Key takeaways Google DeepMind and Yale University developed a 27-billion-parameter AI model that identified a novel drug combination to enhance cancer immunotherapy. The AI successfully predicted that...
Google DeepMind, in collaboration with Yale University, announced Wednesday the release of Cell2Sentence-Scale 27B (C2S-Scale), a groundbreaking artificial intelligence model built on the Gemma family of open models.
The 27-billion parameter foundation model is designed to analyze individual cells and has successfully generated and experimentally validated a novel hypothesis about cancer cellular behavior.
The discovery centers on one of the most challenging problems in cancer immunotherapy: how to convert "cold" tumors, those invisible to the body's immune system, into "hot" tumors that can be detected and attacked by immune cells.
The AI model identified the drug silmitasertib (CX-4945), a kinase inhibitor, as a potential conditional amplifier that could boost immune signaling specifically in environments where low levels of interferon, a key immune-signaling protein, are already present.
The C2S-Scale model simulated the effects of over 4,000 drugs across two distinct immune environments to identify compounds that would selectively enhance antigen presentation, the process by which cells display immune-triggering signals.
The model predicted that silmitasertib would produce a strong increase in antigen presentation only in immune-active settings, with minimal effect in neutral contexts.
Laboratory experiments at Yale confirmed the AI's prediction using human neuroendocrine cell models that were completely unseen by the model during training.
Treating cells with silmitasertib alone produced no effect on antigen presentation, and low-dose interferon alone had only a modest impact.
However, the combination of both treatments resulted in a synergistic amplification, increasing antigen presentation by roughly 50%.
According to Google's blog post announcing the discovery, the model was generating a genuinely novel hypothesis rather than simply identifying known biological relationships.
While the protein kinase CK2, which silmitasertib inhibits, has been implicated in various cellular functions, including immune system modulation, the drug had not previously been reported to explicitly enhance antigen presentation.
The research demonstrates what Google characterizes as an emergent capability of larger-scale AI models, the ability to perform conditional reasoning and generate experimentally testable hypotheses.
According to the company, this finding provides what it calls "a blueprint for a new kind of biological discovery," demonstrating that sufficiently large AI models can run high-throughput virtual drug screens and uncover context-dependent biological interactions.
Teams at Yale are now investigating the underlying mechanism of the immune system effect and testing additional AI-generated predictions in other immune contexts.
The model and accompanying tools have been made publicly available on Hugging Face and GitHub, with a scientific preprint posted on bioRxiv.
While the discovery represents a significant step forward in AI-assisted drug discovery, experts caution that it marks only the beginning of a lengthy process.
The results have not yet undergone peer review, and any therapeutic application would require years of additional preclinical research and clinical trials before the treatment could be made available to patients.
Silmitasertib itself is already in clinical trials for other cancer types, including cholangiocarcinoma and multiple myeloma, and received orphan drug status from the U.S. Food and Drug Administration in 2017 for advanced cholangiocarcinoma.
However, its potential use as an immune-conditional amplifier in combination with interferon represents an entirely new therapeutic application that would need independent validation.
The research builds on previous work published in April 2025, demonstrating that biological AI models follow scaling laws similar to those observed in natural language processing; larger models consistently perform better on biological tasks.
Key takeaways OpenAI has signed a multi-year, $10 billion agreement with AI chipmaker Cerebras Systems to secure computing infrastructure. The deal will deliver 750 megawatts of computing power throug...
KEY TAKEAWAYS: xAI implemented restrictions preventing Grok from editing images of real people in revealing clothing after global backlash California Attorney General Rob Bonta launched investigation...
Key takeaways TSMC reported fourth-quarter net income of NT$505.74 billion, beating analyst expectations and setting a new record. Fourth-quarter revenue reached NT$1.046 trillion, up 20.5% year over...
