biodeg-ml

Raw codes for processing chemical biodegradability data using SMILES and RDKit parameters, used to predict ready biodegradability values using regression and GNN-based methods.

This is the cumulative product of research first started at IHPC in mid 2022. The notebook contains scripts and functions for:

• Loading and saving .sdf and .csv databases containing molecular conformers and SMILES/RDKit parameters respectively.
• Automated extraction of above-mentioend RDKit parameters as identified.
• Pre-processing for sk-learn models.
• Implementation of group-based cross-validataion based on common SMILES (for multiple conformers).

References:

1. Predicting ready biodegradability in the Japanese Ministry of International Trade and Industry test
2. Chemprop