Direct prediction of intrinsically disordered protein conformational properties from sequence

Abstract

Intrinsically disordered regions (IDRs) are ubiquitous across all domains of life and play a range of functional roles. While folded domains are generally well-described by a single 3D structure, IDRs exist in a collection of interconverting states known as an ensemble. This structural heterogeneity means IDRs are largely absent from the PDB, contributing to a lack of computational approaches to predict ensemble conformational properties from sequence. Here we combine rational sequence design, large-scale molecular simulations, and deep learning to develop ALBATROSS, a deep learning model for predicting IDR ensemble dimensions from sequence. ALBATROSS enables the instantaneous prediction of ensemble average properties at proteome-wide scale. ALBATROSS is lightweight, easy-to-use, and accessible as both a locally installable software package and a point-and-click style interface in the cloud. We first demonstrate the applicability of our predictors by examining the generalizability of sequence-ensemble relationships in IDRs. Then, we leverage the high-throughput nature of ALBATROSS to characterize emergent biophysical behavior of IDRs within and between proteomes. Update from previous versionO_LIThis preprint reports an updated version of the ALBATROSS network weights trained on simulations of over 42,000 sequences. C_LIO_LIIn addition, we provide new colab notebooks that enable proteome-wide IDR prediction and annotation in minutes. C_LIO_LIAll conclusions and observations made in versions 1 and 2 of this manuscript remain true and robust. C_LI