"AlphaFold2 revolutionizes conformation prediction with sequence clustering"

Researchers have developed a method called AF-Cluster that uses sequence clustering to predict multiple conformations of proteins using AlphaFold2. This method was used to investigate the evolutionary distribution of predicted structures for the metamorphic protein KaiB and confirmed a surprising prediction about a cyanobacteria KaiB variant. The sensitivity of AF-Cluster to point mutations was also tested, and a putative alternate state was identified for the oxidoreductase Mpt53 in M. tuberculosis. This bioinformatic approach, in combination with experimental validation, has the potential to greatly impact the prediction of protein energy landscapes and understanding biological function.