A new AI tool from NCBS can accurately predict how flexible, shapeshifting proteins interact, opening new avenues in disease research and drug discovery.

Researchers at the National Centre for Biological Sciences (NCBS), Bengaluru, have developed a new artificial intelligence tool that predicts how shapeshifting proteins interact inside cells, a long-standing challenge in biology.

The deep-learning model, called Disobind, focuses on intrinsically disordered proteins (IDPs). These flexible molecules lack a fixed structure but play a central role in cellular signalling, gene regulation, and disease processes. By analysing protein sequences using advanced protein language models, Disobind identifies likely binding regions without relying on structural data. Tested against leading tools such as AlphaFold-multimer, it showed higher accuracy, particularly for previously unseen protein pairs. Scientists stated that this open-source tool could aid disease research and drug design.

(Source:https://www.thehindu.com/news/cities/bangalore/researchers-develop-ai-tool-to-predict-how-shapeshifting-proteins-connect-inside-cells/article70509429.ece)