Google AI predicts the danger of millions of genetic mutations

September 19, 2023 Today at 8:32 p.m.

Google DeepMind researchers have developed AlphaMissense, a tool that can predict the pathogenicity of genetic mutations, focusing on “nonsense” mutations.

Google DeepMind presented an artificial intelligence tool, AlphaMissense, capable of predicting the pathogenic nature of genetic mutations, thus promoting research into rare diseases. Analyzes “nonsense” mutations that affect the genetic code.

The tool studied millions of mutations, and classified 57% as probably benign and 32% as pathogens. Although the tool does not directly determine the disease caused, it can help in the diagnosis and discovery of genes related to the disease. An individual has approximately 9,000 of these mutationsMostly benign, but some can cause diseases, for example cystic fibrosis.

To date, four million of these mutations have already been observed in humans, but only 2% of them have been classified as pathogenic or benign. And in total there are 71 million of these types of mutations possible. They were reviewed by the Google DeepMind tool, called AlphaMissense, which managed decide on 89% among them.

The impact of AI on science

This is “a new step in the recognition of the impact that artificial intelligence has on natural sciences,” said Ziga Avsec, vice president of research at Google DeepMind, in a press conference. This innovation was also praised for its “superior performance” in Science magazine.

Read also

Jakob Uszkoreit, the AI ​​pioneer who wants to revolutionize biotechnology

Indirectly, this could lead to development of new treatmentsestimate the researchers, who warn, however, that AlphaMissense should not be used alone to make a diagnosis.

AlphaMissense was trained with human and primate DNA data and is based on AlphaFold, a previous DeepMind program.