NCBS develops Disobind AI to predict binding of intrinsically disordered proteins

Exam-focused facts from the 19 January 2026 current affairs briefing.

Key facts

Researchers at the National Centre for Biological Sciences (NCBS) have developed an open-source deep-learning model named Disobind that predicts how intrinsically disordered proteins or regions bind to specific partners.
Disobind uses protein language models trained on millions of known protein sequences and considers both proteins in an interaction, achieving higher accuracy than leading predictors including AlphaFold-based multimer models.
The tool applies to immune signalling pathways and proteins involved in cancer and neurodegenerative disorders, aiding disease research, drug design and understanding of complex biological systems.