alterlab-molfeat

What problem does it solve?

MolFeat provides a unified framework to convert molecular structures (SMILES or RDKit molecules) into machine-learning-ready feature vectors, enabling streamlined modeling and data-driven discovery.

Core Features & Use Cases

• Supports a wide range of featurizers including fingerprints (ECFP, MACCS, MAP4), descriptors (RDKit/Mordred), and pretrained transformer/GNN embeddings.
• Enables batched featurization via MoleculeTransformer and feature concatenation via FeatConcat, suitable for QSAR, virtual screening, and similarity search.
• Includes a ModelStore for discovering, loading, and comparing featurizers, and seamless integration with scikit-learn pipelines.
• Example use case: build a QSAR model on a medicinal chemistry dataset or perform large-scale virtual screening with a unified feature space.

Quick Start

Install MolFeat and run a basic featurization pipeline using a simple SMILES list to produce feature vectors.