Yang Tan

VenusGenome improves microbial genome representation learning by using a genome-aware masked autoencoder, achieving high performance in phenotype prediction and gene-level tasks.

This paper introduces a pre-training framework that combines sequential and geometric encoders to improve protein variant prediction for enhanced bioactivity and thermostability.

This paper introduces PROEDIT, a method to remove virus-related knowledge from pre-trained protein models to improve safety in mutation effect predictions without affecting performance on non-virus proteins.

This study introduces a lightweight graph neural network for protein engineering that predicts fitness and guides the design of effective mutants with improved physicochemical properties.

ProSST integrates protein sequences and structures into a Transformer-based model, achieving state-of-the-art performance in mutation effect prediction and downstream tasks.

The paper investigates the onset of anharmonicity in a protein and its surface water around 200 K, finding they are decoupled, with the protein's transition being intrinsic and the water's being a resolution effect.