Russian Scientists Upgrade AI to Predict Protein Mutations With Greater Accuracy

An enhanced version of AlphaFold2 could transform drug discovery and deepen our understanding of disease mechanisms.

Researchers at the Artificial Intelligence Research Institute (AIRI) in Russia have unveiled a major upgrade to AlphaFold2, the groundbreaking AI system for protein modeling. The modification dramatically improves the algorithm’s ability to predict how mutations affect protein stability—opening new possibilities for drug development and the study of genetic disorders.

The original AlphaFold2 struggled to accurately forecast the impact of single amino acid changes. The new version can now handle a wide range of mutations, including complex combinations. Its most important advance: the system no longer requires homologous protein data from evolutionary relatives. That means it can analyze entirely unique proteins with no natural analogues.

According to AIRI, this breakthrough could accelerate the creation of new medicines while offering deeper insights into how inherited and other diseases develop.