Research

Our work combines graph learning, molecular representation learning, protein language models, and biomedical knowledge to study drug discovery and protein function.

AI-driven drug discovery

We develop predictive models for drug–target interaction (DTI) and drug–target affinity (DTA). A central goal is to integrate chemical structures, protein sequences and structures, interaction networks, and prior biological knowledge while improving generalization to new drugs and targets.

Drug-centered association prediction

We study associations among drugs, targets, diseases, phenotypes, and adverse effects. These models support drug repositioning, drug-combination research, mechanism discovery, and the prioritization of candidates for experimental validation.

Protein function and variant effects

We investigate computational approaches for protein-function annotation and missense-variant effect prediction. Our aim is to connect sequence and structural changes with molecular function, drug response, and disease relevance.

From prediction to validation

We are interested in collaborative workflows that connect virtual screening and AI-based prioritization with molecular experiments. This allows computational hypotheses to be evaluated and, in turn, enables experimental evidence to guide model development.