jc32173 / cheminfo_utils

Useful general chemoinformatics functions

cheminfo_utils

Set of useful chemoinformatics functions including curating/filtering datasets, manipulating/modifying molecules and calculating descriptors. Most of the functions rely on RDKit, but some also use OpenEye which requires a license.

Dataset curation and analysis

• apply_lilly_rules.py: Wrapper to call Lilly_Medchem_Rules.rb (https://github.com/IanAWatson/Lilly-Medchem-Rules) from python to apply Lilly rules to a list of SMILES and return the results in a DataFrame.

• clustering.py: Methods to cluster molecules by structures, including:

  • Butina clustering.

• dataset_curation.py: Set of functions for analysing a dataset of molecules, including calculating common representations and descriptors and identifying possible duplicates (including tautomers and stereoisomers).

Processing molecules for ML

• calc_desc.py: Calculate molecular descriptors and fingerprints.

• rdkit_extra_desc.py: Extra molecular descriptors calculated using RDKit, including function to get number and size of fused ring systems (GetFusedRings).

• smi_funcs.py: Code for manipulating SMILES.

  • Canonicalising SMILES
  • Canonicalising tautomers
  • pH conversion using OpenEye or OpenBabel

Modifying molecules

• brics_fns.py: Useful functions for fragmenting molecules using BRICS (see: https://chemistry-europe.onlinelibrary.wiley.com/doi/abs/10.1002/cmdc.200800178) and processing the resulting fragments.

• join_mols.py: Join two SMILES with attachment points with matching labels.

Molecular structure

• fn_grps.py: Various functions for identifying functional groups next to a given substructure within molecules. Includes Ertl's definition (https://jcheminf.biomedcentral.com/articles/10.1186/s13321-017-0225-z).

• struct_fns.py: Code for calculating structural features of molecules, including:

  • The minimum through bond distance between atoms or substructures within a molecule.