Structural biology of single-stranded, positive-sense RNA viruses in the age of accurate atomic-scale predictions of protein structures

For decades atomic structures of proteins could only be determined experimentally and at a very slow pace. This was a particular problem for RNA viruses, for which sequences diverge fast and horizontal transfers are common. This made modeling from known structures difficult and uncertain. Only hard experimental structural data could allow accurate atomic descriptions of viral proteins and subsequent analyses, from mutant phenotype prediction to drug design. This has changed. With the advent of AlphaFold, that allows accurate protein structure prediction from sequence only, it is now possible in most cases to have the structure of a new protein of interest in a matter of minutes. In this mini review we focus on important consequences of this new state of affairs. While most of our conclusions are likely relevant to RNA viruses in general, here we focus on single-stranded, positive-sense RNA viruses. Taking as case studies proteins that are studied in our lab, we highlight why these viruses generally encode proteins that are particularly tough cases, being membrane-associated proteins with alternate conformations, structures, and interactions that may not be conserved even between close relatives. For these proteins AlphaFold may even fail or at least mislead, but with a proper approach it may also allow jump-starting the study of difficult or understudied viruses.