Decoding conformational heterogeneity across disordered proteomes

Abstract

Intrinsically disordered proteins (IDPs) comprise nearly one-third of the human proteome and play key roles in regulation, signaling, and disease, yet their dynamic nature has resisted accurate structural prediction even by the state of the art deep-learning methods. Here we introduce AI-IDP, a framework combining established deep-learning structure prediction of isolated short IDP fragments with their flexible physical restrains-aware assembly to transform sequence information into experiment-consistent conformational ensembles of disordered proteins. AI-IDP reproduces experimental observables across local, medium-range, and global scales, including transient secondary structure, mutation sensitivity, and overall chain dimensions. Applied to more than 3,000 disordered regions across human and non-human proteomes, AI-IDP reveals that transient -helices and polyproline-II conformations are pervasive and evolutionarily tuned features of disorder. By uncovering how sequence encodes conformational heterogeneity, AI-IDP provides a practical framework for understanding the structural and functional logic of disordered proteomes and enables rationally targeting the dynamic protein states that underlie health and disease.