Research
Computational methods that connect sequence to biological function.
Protein function prediction
S2DA-GO predicts Gene Ontology functions from protein sequence. A global sequence branch is paired with a gradient-decoupled cross-attention branch and with semantic priors from GO term text, so rare functional labels interfere less with the shared representation.
Mutation effect prediction
SpaSem (Spatial–Semantic Integration) is a dual-stream model for zero-shot prediction of protein mutation effects. It fuses geometric structure with longer-range evolutionary signal so variants can be scored without building a new structure for every mutant. A formal citation has not been added yet.
Placeholder
[Placeholder] Next research direction
Replace this card in _data/research.yml when the lab wants another public direction listed. Remove the entry if you do not need a spare card.
Research cards live in _data/research.yml. Set home: true to feature a card on the home page.