def main():
### command line args defintions #########################################
parser = argparse.ArgumentParser(
description='Calculate plane of best fit for molecules')
parameter_utils.add_default_io_args(parser)
args = parser.parse_args()
utils.log("PBFEV args: ", args)
input, output, suppl, writer, output_base = rdkit_utils.default_open_input_output(
args.input, args.informat, args.output, 'PBFEV', args.outformat)
i = 0
count = 0
errors = 0
out_results = []
for mol in suppl:
i += 1
AllChem.EmbedMolecule(mol)
if mol is None: continue
out_vector = PBFev(mol)
if out_vector is None: continue
rd = PBFRD(mol)
mol.SetDoubleProp("distance", rd)
for j, angle in enumerate(out_vector):
mol.SetDoubleProp("angle" + "_" + str(j), angle)
out_results.append(mol)
count = write_out(out_results, count, writer, args.outformat)
utils.log("Handled " + str(i) + " molecules, resulting in " + str(count) +
" outputs")
writer.flush()
writer.close()
input.close()
output.close()
def doO3Dalign(i,
mol,
qmol,
threshold,
perfect_score,
writer,
conformerProps=None,
minEnergy=None):
pyO3As = rdMolAlign.GetO3AForProbeConfs(mol, qmol)
best_score = 0
j = 0
conf_id = -1
for pyO3A in pyO3As:
align = pyO3A.Align()
score = pyO3A.Score()
if score > best_score:
best_score = score
conf_id = j
j += 1
#utils.log("Best score = ",best_score)
if not threshold or perfect_score - best_score < threshold:
utils.log(i, align, score, Chem.MolToSmiles(mol, isomericSmiles=True))
mol.SetDoubleProp(field_O3DAScore, score)
if conformerProps and minEnergy:
eAbs = conformerProps[conf_id][(conformers.field_EnergyAbs)]
eDelta = eAbs - minEnergy
if eAbs:
mol.SetDoubleProp(conformers.field_EnergyAbs, eAbs)
if eDelta:
mol.SetDoubleProp(conformers.field_EnergyDelta, eDelta)
writer.write(mol, confId=conf_id)
return 1
return 0
def main():
parser = argparse.ArgumentParser(description='RDKit constrained conformer generator')
parameter_utils.add_default_io_args(parser)
parser.add_argument('-n', '--num', type=int, default=10, help='number of conformers to generate')
parser.add_argument('-r', '--refmol', help="Reference molecule file")
parser.add_argument('--refmolidx', help="Reference molecule index in file", type=int, default=1)
parser.add_argument('-c', '--core_smi', help='Core substructure. If not specified - guessed using MCS', default='')
args = parser.parse_args()
# Get the reference molecule
ref_mol_input, ref_mol_suppl = rdkit_utils.default_open_input(args.refmol, args.refmol)
counter = 0
# Get the specified reference molecule. Default is the first
for mol in ref_mol_suppl:
counter+=1
if counter == args.refmolidx:
ref_mol = mol
break
ref_mol_input.close()
if counter < args.refmolidx:
raise ValueError("Invalid refmolidx. " + str(args.refmolidx) + " was specified but only " + str(counter) + " molecules were present in refmol.")
