The Biohub Platform provides open tools and models for exploring protein biology, helping researchers predict, design, and discover proteins. Its offerings include the ESM family of protein models, an extensive protein database, interactive analysis tools, and an API for developers.

ESMC is a protein language model that learns representations from billions of evolutionary sequences to support functional analysis, structure prediction, and protein design. ESMFold2 predicts high-resolution, all-atom structures of biomolecular complexes from sequences, with optional multiple sequence alignment input to improve accuracy for challenging targets. ESM3 jointly models protein sequence, structure, and function, enabling users to generate novel proteins with controls based on these features.

The platform’s ESM Atlas offers functional annotations for more than 6.8 billion proteins and over 1.1 billion paired predicted structures. Users can search across proteins from all life on Earth by sequence or natural language to investigate structural and functional relationships. Additional tools support complex structure prediction from protein, DNA, and RNA sequences.