Recently, the collaborative research endeavor ‘UniPert–G2CP’ by Tencent Life Sciences Lab and Central South University graced the pages of ‘Cell’, an internationally renowned and top-tier academic journal. This achievement represents a groundbreaking moment, as it is the first instance of AI-driven virtual cell research originating from China to be featured in this esteemed publication.

The algorithm introduces an innovative approach by mapping gene perturbations and chemical drug perturbations into a unified semantic space. This breakthrough effectively tackles the formidable challenge of ‘chemical perturbation × cell-specific response’. At its heart lies the core module, UniPert, which has been made openly accessible to the research community. Meanwhile, G2CP leverages a transfer learning strategy, involving pre-training on gene screening data and subsequent fine-tuning on chemical screening data.

The scope of this research is extensive, encompassing a vast array of elements: 4,994 genes, 7,860 compounds, and 5 cancer cell lines. Furthermore, it successfully achieved closed-loop validation, ranging from prediction to mechanistic explanation, in an ESR1 endocrine resistance case study.