# handle metadata
source = "constrained_conf_gen.py"
datasetMetaProps = {"source":source, "description": "Constrained conformer generation using RDKit " + rdBase.rdkitVersion}
clsMappings = {"EmbedRMS": "java.lang.Float"}
fieldMetaProps = [{"fieldName":"EmbedRMS", "values": {"source":source, "description":"Embedding RMS value"}}]
# Get the molecules
input, suppl = rdkit_utils.default_open_input(args.input, args.informat)
output, WRITER, output_base = rdkit_utils.\
default_open_output(args.output, "const_conf_gen", args.outformat,
valueClassMappings=clsMappings,
datasetMetaProps=datasetMetaProps,
fieldMetaProps=fieldMetaProps)
inputs = 0
totalCount = 0
totalErrors = 0
for mol in suppl:
inputs += 1
if mol:
count, errors = generate_conformers(inputs, mol, args.num, ref_mol, WRITER, args.core_smi)
totalCount += count
totalErrors += errors
input.close()
WRITER.close()
if totalErrors > 0:
utils.log("WARNING:", totalErrors, "conformers failed to generate")
# write metrics
if args.meta:
metrics = {'__InputCount__':inputs, '__OutputCount__':totalCount, 'RDKitConstrainedConformer':totalCount}
if totalErrors > 0:
metrics['__ErrorCount__'] = totalErrors
utils.write_metrics(output_base, metrics)
def main():
"""
Example usage:
python -m pipelines.xchem.split-fragnet-candidates -i ../../data/mpro/expanded-17.json
:return:
"""
parser = argparse.ArgumentParser(
description=
'Split fragnet candidates - Split fragment network expansion into individual sets'
)
parser.add_argument('-i',
'--input',
help='JSON containing the expanded candidates)')
parser.add_argument(
'-g',
'--generate-filenames',
action='store_true',
help=
'Use automatically generated file names instead of the title field)')
args = parser.parse_args()
utils.log("Split fragnet candidates args: ", args)
infile = args.input
execute(infile, args.generate_filenames)
def process(inputs, writer):
total = 0
success = 0
errors = 0
for mol in inputs:
total += 1
if mol is None:
errors += 1
continue
try:
quantScore, qualScore = get_fmap_scores(mol)
# utils.log('Score:', score)
if total % 1000 == 0:
utils.log('Processed molecule', total, '...')
mol.SetDoubleProp(field_FeatureSteinQualityScore, qualScore)
mol.SetDoubleProp(field_FeatureSteinQuantityScore, quantScore)
mol.SetProp('Name', mol.GetProp('_Name'))
writer.write(mol)
success += 1
except:
utils.log("Error scoring molecule", sys.exc_info()[0])
traceback.print_exc()
errors += 1
return total, success, errors
def parse_mol_simple(my_type, txt):
"""Function to parse individual mols given a type.
:param my_type: A type definition (i.e. "mol" or "smiles")
:param txt: The textual definition of the molecule (i.e. a SMILES string)
:return: A mol instance or None if the molecule could not be compiled
"""
# Ignore unexpected parameter values...
if my_type is None or not my_type or txt is None or not txt:
return None
if my_type == "mol":
# Try this way
mol = Chem.MolFromMolBlock(txt.strip())
if mol is None:
mol = Chem.MolFromMolBlock(txt)
if mol is None:
mol = Chem.MolFromMolBlock("\n".join(txt.split("\n")[1:]))
# Now try to do sanifix
if mol is None:
mol = fix_mol(Chem.MolFromMolBlock(txt, False))
# And again
if mol is None:
mol = fix_mol(Chem.MolFromMolBlock(txt.strip(), False))
elif my_type == "smiles":
# Assumes that smiles is the first column -> and splits on chemaxon
mol = Chem.MolFromSmiles(txt.split()[0].split(":")[0])
if mol is None:
utils.log('Failed to parse mol "%s" for my_type %s' % (txt, my_type))
return mol
def main():
### command line args definitions #########################################
parser = argparse.ArgumentParser(description='RDKit Input Splitter')
parameter_utils.add_default_input_args(parser)
parser.add_argument('-o',
'--output',
required=True,
help="Directory name for output files (no extension).")
