Deep learning is becoming prominent in the realm of drug discovery. However, even though there is a significantly large, diverse dataset of protein binding, the models trained on these datasets tend to lack generalization. We expect that when a model is trained on a dataset that has different types of pf proteins and ligands, it should be able to generalize better. One goal of this project is to study this problem and discover a solution to tackle this.

Through this superUROP, I want to strengthen my research skills, expand my knowledge in the application of AI in medical science, and prepare myself for Ph.D. programs.