def main():
parser = argparse.ArgumentParser(
description='Clustering with SuCOS and RDKit')
parser.add_argument(
'-i',
'--input',
help='Input file in SDF format. Can be gzipped (*.gz).')
parser.add_argument(
'-o',
'--output',
default="cluster",
help="Base name for output files in SDF format. " +
"e.g. if value is 'output' then files like output1.sdf, output2.sdf will be created"
)
parser.add_argument(
'--gzip',
action='store_true',
help=
'Gzip the outputs generating files like output1.sdf.gz, output2.sdf.gz'
)
parser.add_argument('-t',
'--threshold',
type=float,
default=0.8,
help='Clustering threshold')
args = parser.parse_args()
utils.log("SuCOS Cluster Args: ", args)
input_file = utils.open_file_for_reading(args.input)
suppl = Chem.ForwardSDMolSupplier(input_file)
mols = list(suppl)
matrix = calc_distance_matrix(mols)
clusters = cluster(matrix, threshold=args.threshold)
write_clusters_to_sdfs(mols, clusters, args.output, gzip=args.gzip)
def process(
refmol_filename,
inputs_filename,
outputs_filename,
refmol_index=None,
refmol_format=None,
tani=False,
score_mode=FeatMaps.FeatMapScoreMode.All,
):
ref_mol = utils.read_single_molecule(refmol_filename,
index=refmol_index,
format=refmol_format)
# utils.log("Reference mol has", ref_mol.GetNumHeavyAtoms(), "heavy atoms")
ref_features = getRawFeatures(ref_mol)
input_file = utils.open_file_for_reading(inputs_filename)
suppl = Chem.ForwardSDMolSupplier(input_file)
output_file = utils.open_file_for_writing(outputs_filename)
writer = Chem.SDWriter(output_file)
count = 0
total = 0
errors = 0
for mol in suppl:
count += 1
if mol is None:
continue
# utils.log("Mol has", str(mol.GetNumHeavyAtoms()), "heavy atoms")
try:
sucos_score, fm_score, val3 = get_SucosScore(
ref_mol,
mol,
tani=tani,
ref_features=ref_features,
score_mode=score_mode,
)
mol.SetDoubleProp("SuCOS_Score", sucos_score)
mol.SetDoubleProp("SuCOS_FeatureMap_Score", fm_score)
if tani:
mol.SetDoubleProp("SuCOS_Tanimoto_Score", val3)
else:
mol.SetDoubleProp("SuCOS_Protrude_Score", val3)
utils.log("Scores:", sucos_score, fm_score, val3)
writer.write(mol)
total += 1
except ValueError as e:
errors += 1
utils.log("Molecule", count, "failed to score:", e.message)
input_file.close()
writer.flush()
writer.close()
output_file.close()
utils.log("Completed.", total, "processed, ", count, "succeeded, ", errors,
"errors")
def process(inputfilename, clusterfilenames, outputfilename, mode):
all_clusters = {}
for filename in clusterfilenames:
cluster = []
cluster_file = utils.open_file_for_reading(filename)
suppl = Chem.ForwardSDMolSupplier(cluster_file)
i = 0
for mol in suppl:
i += 1
if not mol:
utils.log("WARNING: failed to generate molecule", i,
"in cluster", filename)
continue
try:
features = sucos.getRawFeatures(mol)
cluster.append((mol, features))
except:
utils.log("WARNING: failed to generate features for molecule",
i, "in cluster", filename)
cluster_file.close()
all_clusters[filename] = cluster
input_file = utils.open_file_for_reading(inputfilename)
suppl = Chem.ForwardSDMolSupplier(input_file)
output_file = utils.open_file_for_writing(outputfilename)
writer = Chem.SDWriter(output_file)
comparisons = 0
mol_num = 0
for mol in suppl:
mol_num += 1
if not mol:
utils.log("WARNING: failed to generate molecule", mol_num,
"in input")
continue
try:
query_features = sucos.getRawFeatures(mol)
except:
utils.log("WARNING: failed to generate features for molecule",
mol_num, "in input")
continue
scores = [0, 0, 0]
for clusterfilename in all_clusters:
cluster = all_clusters[clusterfilename]
index = 0
for entry in cluster:
hit = entry[0]
ref_features = entry[1]
index += 1
comparisons += 1
sucos_score, fm_score, vol_score = sucos.get_SucosScore(
hit,
mol,
tani=False,
ref_features=ref_features,
query_features=query_features)
if mode == 'max':
if sucos_score > scores[0]:
scores[0] = sucos_score
scores[1] = fm_score
scores[2] = vol_score
cluster_name = clusterfilename
cluster_index = index
elif mode == 'cum':
scores[0] += sucos_score