parser.add_argument(
'-f',
'--field',
required=True,
help=
"field to use to split input. Output files will have the name of this field's value"
)
parser.add_argument('--meta',
action='store_true',
help='Write metadata and metrics files')
args = parser.parse_args()
utils.log("Splitter Args: ", args)
filenames = split(args.input, args.informat, args.field, args.output,
args.meta)
utils.log("Files generated:", " ".join(filenames))
def patch_scores_sdf(sdf_in, outfile, scores):
global work_dir
counter = 0
sdf_path = "{0}{1}{2}.sdf".format(work_dir, os.path.sep, outfile)
tsv_path = "{0}{1}{2}.tsv".format(work_dir, os.path.sep, outfile)
utils.log("Writing results to {0} and {1}".format(sdf_path, tsv_path))
with open(tsv_path, 'w') as tsv_file:
sdf_file = pybel.Outputfile("sdf", sdf_path)
for mol in pybel.readfile("sdf", sdf_in):
if counter in scores:
score = scores[counter]
# utils.log("Score for record {0} is {1}".format(counter, score))
mol.data['dls_deep_score'] = score
if 'SCORE' in mol.data:
rdock_score = mol.data['SCORE']
else:
rdock_score = ''
if 'SCORE.norm' in mol.data:
rdock_nscore = mol.data['SCORE.norm']
else:
rdock_nscore = ''
sdf_file.write(mol)
tsv_file.write("{0}\t{1}\t{2}\t{3}\n".format(
counter, rdock_score, rdock_nscore, score))
else:
utils.log("No score found for record", counter)
counter += 1
sdf_file.close()
def filter_value(value, min, max, key, quiet=False):
if value is not None and value < min:
if not quiet:
utils.log(key, value, "<", min)
return False
if value is not None and value > max:
if not quiet:
utils.log(key, value, ">", max)
return False
return True
def run_predictions():
global types_file_name
global predict_file_name
global work_dir
# python3 scripts/predict.py -m resources/dense.prototxt -w resources/weights.caffemodel -i work_0/test_set.types >> work_0/caffe_output/predictions.txt
cmd1 = "python3 /root/train/fragalysis_test_files/scripts/predict.py -m /root/train/fragalysis_test_files/resources/dense.prototxt" +\
" -w /root/train/fragalysis_test_files/resources/weights.caffemodel" +\
" -i {0}/{1} -o {0}/{2}".format(work_dir, types_file_name, predict_file_name)
utils.log("CMD:", cmd1)
os.system(cmd1)
def determine_protein_format(protein_file, protein_format):
if protein_format:
return protein_format
elif protein_file.endswith('.pdb') or protein_file.endswith('.pdb.gz'):
return 'pdb'
elif protein_file.endswith('.mol2') or protein_file.endswith('.mol2.gz'):
return 'mol2'
else:
utils.log(
"Can't determine file format, so assuming pdb. Please use the --protein-format parameter"
)
return 'pdb'
def read_next_protein(proteins, format, previous, index, keep_hs=False):
if previous and index >= len(proteins):
return previous
protein = next(
toolkit.readfile(format, proteins[index], removeHs=not keep_hs))
if not protein:
raise ValueError('Unable to read protein')
else:
utils.log('Read protein', index + 1)
protein.protein = True
protein.removeh()
return protein
def create_featuremap(fragments):
mols = [m for m in fragments if m]
fmaps, scores = build_feat_data(mols)
merged_fmaps = fmaps.copy()
utils.log('Processing', len(fmaps), 'molecules')
while len(merged_fmaps) > 1:
merge_feat_maps(merged_fmaps, scores)
merged_fmap = merged_fmaps[0]
utils.log('Created merged feature map with', merged_fmap.GetNumFeatures(),
'features')
return merged_fmap
def process(inputs, fname):
mols = [m for m in inputs if m]
fmaps, scores = build_feat_data(mols)
merged_fmaps = fmaps.copy()
utils.log('Processing', len(fmaps), 'molecules')
while len(merged_fmaps) > 1:
merge_feat_maps(merged_fmaps, scores)
merged_fmap = merged_fmaps[0]
pickle.dump(merged_fmap, open(fname, "wb"))
utils.log('Wrote merged feature map with', merged_fmap.GetNumFeatures(),
