Jin Su

This paper introduces Pinal, a foundation model that designs proteins from natural language descriptions, achieving functional alignment and experimental success in four design tasks.

DOMINO generates multidomain proteins by learning domain co-occurrence patterns and using them to design novel sequences with high structural confidence.

The paper introduces SaprotHub, a platform that simplifies training and sharing of protein language models, enabling non-experts to build and collaborate on customized models.

ProTrek introduces a tri-modal contrastive learning model for protein sequence, structure, and function, achieving significant improvements in speed and accuracy over existing tools.

PRIME is a multi-agent system that dynamically synthesizes computational workflows for protein engineering, outperforming existing AI agents in complex tasks.

Evolla, a protein-language model, improves functional inference and discovers novel proteins through natural language queries and benchmarking.

Pairing each residue with its Foldseek 3D-structure token and pre-training an ESM-2-sized model on about 40 million AlphaFold2 structures beat ESM-2 and structure-based baselines on variant-effect and function tasks.

The paper compares sequence-based and structure-based methods for predicting protein function, finding that structure-based approaches can offer better accuracy when using high-quality protein structures.

This study shows that pre-trained protein language models can predict mutation pathogenicity and patient survival in cancer, outperforming existing methods.