deal with wrong molecule inputs, set to None

ersilia/io/readers/file.py

@@ -189,6 +189,9 @@ def resolve_columns(self):
  reader = csv.reader(f, delimiter=self._column_delimiter)
  N = 0
  for i, r in enumerate(reader):
+ if len(r) == 1:
+ self.matching = {"input": [0], "key": None}
+ return self.matching
  if i > self.sniff_line_limit:
  self.logger.debug("Stopping sniffer for resolving column types")
  break

ersilia/utils/identifiers/compound.py

@@ -23,6 +23,7 @@ def __init__(self, local=True):
  else:
  self.Chem = None
  self.unichem = unichem
+ self.default_type = "smiles"
 
  def _is_smiles(self, text):
  if self.Chem is None:
@@ -53,6 +54,8 @@ def _is_inchikey(text):
  return True
 
  def guess_type(self, text):
+ if text is None:
+ return self.default_type
  if self._is_inchikey(text):
  return "inchikey"
  if self._is_smiles(text):
@@ -111,14 +114,17 @@ def encode(self, smiles):
  if inchikey is None:
  inchikey = self._nci_smiles_to_inchikey(smiles)
  else:
- mol = self.Chem.MolFromSmiles(smiles)
- if mol is None:
- raise Exception(
- "The SMILES string: %s is not valid or could not be converted to an InChIKey"
- % smiles
- )
- inchi = self.Chem.rdinchi.MolToInchi(mol)[0]
- if inchi is None:
- raise Exception("Could not obtain InChI")
- inchikey = self.Chem.rdinchi.InchiToInchiKey(inchi)
+ try:
+ mol = self.Chem.MolFromSmiles(smiles)
+ if mol is None:
+ raise Exception(
+ "The SMILES string: %s is not valid or could not be converted to an InChIKey"
+ % smiles
+ )
+ inchi = self.Chem.rdinchi.MolToInchi(mol)[0]
+ if inchi is None:
+ raise Exception("Could not obtain InChI")
+ inchikey = self.Chem.rdinchi.InchiToInchiKey(inchi)
+ except:
+ inchikey = None
  return inchikey