'features as pickle to', fname)
return len(mols), merged_fmap.GetNumFeatures()
def split(input, informat, fieldName, outputBase, writeMetrics):
"""Splits the input into separate files. The name of each file and the file the each record is written to
is determined by the fieldName parameter
"""
input, suppl = rdkit_utils.default_open_input(input, informat)
i = 0
written = 0
writers = {}
outputs = []
filenames = []
for mol in suppl:
i += 1
if mol is None: continue
if not mol.HasProp(fieldName):
utils.log("Skipping molecule", i, "- did not contain field",
fieldName)
continue
value = mol.GetProp(fieldName)
if value:
s = str(value)
if writers.has_key(s):
writer = writers[s]
else:
name = outputBase + s
output, writer = rdkit_utils.default_open_output_sdf(
name, outputBase, False, False)
filenames.append(name + '.sdf')
outputs.append(output)
writers[s] = writer
writer.write(mol)
written += 1
utils.log("Generated", len(writers), "outputs from", i, "records")
input.close()
for k in writers:
writers[k].close()
for o in outputs:
o.close()
if writeMetrics:
utils.write_metrics(outputBase, {
'__InputCount__': i,
'__OutputCount__': written,
'Splitter': i
})
return filenames
def generate_mols_from_json(input):
"""Create a supplier of RDKit Mol objects from the json
:param input: file like object containing the json representation of the molecules
"""
j = 0
for item in input:
j += 1
mol = create_mol_from_props(item)
if not mol:
# TODO - get a count of the errors and report
utils.log("Failed to create molecule - skipping. Data was ", item)
continue
yield mol
def main():
global PDB_PATH,WRITER,THRESHOLD
parser = argparse.ArgumentParser(description='Open babel PDB prepare')
parser.add_argument('--no-gzip', action='store_true', help='Do not compress the output')
parser.add_argument('-i', '--input', help="PDB file for converting")
parser.add_argument('-o', '--output', help="Base name for output files (no extension).")
parser.add_argument('-mol2', '--mol2', action='store_true', help='Output as Mol2 format.')
parser.add_argument('-pdbqt', '--pdbqt', action='store_true', help='Output as pdbqt format.')
parser.add_argument('--meta', action='store_true', help='Write metrics files')
parser.add_argument('-prot', '--protonate', type=float, help="protonate at this pH (optional)")
args = parser.parse_args()
utils.log("Prepare Args: ", args)
if not (args.mol2 or args.pdbqt):
raise ValueError("Must specify at least one output fromat: mol2 and/or pdbqt")
if args.pdbqt:
utils.log("Preparing as pdbqt")
execute(args.input, args.output, "pdbqt", "-opdbqt", args.protonate, args.no_gzip)
if args.mol2:
utils.log("Preparing as mol2")
execute(args.input, args.output, "mol2", "-omol2", args.protonate, args.no_gzip)
utils.log("Preparation complete")
def SelectDiverseSubset(mols, clusters, distances, count, field, maximise,
score, quiet):
total = len(mols)
num_clusters = len(clusters)
pickedList = []
clustersList = []
for i in range(0, num_clusters):
pickedList.append([])
if field:
filteredByValue = [
x for x in clusters[i] if mols[x].HasProp(field)
]
sortedByValue = sorted(
filteredByValue,
key=lambda idx: FetchScore(idx, mols, field, maximise))
clustersList.append(sortedByValue)
else:
allRecords = [x for x in clusters[i]]
clustersList.append(allRecords)
totalIter = 0
clusterIter = 0
pickedCount = 0
while totalIter < total and pickedCount < count:
clusterNum = totalIter % num_clusters
clus = clustersList[clusterNum]
pick = pickedList[clusterNum]
#utils.log("iter",totalIter,"cluster",clusterNum,"length",len(clus))
if len(clus) > 0:
# remove that item from the cluster so that it's not tried again
molIndex = clus.pop(0)
if len(pick) == 0: # first time for this cluster
pick.append(molIndex)
pickedCount += 1
