DeepMind launched Alpha Missense, an AI tool, to predict harmful gene mutations in humans.

Thanks to CTOnews.com netizen Alejandro86 for the clue delivery! CTOnews.com, September 20 / PRNewswire-FirstCall-Asianet /-- Google's DeepMind recently launched an AI tool called AlphaMissense, which evaluates all 71 million "missense" mutations. CTOnews.com found that the tool has been on GitHub shelves, and the results have been published in the journal Science.

It is reported that human experts have so far found only 0.1% of the clinical effects of missense variants that change the structure of the body's main working molecules (proteins). It is reported that human experts have so far found only 0.1% of missense variants that change the structure of the body's main working molecules (proteins). In these mutations, a letter in the human genetic code has changed. Of these, 32% were classified as likely to cause disease, 57% were benign, and the rest were uncertain.

AlphaMissense predictions show mutations in the structure of two proteins, of which red is harmful, blue is benign and gray is uncertain.

"previous experiments to detect disease-causing mutations were expensive and laborious, each protein was unique, and each experiment had to be designed separately, which could take months," said iga Avsec iga Avsec, a project researcher at ▲ 's London-based paper. Now by using AI predictions, researchers can preview the results of thousands of proteins at a time, which helps to prioritize resources and accelerate more complex research. "

At the same time, iga Avsec said that the current AlphaMissense also has limitations, and the tool's prediction of pathogenicity does not tell researchers about the "biophysical properties of the mutants," but he remains optimistic about the future of the tool.

"these AI predictions were never really intended for clinical diagnosis alone, and they should always be used in conjunction with other evidence, but these predictions will still help improve the diagnosis rate of rare diseases and help researchers find new pathogenic genes," said Cheng Jun, a researcher with the project. "

Referenc

Accurate proteome-wide missense variant effect prediction with AlphaMissense

Deepmind/alphamissense-GitHub

DeepMind's New AI Can Predict Genetic Diseases