graphein

Graphein is an open-source Python package designed to construct and analyze biological graphs from heterogeneous data sources. It provides a unified framework to fetch protein, structure, and interaction data and to convert them into network representations. Graphs created with graphein can represent entities such as proteins, residues, genes, and molecular interactions, with nodes and edges annotated by features like sequence information, structural properties, interaction confidence, and functional annotations.

The library supports multiple granularity levels, including residue-level graphs derived from three-dimensional structures, sequence-based neighborhood graphs, and large-scale interaction networks. It integrates with common graph and machine learning ecosystems, enabling export to formats compatible with NetworkX, PyTorch Geometric, and other libraries, and can be used with data pipelines that feed into graph neural networks or traditional network analysis workflows.

Key data sources include public biological databases and resources. Users can define graph-building workflows through configuration objects that specify data sources, content filters, and graph construction rules. Graphein emphasizes reproducibility and modularity, allowing researchers to compose reusable pipelines for tasks such as contact-map extraction, protein–protein interaction networks, and regulatory networks.

Installation is via Python package managers, and the project is maintained as an open-source effort with documentation and examples available on its repository. The library is used in computational biology, drug discovery, and systems biology to enable rapid generation of testable networks from diverse datasets.