clusterIter += 1
if not quiet:
utils.log("Cluster", clusterNum, "initialised with",
molIndex)
else:
closestDist = GetClosestDistance(distances, molIndex, pick)
#utils.log("Closest score",closestDist)
if closestDist < score:
pick.append(molIndex)
pickedCount += 1
clusterIter += 1
if not quiet:
utils.log("Cluster", clusterNum, "added", molIndex,
"with score", closestDist)
elif not quiet:
utils.log("Cluster", clusterNum, "discarded", molIndex,
"with score", closestDist)
else: # cluster has been exhausted
#utils.log("Cluster",clusterNum,"exhasted")
clusterIter += 1
totalIter += 1
utils.log("Picked", pickedCount, "using", totalIter, "iterations")
return pickedList
def main():
# Example usage
# python -m pipelines.xchem.featurestein_score -i ../../data/mpro/poses.sdf.gz -f mpro-fstein.p -o fstein
global fmaps
parser = argparse.ArgumentParser(description='FeatureStein scoring with RDKit')
parameter_utils.add_default_io_args(parser)
parser.add_argument('-f', '--feat-map', help='Feature Map pickle to score with')
parser.add_argument('--no-gzip', action='store_true', help='Do not compress the output (STDOUT is never compressed')
parser.add_argument('--metrics', action='store_true', help='Write metrics')
args = parser.parse_args()
utils.log("FeatureStein Args: ", args)
source = "featurestein_score.py"
datasetMetaProps = {"source":source, "description": "FeatureStein scoring using RDKit " + rdBase.rdkitVersion}
clsMappings = {}
fieldMetaProps = []
clsMappings[field_FeatureSteinQualityScore] = "java.lang.Float"
clsMappings[field_FeatureSteinQuantityScore] = "java.lang.Float"
fieldMetaProps.append({"fieldName":field_FeatureSteinQualityScore, "values": {"source":source, "description":"FeatureStein quality score"},
"fieldName":field_FeatureSteinQuantityScore, "values": {"source":source, "description":"FeatureStein quantity score"}})
pkl_file = open(args.feat_map, 'rb')
fmaps = pickle.load(pkl_file)
utils.log('FeatureMap has', fmaps.GetNumFeatures(), "features")
inputs_file, inputs_supplr = rdkit_utils.default_open_input(args.input, args.informat)
output, writer, output_base = rdkit_utils.default_open_output(args.output,
'featurestein', args.outformat,
valueClassMappings=clsMappings,
datasetMetaProps=datasetMetaProps,
fieldMetaProps=fieldMetaProps,
compress=not args.no_gzip)
# this does the processing
total, success, errors = process(inputs_supplr, writer)
inputs_file.close()
writer.flush()
writer.close()
output.close()
if args.metrics:
utils.write_metrics(output_base, {'__InputCount__':total, '__OutputCount__':success, '__ErrorCount__':errors, 'RDKitFeatureMap':success})
def mock_predictions():
global work_dir
global predict_file_name
global inputs_protein
global inputs_ligands
utils.log("WARNING: generating mock results instead of running on GPU")
outfile = generate_predictions_filename(work_dir, predict_file_name)
with open(outfile, 'w') as predictions:
for protein in inputs_protein:
for ligand in inputs_ligands:
score = random.random()
line = "{0} | 0 {1}{4}proteins{4}{2} {1}{4}ligands{4}{3}\n".format(
score, work_dir, protein, ligand, os.path.sep)
# utils.log("Writing", line)
predictions.write(line)
def main():
### command line args defintions #########################################
parser = argparse.ArgumentParser(description='RDKit cluster 3D')
parameter_utils.add_default_io_args(parser)
args = parser.parse_args()
utils.log("Cluster_3d Args: ", args)
source = "cluster_3d.py"
datasetMetaProps = {
"source": source,
"description": "Cluster 3D using RDKit " + rdBase.rdkitVersion
}
clsMappings = {
# "RMSToCentroid": "java.lang.Float",
# "EnergyDelta": "java.lang.Float",
# "EnergyAbs": "java.lang.Float",
# "ConformerNum": "java.lang.Integer",
# "ClusterCentroid": "java.lang.Integer",
# "ClusterNum": "java.lang.Integer",
# "StructureNum": "java.lang.Integer"
}
