argument('-l',
'--limit',
type=int,
help='Limit output to this many records')
parser.add_argument(
'-c',
'--chunksize',
type=int,
help=
'Split output into chunks of size c. Output will always be files. Names like filter1.sdf.gz, filter2.sdf.gz ...'
)
parser.add_argument(
'-d',
'--digits',
type=int,
default=0,
help=
'When splitting zero pad the file name to this many digits so that they are in sorted order. Names like filter001.sdf.gz, filter002.sdf.gz ...'
)
parser.add_argument('-r',
'--rename',
action='append',
help='Rename field (fromname:toname)')
parser.add_argument(
'-t',
'--transform',
action='append',
help='Transform field value(fieldname:regex:type). ' +
'Regex is in the form of /regex/substitution/ (the 3 slashes are required). '
+
'Type is of int, float, boolean or string. The type is optional and if not specified then string is assumed. '
+
'Transformation occurs after field renaming so specify the new name.')
parser.add_argument('--delete', action='append', help='Delete field')
parser.add_argument(
'--no-gzip',
action='store_true',
help='Do not compress the output (STDOUT is never compressed')
# WARNING: thin output is not appropriate when using --fragment
parser.add_argument('--thin', action='store_true', help='Thin output mode')
parser.add_argument(
'-q',
'--quiet',
action='store_true',
help='Quiet mode - suppress reporting reason for filtering')
parameter_utils.add_default_io_args(parser)
args = parser.parse_args()
utils.log("Filter Args: ", args)
field_renames = {}
if args.rename:
for t in args.rename:
parts = t.split(':')
if len(parts) != 2:
raise ValueError('Invalid field rename argument:', t)
field_renames[parts[0]] = parts[1]
if args.delete:
for f in args.delete:
field_renames[f] = None
field_regexes = {}
field_replacements = {}
field_types = {}
if args.transform:
for t in args.transform:
parts = t.split(':')
if len(parts) < 2 or len(parts) > 3:
raise ValueError('Invalid field transform argument:', t)
terms = parts[1].split('/')
utils.log("|".join(terms) + str(len(terms)))
field_regexes[parts[0]] = re.compile(terms[1])
field_replacements[parts[0]] = terms[2]
if len(parts) == 3:
t = parts[2]
else:
t = 'string'
field_types[parts[0]] = t
utils.log("Created transform of " + terms[1] + " to " + terms[2] +
" using type of " + t)
if args.delete:
for f in args.delete:
field_renames[f] = None
input, suppl = rdkit_utils.default_open_input(args.input, args.informat)
if args.chunksize:
chunkNum = 1
if args.output:
output_base = args.output
else:
output_base = 'filter'
output_base_chunk = output_base + str(chunkNum).zfill(args.digits)
output, writer, output_base_chunk = rdkit_utils.default_open_output(
output_base_chunk,
output_base_chunk,
args.outformat,
thinOutput=args.thin,
compress=not args.no_gzip)
else:
output, writer, output_base_chunk = rdkit_utils.default_open_output(
args.output,
"filter",
args.outformat,
thinOutput=args.thin,
compress=not args.no_gzip)
output_base = output_base_chunk
utils.log("Writing to " + output_base_chunk)
i = 0
count = 0
chunkNum = 1
for mol in suppl:
if args.limit and count >= args.limit:
break
i += 1
if mol is None: continue
if args.fragment:
mol = mol_utils.fragment(mol, args.fragment, quiet=args.quiet)
if not filter(mol,
minHac=args.hacmin,
maxHac=args.hacmax,
minMw=args.mwmin,
maxMw=args.mwmax,
minRotb=args.rotbmin,
maxRotb=args.rotbmax,
minLogp=args.logpmin,
maxLogp=args.logpmax,
quiet=args.quiet):
continue
if args.chunksize:
if count > 0 and count % args.chunksize == 0:
# new chunk, so create new writer
writer.close()
output.close()
chunkNum += 1
output_chunk_base = output_base + str(chunkNum).zfill(
args.digits)
utils.log("Writing to " + output_chunk_base)
output, writer, output_chunk_base = rdkit_utils.default_open_output(
output_chunk_base,
output_chunk_base,
args.outformat,
thinOutput=args.thin,
compress=not args.no_gzip)
for from_name in field_renames:
to_name = field_renames[from_name]
if mol.HasProp(from_name):
val = mol.GetProp(from_name)
mol.ClearProp(from_name)
if to_name:
mol.SetProp(to_name, val)
for fieldname in field_regexes:
p = mol.GetProp(fieldname)
if p is not None:
regex = field_regexes[fieldname]
q = regex.sub(field_replacements[fieldname], p)
t = field_types[fieldname]
if t == 'int':
mol.SetIntProp(fieldname, int(q))
elif t == 'float':
mol.SetDoubleProp(fieldname, float(q))
elif t == 'boolean':
mol.SetBoolProp(fieldname, bool(q))
else:
mol.SetProp(fieldname, q)
count += 1
writer.write(mol)
utils.log("Filtered", i, "down to", count, "molecules")
if args.chunksize:
utils.log("Wrote", chunkNum, "chunks")
if (args.digits > 0 and len(str(chunkNum)) > args.digits):
utils.log(
"WARNING: not enough digits specified for the number of chunks"
)
writer.flush()
writer.close()
input.close()
output.close()
if args.meta:
utils.write_metrics(output_base, {
'__InputCount__': i,
'__OutputCount__': count,
'RDKitFilter': i
})