Google DeepMind has recently collaborated with Yale University to launch a new artificial intelligence model called C2S-Scale27B. This model is based on the open Gemma model series and focuses on complex single-cell analysis, successfully discovering a previously unknown cancer treatment pathway.
The main discovery of C2S-Scale27B is identifying the drug Silmitasertib (CX-4945) as a "conditional enhancer". This means that under specific conditions, this drug can make tumor cells more recognizable and eliminable by the immune system.
According to the DeepMind team, this achievement provides an important blueprint for new biological discoveries. They stated: "By following scale laws and building large models like C2S-Scale27B, we can create powerful cell behavior prediction models capable of performing high-throughput virtual screening, discovering biology under contextual conditions, and generating biology-based hypotheses."
After the model identified this potential therapeutic pathway, the research team conducted laboratory experiments using a human neuroendocrine cell model, successfully confirming the prediction. Before this, the C2S-Scale model had already simulated the effects of more than 4,000 drugs in two different immune environments.
The model's technology and research findings are now available to the public. The model code can be accessed on GitHub, and the model itself can be found on the Hugging Face platform. More detailed information has been published as a preprint on bioRxiv.