fieldMetaProps = [
# {"fieldName":"RMSToCentroid", "values": {"source":source, "description":"RMS distance to the cluster centroid"}},
# {"fieldName":"EnergyDelta", "values": {"source":source, "description":"Energy difference to lowest energy structure"}},
# {"fieldName":"EnergyAbs", "values": {"source":source, "description":"Absolute energy"}},
# {"fieldName":"ConformerNum", "values": {"source":source, "description":"Conformer number"}},
# {"fieldName":"ClusterCentroid", "values": {"source":source, "description":"Conformer number of the cluster centroid"}},
# {"fieldName":"ClusterNum", "values": {"source":source, "description":"Cluster number"}},
# {"fieldName":"StructureNum", "values": {"source":source, "description":"Structure number this conformer was generated from"}}
]
input, output, suppl, writer, output_base = rdkit_utils. \
default_open_input_output(args.input, args.informat, args.output,
'conformers', args.outformat,
valueClassMappings=clsMappings,
datasetMetaProps=datasetMetaProps,
fieldMetaProps=fieldMetaProps)
basemol = combine_conformers(suppl)
if input:
input.close()
writer.flush()
writer.close()
output.close()
def get_FeatureMapScore(small_feats,
large_feats,
tani=False,
score_mode=FeatMaps.FeatMapScoreMode.All):
"""
Generate the feature map score.
:param small_feats:
:param large_feats:
:param tani:
:return:
"""
featLists = []
for rawFeats in [small_feats, large_feats]:
# filter that list down to only include the ones we're interested in
featLists.append(rawFeats)
fms = [
FeatMaps.FeatMap(feats=x, weights=[1] * len(x), params=fmParams)
for x in featLists
]
# set the score mode
fms[0].scoreMode = score_mode
try:
if tani:
c = fms[0].ScoreFeats(featLists[1])
A = fms[0].GetNumFeatures()
B = len(featLists[1])
if B != fms[1].GetNumFeatures():
utils.log("Why isn't B equal to number of features...?!")
tani_score = float(c) / (A + B - c)
return tani_score
else:
fm_score = fms[0].ScoreFeats(featLists[1]) / min(
fms[0].GetNumFeatures(), len(featLists[1]))
return fm_score
except ZeroDivisionError:
utils.log("ZeroDivisionError")
return 0
if tani:
tani_score = float(c) / (A + B - c)
return tani_score
else:
fm_score = fms[0].ScoreFeats(featLists[1]) / min(
fms[0].GetNumFeatures(), len(featLists[1]))
return fm_score
def process(molecules, fragments, writer, threshold=0.4):
frag_mol_list = []
errors = 0
for m in fragments:
if m is None:
errors += 1
else:
frag_mol_list.append(m)
if errors:
utils.log(errors, 'molecules failed to load. Using', len(frag_mol_list), 'fragments.')
else:
utils.log('Using', len(frag_mol_list), 'fragments. No errors')
#mols, frags, score_threshold, writer
getReverseScores(molecules, frag_mol_list, threshold, writer)
def execute(input, output, extension, format, ph, noGzip):
# TODO - convert this to use the Python API rather than an external process
filename = output + "." + extension
base_args = ["obabel", "-ipdb", input, format, "-O", filename]
if ph:
base_args.append("-p")
base_args.append(str(ph))
utils.log("Command: " + " ".join(base_args))
subprocess.check_call(base_args, stdout=sys.stderr, stderr=sys.stderr)
# NOTE the -z argument does not seem to work correctly with obabel (truncated files generated) so we
# fall back to good old gzip to handle the compression once the uncompressed file is created
if not noGzip:
subprocess.check_call(['gzip', filename], stdout=sys.stderr, stderr=sys.stderr)
def get_SucosScore(ref_mol,
query_mol,
tani=False,
ref_features=None,
query_features=None,
score_mode=FeatMaps.FeatMapScoreMode.All):
"""
This is the key function that calculates the SuCOS scores and is expected to be called from other modules.
