Conformational ensembles of the human intrinsically disordered proteome

Intrinsically disordered regions (IDRs) constitute about one third of the human proteome and play important roles in biological processes. While lacking well-defined 3D structures, IDRs adopt highly diverse conformational ensembles which are determined by the amino acid sequence, yet not captured by existing structure prediction methods. I will first present CALVADOS, an efficient molecular model of IDRs which we trained via a data-driven parameter-learning procedure and tested against experimental data on both conformational and phase properties. Second, I will describe how we investigated sequence-ensemble-function relationships of IDRs by constructing conformational ensembles of 28,058 human IDRs using CALVADOS. Through bioinformatics analyses, we explored the relationship between the compaction of the IDRs and the biological function and cellular localization of the proteins. Further, we gained insights into how sequence features relate to chain compaction and analyzed the evolutionary conservation of conformational properties across homologous IDRs. Our work exemplifies how the proteome-wide and freely-available database of conformational ensembles which we generated can be used to formulate hypotheses on the biological role and evolution of human IDRs.