Virus structure predictions

Predicting the structure of viral proteins opens up a new field of research for the comparative genomics of viruses. The reason for this is that the folding of many viral enzymes or virion structural proteins is restricted, while the sequence itself is often so diverse that it cannot be aligned. New tools for structure prediction and analysis are shedding light on the dark proteome of viruses.

Structure prediction model

NametypealgorithmnotesPublication
AlphaFold2ColabAlignementMay 2022
ESMfoldWeb resourceProtein language modelMax length of 400 aaMarch 2023
AlphaFold3Web resourceMix alignement/ PLM??licence restrictions, max 20 models/dayJune 2024

More on structure prediction software: PDB

Tools

NameTypeMethodPublication
FoldseekStructure alignment searchStructural alphabetFeb 2024
FoldMasonStructural phylogeneticsStructural alphabetAug 2024
FoldtreeStructural phylogeneticsStructural alphabet + sequence MSAOct 2023*


Virus model database

Structure databaseDatasetprediction modelproteinsvirusessizePublication
Norburg's Virome at ModelArchiveRefseqAlphaFold267,715 4,463 Eukaryotic<1500 aaSept 2024
BFVDUniref30AlphaFold2351,242All<1500 aaJan 2025
Viro3DVMR ICTVAlphaFold2 and ESMfold85,0004,000 human and animal2025



Norburg's virome predictions in ViralZone

47841 predicted structures are available in the ViralZone datasheets via the "3D Models" tab. All available structures are sorted by virus. For each structure, a link to the corresponding UniProt entry shows the name of the protein, and a link to the ModelArchive provides access to the structure data. The pop-up window that appears when hovering over the links allows you to preview the structure (ModelArchive) or the UniProt entry displaying a PDB structure, if available. image