qsar-modeling

What problem does it solve?

This Skill automates the process of building QSAR/QSPR models, enabling the prediction of bioassay data, ADMET endpoints, and selectivity profiles.

Core Features & Use Cases

• QSAR/QSPR Modeling: Utilizes chemprop D-MPNN, MolFormer, Uni-Mol, ChemBERTa, random forest baselines, and Gaussian processes.
• Applicability Domain Handling: Implements OECD 5 principles, kNN, leverage, conformal prediction, Mahalanobis, and scaffold-balanced splits.
• Ensemble Uncertainty and Calibration: Incorporates Platt scaling, isotonic regression, and SHAP for feature importance.
• Use Case: For a pharmaceutical company developing new drug compounds, this Skill can predict the ADME properties of compounds based on their molecular structure.

Quick Start

Use the qsar-modeling skill to train a QSAR model on your bioassay data using chemprop 2.0 D-MPNN.