American tech company Google says it has taken an important step toward identifying disease-causing genes that could help identify rare genetic disorders.
A new model called AlphaMissense is able to accurately classify 89 percent of all possible missense variants in genes, indicating whether they are disease-causing or harmless.
Compared to human experts, only 0.1% of the missense variants were reliably classified.
Missense variants arise when a single letter in the DNA is changed, resulting in proteins with different amino acids. This small change can have significant effects as Google likened it to how changing one letter in a word can change the meaning of an entire sentence.
Most of these variants are harmless, and the average person has more than nine thousand variants, but some of them can be devastating, causing rare genetic diseases.
The new alpha missense looked at the current knowledge about the different missense variants and how commonly they are seen in humans and their closest relatives, the primates. He looked for variants that were rarely seen. By ranking them as pathogenic and using this information to analyze the sequences of other proteins to decide not only whether they are likely to cause problems, but Even how certain it was.
Human experiments to detect these mutations are expensive and slow because people need to test and design each unique protein separately. Google says the new system means researchers can observe the results for thousands of proteins at a time, helping them decide where to focus.
This section contains related reference points (Related Nodes field).
The company has used its systems to release an extensive catalog of missense mutations so that researchers can learn what effect they have. In some cases, these mutations can lead to diseases such as cystic fibrosis, sickle cell, anemia or cancer, so understanding them is key for researchers studying ways to treat or prevent these diseases.
This is the latest development in the field of health by Google’s ‘Deep Mind’ division to identify and treat a variety of medical conditions. Artificial intelligence uses The new system was built on the alphafold, an advanced model that helped probe proteins that are the building blocks of life.
The research is described in a new paper titled ‘Accurate proteome-wide missense variant effect prediction with AlphaMissense’, published in the journal Science. Google said the catalog is being made freely available to the research community and the company will also release the code behind the AI system.
#Google #artificial #intelligence #identify #genetic #diseases
