Jiahua Rao

By unifying contrastive learning with pose-conditioned generative modeling, BindCLIP produces interaction-aware embeddings that substantially improve virtual screening, especially in challenging out-of-distribution scenarios.