Merge pull request #118 from rdkit-rs/validate_standardized_mols

Validate standardized molecules

src/search/compound_processing.rs

@@ -152,6 +152,10 @@ pub fn standardize_mol(romol: &ROMol) -> eyre::Result<ROMol> {
     let te = TautomerEnumerator::new();
     let canon_taut = te.canonicalize(&parent_rwmol.to_ro_mol())?;
     let neutralized_canon = neutralize_atoms(&canon_taut)?;
+
+    // Validate
+    let _ = ROMol::from_smiles(&neutralized_canon.as_smiles())
+        .map_err(|_| eyre::eyre!("Canonicalization failed validation"))?;
     Ok(neutralized_canon)
 }
 

tests/cpd_processing_tests.rs

@@ -99,6 +99,17 @@ fn test_standardize_bad_smiles() {
     assert!(standardize_smiles(smiles, false).is_err());
 }
 
+#[test]
+fn test_bad_standardization() {
+    // The smiles below is technically a perfectly valid smiles
+    // but at the moment our standardization procedure can mess up isotopic hydrogens.
+    // This is a rare occurrence so it's maybe not worth fixing these cases specifically just yet.
+    // That said, we should at least force an error in these cases.
+    let smiles = "O=C(O[2H])C(F)(F)F";
+    let result = standardize_smiles(smiles, false);
+    assert!(result.is_err());
+}
+
 #[test]
 fn test_get_tautomers() {
     let smiles = "Oc1c(cccc3)c3nc2ccncc12";