the company wrote<\/a> in a blog post today.<\/p>\nThe finding centers on one of the hardest problems in cancer immunotherapy: how to make so-called cold tumors, which are invisible to the immune system, more hot and thus more responsive to treatment. DeepMind said its model successfully identified a conditional amplifier drug that could boost immune visibility only in certain biological contexts.<\/p>\n
\ufeff<\/p>\n
To test the idea, C2S-Scale analyzed patient tumor data and simulated the effects of more than 4,000 drug candidates under two conditions: one where immune signaling was active and one where it was not. The model predicted that silmitasertib (CX-4945), a kinase CK2 inhibitor, would dramatically increase antigen presentation\u2014a key immune trigger\u2014but only in the immune-active setting.<\/p>\n
\u201cWhat made this prediction so exciting was that it was a novel idea,\u201d Google wrote. “Although CK2 has been implicated in many cellular functions, including as a modulator of the immune system, inhibiting CK2 via silmitasertib has not been reported in the literature to explicitly enhance MHC-I expression or antigen presentation. This highlights that the model was generating a new, testable hypothesis, and not just repeating known facts.”<\/p>\n
Laboratory experiments confirmed the prediction. When human neuroendocrine cells were treated with both silmitasertib and low-dose interferon, antigen presentation rose by roughly 50 percent, effectively making the tumor cells more visible to the immune system.<\/p>\n
DeepMind researchers described the discovery as evidence that scaling up biological AI models doesn\u2019t just improve accuracy\u2014it can produce entirely new hypotheses. \u201cThe true promise of scaling lies in the creation of new ideas, and the discovery of the unknown,\u201d the post said.<\/p>\n
Teams at Yale are now probing the mechanism that underlies this immune-system effect and testing other AI-generated predictions. DeepMind said the work \u201cprovides a blueprint for a new kind of biological discovery,\u201d one that uses large-scale AI systems to run virtual drug screens and propose biologically grounded hypotheses for lab testing.<\/p>\n
The model and accompanying tools are publicly available on Hugging Face<\/a> and GitHub<\/a>, with a scientific preprint posted on bioRxiv<\/a>.<\/p>\nStill, experts caution that such findings represent only the first step in a long process. The results have yet to undergo peer review or clinical validation, and any therapeutic application would require years of additional research and trials.<\/p>\n
Generally Intelligent Newsletter<\/p>\n
A weekly AI journey narrated by Gen, a generative AI model.<\/p>\n