A deep learning method for predicting interactions for intrinsically disordered regions of proteins

Intrinsically disordered proteins or regions (IDPs or IDRs) exist as ensembles of conformations in the monomeric state. Upon binding to a partner protein, they can adopt a variety of binding modes, ranging from becoming ordered upon binding, to binding in a multivalent manner, to remaining heterogeneous or fuzzy in the bound state. Moreover, the same IDR can adopt a different binding mode, depending on the partner. Characterizing the interfaces of IDRs in complexes is challenging, both experimentally and computationally. Here, we developed Disobind, a deep-learning method that predicts inter-protein contact maps and interface residues for an IDR and a partner protein, given their sequences. Most current methods, in contrast, are agnostic to the partner, require the structure of either protein, and/or are limited by the quality of multiple sequence alignments. Disobind performs better than AlphaFold-multimer and AlphaFold3. Combining the Disobind and AlphaFold-multimer predictions further improves the performance. We demonstrate the use of Disobind on a prion protein complex and a chemokine-receptor complex. The predictions can be used to localize IDRs in integrative structures of large assemblies, characterize protein-protein interactions involving IDRs, and modulate IDR-mediated interactions.