a-r-j/graphein

What it solves

Graphein is a library designed to bridge the gap between raw biomolecular data and geometric deep learning. It simplifies the process of converting complex biological structures—such as proteins, RNA, and molecular interaction networks—into graph representations that can be used as input for machine learning models.

How it works

The library provides a set of tools to transform biological data from various sources (like PDB codes, SMILES strings, or AlphaFold2 structures) into geometric representations. It supports multiple levels of granularity, including residue-level and atomic-level graphs, as well as protein meshes. It also handles the construction of interaction networks, such as protein-protein interaction (PPI) graphs and gene regulatory networks (GRN).

Who it’s for

It is primarily intended for researchers and developers working in bioinformatics, structural biology, and geometric deep learning who need to prepare biological data for training and evaluating AI models.

Highlights

• Multi-modal biological support: Handles proteins, RNA, and small molecules.
• Deep learning integration: Provides ready-to-go dataloaders for PyTorch Geometric.
• Flexible API: Offers both a programmatic Python API and a command-line interface (CLI) for batch processing.
• Diverse graph types: Supports residue-level graphs, atomic graphs, protein meshes, and interaction networks (PPI and GRN).