Leveraging sequence purification for accurate prediction of multiple conformational states with AlphaFold2
In the authors' words
Abstract AlphaFold2 (AF2) has transformed protein structure prediction by harnessing co-evolutionary constraints embedded in multiple sequence alignments (MSAs). MSAs not only encode static structural information, but also contain evolutionary signatures linked to protein dynamics and conformational heterogeneity, which are central to biological function. However, the subtle co-evolutionary signals that bias proteins toward distinct conformational states are often obscured by noise within MSA data and remain challenging to decipher. Here, we introduce AF-ClaSeq, a systematic framework that enriches and purifies state-specific evolutionary signals through iterative sequence classification and selection. By identifying sequence subsets that preferentially encode distinct conformational states, AF-ClaSeq enables robust predictions of alternative conformations across diverse protein systems. Our findings indicate that the accuracy of alternative-state prediction depends more on the purity of the evolutionary signal within an MSA than on sequence depth. Notably, purified sequences encoding specific structural states are distributed across phylogenetic clades and superfamilies, rather than being confined to individual lineages. By extending AF2 beyond single-state structure prediction, AF-ClaSeq provides a promising approach for exploring protein structural plasticity, establishing a foundation for future studies of the sequence determinants underlying conformational switching and allosteric regulation.
Appeared: Wednesday, November 18. Nature Communications. Peer-reviewed journal.