FoldaVirus, a knowledge-based icosahedral capsid prediction tool using AlphaFold

Abstract

Significance Currently, there exists a huge gap between the number of available viral coat protein (CP) sequences and the experimentally determined capsid structures. Moreover, there are no resources available that provide full capsid models. By leveraging the observation that CP tertiary structures and their capsid organization in spherical viruses are often well conserved within each virus family, we developed a knowledge-based approach that combines AlphaFold predictions of the CPs with the information on icosahedral capsid architectures (e.g., T = 1, 3, 4…) that they are likely to form based on the known structures. Furthermore, we developed a webtool (https://foldavirus.org) that is freely available to the public for generating the models of icosahedral capsids given the a.a. sequences of CPs.