Google AI May Help Scientists To Discover Promising Therapy To Fight Cancer

Google's artificial intelligence, build on its Gemma family of models, developed novel cancer cellular path, which may lead to a potential cancer therapy pathway, Google Inc said. CEO Sundar Pichai called it a milestone for AI in science.

Google Inc collaborated with Yale University and used one of the AI built on its Gemma models for the breakthrough.

Google built with Yale University a foundational model, Cell2Sentence-Scale 27B (C2S-Scale), based on Gemma, generated a novel hypothesis about cencer cellular behaviour, according Pichai's post on X.

Scientists experimentally validated in livin cells. More pre-clinical and clinical tests may lead to a discovery of developing therapies to fight cancer, Pichai said.

C2S-Scale represents a new innovation in single-cell analysis, according to Google Inc. This breakthrough follows earlier work, where the collaborator found that biological models follow clear scaling laws — just like with natural language, larger models perform better on biology.

A major challenge in cancer immunotherapy is that many tumors are cold or invisible to the body's immune system. A key strategy to make them hot is by forcing them to display immune-triggering signals through a process called antigen presentation, Google's article said.

C2S-Scale 27B model was given a task to find a drug that acts as a conditional amplifier. It would boost the immune signal only in specific 'immune-context-positive' environment where low levels of interferon were already present, but inadequate to induce antigen presentation on their own, according to the article.

Smaller models could not resolve the context-dependent effect, as this requires a level of conditional reasoning that appeared to be an emergent capability effect, according to the article.

Next, they designed a dual-context virtual screen to find this specific synergistic effect. The virtual screen involved two stages such as immune-context positive and immune-context neutral.

Next, the Gemma-based AI model simulated the effect of over 4,000 drugs across both contexts and asked the model to predict which drugs would only boost antigen presentation in the first context, to bias the screen towards the patient relevant setting, the article said.

A fraction of drug hits are already known in prior literature, while remaining drugs are surprising hits with no prior known link to the screen, the article said. Although CK2 has been implicated in many cellular functions, including 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, the article said.