This repository contains the model and data files for predicting solvent sustainability based on the GlaxoSmithKline (GSK) Solvent Sustainability Guide. Solvent sustainability is quantified using the Composite Greenness Score (G-Score), which incorporates Environment, Health, Safety, and Waste (EHSW) considerations and Life Cycle Assessment (LCA) factors.
- GSK_dataset.csv: This csv file contains the GSK-based Composite Greenness Score (G-score) data for 154 solvents.
- Greener_Replacement_Candidate_Sets: A folder containing greener replacements for the target undesirable solvent, ranked based on Hansen Solubility Parameter (HSP) based Relative Energy Difference (RED) along with Gaussian Process Regression ML predicted G-scores. A lower RED value indicates greater similarity to the target solvent, and a higher G-score indicates a more sustainable solvent.
- Each alternative is labeled as “high-fidelity RED” (based on validated HSP dataset) or “low-fidelity RED” (based on estimated HSP dataset), serving as an indicator of HSP-derived confidence and to caution against potential inaccuracies associated with lower-fidelity data.
- GreenSolventDB: A folder containing machine learning-predicted Greenness Scores (G-scores).
Please cite: Datta, R., Nistane, J., Sose, A., Sahu, H., & Ramprasad, R. Machine Learning for Green Solvents: Assessment, Selection and Substitution. Advanced Science, e16851 (2025). https://doi.org/10.1002/advs.202516851
All recipients must adhere to the terms and conditions outlined in the GTRC Academic Research Use License.