To improve performance you can pre-calculate the features and pass them in as optional parameters to avoid having
to recalculate them. Use the getRawFeatures function to pre-calculate the features.
:param ref_mol: The reference molecule to compare to
:param query_mol: The molecule to compare to the reference
:param tani: Whether to calculate Tanimoto distances
:param ref_features: An optional feature map for the reference molecule, avoiding the need to re-calculate it.
:param query_features: An optional feature map for the query molecule, avoiding the need to re-calculate it.
:return: A tuple of 3 values. 1 the sucos score, 2 the feature map score,
3 the Tanimoto distance or 1 minus the protrude distance
"""
if not ref_features:
ref_features = getRawFeatures(ref_mol)
if not query_features:
query_features = getRawFeatures(query_mol)
fm_score = get_FeatureMapScore(ref_features, query_features, tani,
score_mode)
fm_score = np.clip(fm_score, 0, 1)
try:
if tani:
tani_sim = 1 - float(
rdShapeHelpers.ShapeTanimotoDist(ref_mol, query_mol))
tani_sim = np.clip(tani_sim, 0, 1)
SuCOS_score = 0.5 * fm_score + 0.5 * tani_sim
return SuCOS_score, fm_score, tani_sim
else:
protrude_dist = rdShapeHelpers.ShapeProtrudeDist(
ref_mol, query_mol, allowReordering=False)
protrude_dist = np.clip(protrude_dist, 0, 1)
protrude_val = 1.0 - protrude_dist
SuCOS_score = 0.5 * fm_score + 0.5 * protrude_val
return SuCOS_score, fm_score, protrude_val
except:
utils.log("Failed to calculate SuCOS scores. Returning 0,0,0")
return 0, 0, 0
def read_predictions():
global predict_file_name
global work_dir
scores = {}
with open("{0}{1}{2}".format(work_dir, os.path.sep, predict_file_name),
'r') as input:
for line in input:
#utils.log(line)
tokens = line.split()
if len(tokens) == 5 and tokens[1] == '|':
# utils.log(len(tokens), tokens[0], tokens[3], tokens[4])
record_no = match_ligand(tokens[4])
if record_no is not None:
# utils.log(record_no, tokens[0])
scores[record_no] = tokens[0]
utils.log("Found", len(scores), "scores")
return scores
def main():
# Example usage:
# python -m pipelines.xchem.xcos -f ../../data/mpro/hits-17.sdf.gz -i ../../data/mpro/poses.sdf.gz -o xcos
parser = argparse.ArgumentParser(description='XCos scoring with RDKit')
parameter_utils.add_default_io_args(parser)
parser.add_argument('-f', '--fragments', required=True, help='Fragments to compare')
parser.add_argument('-ff', '--fragments-format', help='Fragments format')
parser.add_argument('-t', '--score-threshold', type=float, default=0.4,
help='Minimum shape overlay and feature map score required for scoring a bit to a fragment')
parser.add_argument('--no-gzip', action='store_true', help='Do not compress the output (STDOUT is never compressed')
parser.add_argument('--metrics', action='store_true', help='Write metrics')
args = parser.parse_args()
utils.log("XCos Args: ", args)
source = "xcos.py"
datasetMetaProps = {"source":source, "description": "XCos scoring using RDKit " + rdBase.rdkitVersion}
clsMappings = {}
fieldMetaProps = []
clsMappings[field_XCosRefMols] = "java.lang.String"
clsMappings[field_XCosNumHits] = "java.lang.Integer"
clsMappings[field_XCosScore1] = "java.lang.Float"
fieldMetaProps.append({"fieldName":field_XCosRefMols, "values": {"source":source, "description":"XCos reference fragments"}})