scores[1] += fm_score
scores[2] += vol_score
else:
raise ValueError("Invalid mode: " + mode)
if scores[0] > 0:
if mode == 'max':
cluster_file_name_only = cluster_name.split(os.sep)[-1]
#utils.log("Max SuCOS:", scores[0], "FM:", scores[1], "P:", scores[2],"File:", cluster_file_name_only, "Index:", cluster_index)
mol.SetDoubleProp("Max_SuCOS_Score", scores[0])
mol.SetDoubleProp("Max_SuCOS_FeatureMap_Score", scores[1])
mol.SetDoubleProp("Max_SuCOS_Protrude_Score", scores[2])
mol.SetProp("Max_SuCOS_Cluster", cluster_file_name_only)
mol.SetIntProp("Max_SuCOS_Index", cluster_index)
else:
#utils.log("Cum SuCOS:", scores[0], "FM:", scores[1], "P:", scores[2])
mol.SetDoubleProp("Cum_SuCOS_Score", scores[0])
mol.SetDoubleProp("Cum_SuCOS_FeatureMap_Score", scores[1])
mol.SetDoubleProp("Cum_SuCOS_Protrude_Score", scores[2])
writer.write(mol)
else:
utils.log("Molecule", mol_num,
"did not overlay. Omitting from results")
input_file.close()
writer.flush()
writer.close()
output_file.close()
utils.log("Completed", comparisons, "comparisons")
def process(inputfilename, clusterfilenames, outputfilename, filter_value,
filter_field):
all_clusters = {}
for filename in clusterfilenames:
cluster = []
cluster_file = utils.open_file_for_reading(filename)
suppl = Chem.ForwardSDMolSupplier(cluster_file)
i = 0
for mol in suppl:
i += 1
if not mol:
utils.log("WARNING: failed to generate molecule", i,
"in cluster", filename)
continue
try:
features = sucos.getRawFeatures(mol)
cluster.append((mol, features))
except:
utils.log("WARNING: failed to generate features for molecule",
i, "in cluster", filename)
cluster_file.close()
all_clusters[filename] = cluster
input_file = utils.open_file_for_reading(inputfilename)
suppl = Chem.ForwardSDMolSupplier(input_file)
output_file = utils.open_file_for_writing(outputfilename)
writer = Chem.SDWriter(output_file)
comparisons = 0
mol_num = 0
for mol in suppl:
mol_num += 1
if not mol:
utils.log("WARNING: failed to generate molecule", mol_num,
"in input")
continue
try:
query_features = sucos.getRawFeatures(mol)
except:
utils.log("WARNING: failed to generate features for molecule",
mol_num, "in input")
continue
scores_max = [0, 0, 0]
scores_cum = [0, 0, 0]
cluster_name = None
for clusterfilename in all_clusters:
cluster = all_clusters[clusterfilename]
index = 0
for entry in cluster:
hit = entry[0]
ref_features = entry[1]
index += 1
comparisons += 1
sucos_score, fm_score, vol_score = sucos.get_SucosScore(
hit,
mol,
tani=False,
ref_features=ref_features,
query_features=query_features)
if sucos_score > scores_max[0]:
scores_max[0] = sucos_score
scores_max[1] = fm_score
scores_max[2] = vol_score
cluster_name = clusterfilename
cluster_index = index
scores_cum[0] += sucos_score
scores_cum[1] += fm_score
scores_cum[2] += vol_score
# utils.log("Max SuCOS:", scores[0], "FM:", scores[1], "P:", scores[2],"File:", cluster_file_name_only, "Index:", cluster_index)
mol.SetDoubleProp("Max_SuCOS_Score",
scores_max[0] if scores_max[0] > 0 else 0)
mol.SetDoubleProp("Max_SuCOS_FeatureMap_Score",
scores_max[1] if scores_max[1] > 0 else 0)
mol.SetDoubleProp("Max_SuCOS_Protrude_Score",
scores_max[2] if scores_max[2] > 0 else 0)
if cluster_name:
cluster_file_name_only = cluster_name.split(os.sep)[-1]
mol.SetProp("Max_SuCOS_Cluster", cluster_file_name_only)
mol.SetIntProp("Max_SuCOS_Index", cluster_index)
# utils.log("Cum SuCOS:", scores[0], "FM:", scores[1], "P:", scores[2])
mol.SetDoubleProp("Cum_SuCOS_Score",
scores_cum[0] if scores_cum[0] > 0 else 0)
mol.SetDoubleProp("Cum_SuCOS_FeatureMap_Score",
scores_cum[1] if scores_cum[1] > 0 else 0)
mol.SetDoubleProp("Cum_SuCOS_Protrude_Score",
scores_cum[2] if scores_cum[2] > 0 else 0)
if filter_value and filter_field:
if mol.HasProp(filter_field):
val = mol.GetDoubleProp(filter_field)
if val > filter_value:
writer.write(mol)
else:
writer.write(mol)
input_file.close()
writer.flush()
writer.close()
output_file.close()
utils.log("Completed", comparisons, "comparisons")