ColabFold predicts alternative protein structures from single sequences, coevolution unnecessary for AF-cluster

Lauren L Porter; Devlina Chakravarty; Joseph W Schafer; Ethan A Chen

doi:10.1101/2023.11.21.567977

ColabFold predicts alternative protein structures from single sequences, coevolution unnecessary for AF-cluster

bioRxiv [Preprint]. 2023 Nov 25:2023.11.21.567977. doi: 10.1101/2023.11.21.567977.

Authors

Lauren L Porter^{1

2}, Devlina Chakravarty¹, Joseph W Schafer¹, Ethan A Chen¹

Affiliations

¹ National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Bethesda, MD 20894.
² Biochemistry and Biophysics Center, National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, MD, 20892.

Abstract

Though typically associated with a single folded state, globular proteins are dynamic and often assume alternative or transient structures important for their functions^1,2. Wayment-Steele, et al. steered ColabFold³ to predict alternative structures of several proteins using a method they call AF-cluster⁴. They propose that AF-cluster "enables ColabFold to sample alternate states of known metamorphic proteins with high confidence" by first clustering multiple sequence alignments (MSAs) in a way that "deconvolves" coevolutionary information specific to different conformations and then using these clusters as input for ColabFold. Contrary to this Coevolution Assumption, clustered MSAs are not needed to make these predictions. Rather, these alternative structures can be predicted from single sequences and/or sequence similarity, indicating that coevolutionary information is unnecessary for predictive success and may not be used at all. These results suggest that AF-cluster's predictive scope is likely limited to sequences with distinct-yet-homologous structures within ColabFold's training set.

Publication types

Preprint