better error handling for O3DAlign

Author: tdudgeon

src/python/pipelines/rdkit/o3dAlign.py

@@ -27,13 +27,17 @@
 
 field_O3DAScore = "O3DAScore"
 
-### start function defintions #########################################
+### start function definitions #########################################
 
 def doO3Dalign(i, mol, qmol, threshold, perfect_score, writer, conformerProps=None, minEnergy=None):
     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()
@@ -44,7 +48,7 @@ def doO3Dalign(i, mol, qmol, threshold, perfect_score, writer, conformerProps=No
 
     #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))
+        utils.log("Molecule", i, align, score)
         mol.SetDoubleProp(field_O3DAScore, score)
         if conformerProps and minEnergy:
             eAbs = conformerProps[conf_id][(conformers.field_EnergyAbs)]
@@ -68,12 +72,15 @@ def main():
     parser.add_argument('-n', '--num', default=0, type=int, help='number of conformers to generate, if None then input structures are assumed to already be 3D')
     parser.add_argument('-a', '--attempts', default=0, type=int, help='number of attempts to generate conformers')
     parser.add_argument('-r', '--rmsd', type=float, default=1.0, help='prune RMSD threshold for excluding conformers')
-    parser.add_argument('-e', '--emin', type=int, default=0, help='energy minimisation iterations for generated confomers (default of 0 means none)')
+    parser.add_argument('-e', '--emin', type=int, default=0, help='energy minimisation iterations for generated conformers (default of 0 means none)')
     parameter_utils.add_default_io_args(parser)
 
     args = parser.parse_args()
     utils.log("o3dAlign Args: ", args)
 
+    # TODO - handle molecules with multiple fragments
+    # TODO - allow to specify threshold as fraction of perfect score?
+
     qmol = rdkit_utils.read_single_molecule(args.query, index=args.qmolidx)
     qmol = Chem.RemoveHs(qmol)
     qmol2 = Chem.Mol(qmol)
@@ -107,26 +114,33 @@ def main():
     i=0
     count = 0
     total = 0
+    errors = 0
     for mol in suppl:
-        if mol is None: continue
-        if args.num > 0:
-            mol.RemoveAllConformers()
-            conformerProps, minEnergy = conformers.process_mol_conformers(mol, i, args.num, args.attempts, args.rmsd, None, None, 0)
-            mol = Chem.RemoveHs(mol)
-            count += doO3Dalign(i, mol, qmol, args.threshold, perfect_score, writer, conformerProps=conformerProps, minEnergy=minEnergy)
-        else:
-            mol = Chem.RemoveHs(mol)
-            count += doO3Dalign(i, mol, qmol, args.threshold, perfect_score, writer)
+        if mol is None:
+            i +=1
+            continue
+        try:
+            if args.num > 0:
+                mol.RemoveAllConformers()
+                conformerProps, minEnergy = conformers.process_mol_conformers(mol, i, args.num, args.attempts, args.rmsd, None, None, 0)
+                mol = Chem.RemoveHs(mol)
+                count += doO3Dalign(i, mol, qmol, args.threshold, perfect_score, writer, conformerProps=conformerProps, minEnergy=minEnergy)
+            else:
+                mol = Chem.RemoveHs(mol)
+                count += doO3Dalign(i, mol, qmol, args.threshold, perfect_score, writer)
+            total += mol.GetNumConformers()
+        except ValueError as e:
+            errors +=1
+            utils.log("Molecule", i, "failed to align:", e.message)
         i +=1
-        total += mol.GetNumConformers()
-    
+
     input.close()
     writer.flush()
     writer.close()
     output.close()
 
     if args.meta:
-        utils.write_metrics(output_base, {'__InputCount__':i, '__OutputCount__':count, 'RDKitO3DAlign':total})
+        utils.write_metrics(output_base, {'__InputCount__':i, '__OutputCount__':count, '__ErrorCount__':errors, 'RDKitO3DAlign':total})
 
 if __name__ == "__main__":
     main()