Our group has introduced GerNA-Bind, a next-generation geometric deep learning platform for discovering small-molecule drugs targeting RNA. GerNA-Bind sets new state-of-the-art performance across multiple benchmarks and improves binding-site prediction accuracy by 20.8% compared with AlphaFold3. In a large-scale virtual screen, the platform identified 18 structurally diverse compounds targeting the oncogenic RNA MALAT1, with several showing submicromolar affinity. One lead molecule selectively binds the MALAT1 triple helix, reduces transcript levels, and inhibits cancer cell migration. These results highlight GerNA-Bind as a powerful tool for RNA-focused drug discovery, offering both high accuracy and meaningful biological insight.

Congratulations to our two talented research interns, Yunpeng (now a PhD student at SJTU) and Jiayi (a master’s student at Harvard)!