def calc_distance_matrix(mols):
"""
Calculate a full distance matrix for the given molecules. Identical molecules get a score of 0.0 with the maximum
distance possible being 1.0.
:param mols: A list of molecules. It must be possible to iterate through this list multiple times
:return: A NxN 2D array of distance scores, with N being the number of molecules in the input
"""
# TODO - do we need to calculate both sides of the matrix? Tanimoto is supposed to be a symmetric distance measure,
# but the matrix that is generated does not seem to be symmetric.
mol_fm_tuples = []
for mol in mols:
features = sucos.getRawFeatures(mol)
mol_fm_tuples.append((mol, features))
matrix = []
for tuple1 in mol_fm_tuples:
tmp = []
for tuple2 in mol_fm_tuples:
if tuple1[0] == tuple2[0]:
tmp.append(0.0)
else:
#utils.log("Calculating SuCOS between", mol1, mol2)
sucos_score, fm_score, tani_score = sucos.get_SucosScore(
tuple1[0],
tuple2[0],
tani=True,
ref_features=tuple1[1],
query_features=tuple2[1])
tmp.append(1.0 - sucos_score)
matrix.append(tmp)
return matrix
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")