NLM DIR Seminar Schedule

Abstract:

Though typically associated with a single folded state, some globular proteins remodel their secondary and/or tertiary structures in response to cellular stimuli. AlphaFold2 (AF2) readily generates one dominant protein structure for these fold-switching (a.k.a. metamorphic) proteins, but it often fails to predict their alternative experimentally observed structures. Wayment-Steele, et al. steered AF2 to predict alternative structures of a few metamorphic proteins using a method they call AF-cluster. However, their paper lacks some essential controls needed to assess AF-cluster’s reliability. We find that using ColabFold-based random sequence sampling–a method we call CF-random–is a more accurate and less computationally intense alternative to AF-cluster. In addition, CF-random effectively captures the alternative conformations of functional and membrane transport proteins with fewer predicted samples than other AF2-based enhanced sampling approaches. We suggest that CF-random predicts the alternative conformations of proteins using associative sequence homology rather than generative coevolutionary inference.