PoET-2: Understanding protein function with a multimodal retrieval-augmented foundation model
Tuesday October 28th, 4:00-5:00pm EST | Tristan Bepler, PhD (OpenProtein.AI)
Abstract: Protein language models (PLMs) learn probability distributions over natural protein sequences. By learning from hundreds of millions of natural protein sequences, protein understanding and design capabilities emerge. Recent works have shown that scaling these models improves structure prediction, but does not seem to improve mutation understanding and representation quality for protein function prediction. We introduce PoET-2, a multimodal, retrieval-augmented protein foundation model that incorporates in-context learning of family-specific evolutionary constraints with optional structure conditioning to learn generative distributions over protein sequences. PoET-2 uses a hierarchical transformer encoder that is equivariant to sequence context ordering and a dual decoder architecture with both causal and masked language modeling objectives, allowing PoET-2 to operate in both fully generative and bidirectional representation learning modes. PoET-2 achieves state-of-the-art performance on zero-shot variant effect prediction, excelling at scoring variants with multiple mutations and challenging indel mutations. In supervised settings, PoET-2 embeddings outperform previous methods for learning sequence-function relationships, especially with small datasets. This work highlights the benefits of combining retrieval augmentation with multimodal, family-centric modeling for advancing protein foundation models.
Tristan is a machine learning scientist, group leader of the Simons Machine Learning Center at the New York Structural Biology Center, and CEO and co-founder of OpenProtein.AI. He received his PhD from MIT in the Computational and Systems Biology Program. Before starting OpenProtein.AI, Tristan pioneered large language models for learning protein sequence representations and their application to protein property prediction. He is also passionate about machine learning methods for understanding protein structures in native states and accelerating cryo-EM.