Research on AI4biology
- Time: from 2022~current
- Developed flexible protein docking using denoising and equivariant neural networks.
- Developed reinforcement learning framework for protein sequence design.
- Developed SOTA inverse-folding model via denoising training.
- Developed a new type of antibody-antigen complex prediction model.
- Developed multi-modal foundation model for protein representation learning.
- Developed a new type of protein representation learning framework focusing on surfaces.
- Developed a new type of protein generative model focusing on surfaces.