Google Deepmind tool could help identify disease-causing genes
BNN SCREENSHOT
Researchers at Google DeepMind, the artificial intelligence (AI) division of the US tech giant, have just introduced a groundbreaking tool to predict dangerous genetic mutations that could help research into rare diseases.
Google DeepMind vice president of research Pushmeet Kohli said the discovery was "another step in documenting the impact of AI on the natural sciences ," AFP reported on September 20.
The tool focuses on so-called “missense mutations,” in which a single letter of the genetic code is affected.
The average person has 9,000 such mutations in their entire genome. They can be harmless or cause diseases such as cystic fibrosis, cancer or damage brain development.
To date, about four million of these mutations have been observed in humans, but only 2% of them are classified as pathogenic or benign.
There are an estimated 71 million such mutations. Google DeepMind's AlphaMissense tool looked at these mutations and was able to predict 89% of them with 90% accuracy.
Results showed that 57% were classified as possibly benign and 32% as possibly pathogenic, while the remainder were uncertain.
The database has been made public and available to scientists, and an accompanying study has just been published in the journal Science . Experts say AlphaMissense shows “superior performance” compared to previous tools.
Google Deepmine expert Jun Cheng emphasizes that the predictions have never actually been used for clinical diagnostic purposes.
“However, we think our prediction could be useful to increase the rate of diagnosis of rare diseases and also potentially help us find new disease-causing genes,” the expert said, adding that the new tool could indirectly help develop new therapies.
Source link
Comment (0)