fieldMetaProps.append({"fieldName":field_XCosNumHits, "values": {"source":source, "description":"XCos number of hits"}})
fieldMetaProps.append({"fieldName":field_XCosScore1, "values": {"source":source, "description":"XCos score 1"}})
frags_input,frags_suppl = rdkit_utils.default_open_input(args.fragments, args.fragments_format)
inputs_file, inputs_supplr = rdkit_utils.default_open_input(args.input, args.informat)
output, writer, output_base = rdkit_utils.default_open_output(args.output,
'xcos', args.outformat,
valueClassMappings=clsMappings,
datasetMetaProps=datasetMetaProps,
fieldMetaProps=fieldMetaProps,
compress=not args.no_gzip)
# this does the processing
process(inputs_supplr, frags_suppl, writer, threshold=args.score_threshold)
writer.close()
def doO3Dalign(i,
mol,
qmol,
use_crippen,
threshold,
perfect_score,
writer,
conformerProps=None,
minEnergy=None):
if use_crippen:
pyO3As = rdMolAlign.GetCrippenO3AForProbeConfs(mol, qmol)
else:
pyO3As = rdMolAlign.GetO3AForProbeConfs(mol, qmol)
if len(pyO3As) == 0:
return 0
best_score = 0
j = 0
conf_id = -1
for pyO3A in pyO3As:
align = pyO3A.Align()
score = pyO3A.Score()
if score > best_score:
best_score = score
conf_id = j
j += 1
#utils.log("Best score = ",best_score)
if not threshold or perfect_score - best_score < threshold:
utils.log("Molecule", i, align, score)
mol.SetDoubleProp(field_O3DAScore, score)
if conformerProps and minEnergy:
eAbs = conformerProps[conf_id][(conformers.field_EnergyAbs)]
eDelta = eAbs - minEnergy
if eAbs:
mol.SetDoubleProp(conformers.field_EnergyAbs, eAbs)
if eDelta:
mol.SetDoubleProp(conformers.field_EnergyDelta, eDelta)
writer.write(mol, confId=conf_id)
return 1
return 0
def fragment(mol, mode, quiet=False):
frags = Chem.GetMolFrags(mol, asMols=True)
if len(frags) == 1:
return mol
else:
# TODO - handle ties
biggest_index = -1
i = 0
if mode == 'hac':
biggest_count = 0
for frag in frags:
hac = frag.GetNumHeavyAtoms()
if hac > biggest_count:
biggest_count = hac
biggest_mol = frag
biggest_index = i
i+=1
if not quiet:
utils.log("Chose fragment", biggest_index, "from", len(frags), "based on HAC")
elif mode == 'mw':
biggest_mw = 0
for frag in frags:
mw = Descriptors.MolWt(frag)
if mw > biggest_mw:
biggest_mw = mw
biggest_mol = frag
biggest_index = i
i+=1
if not quiet:
utils.log("Chose fragment", biggest_index, "from", len(frags), "based on MW")
else:
raise ValueError('Invalid fragment mode:',mode)
# copy the properties across
for name in mol.GetPropNames():
biggest_mol.SetProp(name, mol.GetProp(name))
# _Name is a magical property that is not in the ones returned by GetPropNames
if '_Name' in mol.GetPropNames():
biggest_mol.SetProp("_Name", mol.GetProp("_Name"))
return biggest_mol
def run_dock(mol):
global WRITER, COUNTER, SUCCESS, THRESHOLD
answer_dict = run_and_get_ans(mol, PDB_PATH)
COUNTER += 1
if not answer_dict:
utils.log("FAILED MOL", Chem.MolToSmiles(mol))
return
if THRESHOLD is not None:
if answer_dict["system"]["pliff_score"] > THRESHOLD:
return
for ans in answer_dict["system"]:
if ans.startswith(u"pliff"):
mol.SetDoubleProp(str(ans), answer_dict["system"][ans])
utils.log("SCORED MOL:", Chem.MolToSmiles(mol), answer_dict)
lock.acquire()
WRITER.write(mol)
SUCCESS += 1
WRITER.flush()
lock.release()