Today, the most powerful approach to detect distant homologs of a protein is based on structure prediction and comparison. Yet this approach is still inapplicable to many viral proteins. Therefore, we developed a powerful sequence-based procedure to identify distant homologs of viral proteins. It relies on 3 main principles: 1) Traces of sequence similarity with a protein can persist beyond the significance cutoff of homology detection programs; 2) Candidate homologs can be identified among proteins with weak sequence similarity to the query, by using "contextual" information, e.g. taxonomy or type of host infected; 3) These candidate homologs can be validated using highly sensitive profile-profile comparison.As a test case, we applied our approach to a protein without known homologs, ORF4 of Lake Sinai virus (which infects bees). We discovered that ORF4 is composed of a domain that has homologs in proteins from >20 taxa of viruses infecting arthropods. We called it “Widespread, Intriguing, Versatile” (WIV) domain because it is found in proteins with a wide variety of domain organizations and functions. For example, WIV is encoded by the NSs protein of tospoviruses, a global threat to food security, which infect plants through arthropod vectors; by the protein encoded by RNA2 ORF1 of chronic bee paralysis virus, a widespread virus of bees; and by various proteins of cypoviruses, which infect the silkworm bombyx mori. WIV has a previously unknown structural fold, according to Alphafold predictions. In some viral species, WIV facilitates infection of arthropods, according to bibliographical evidence

30 years ago, researchers noticed that the capsid (VP1) gene of B19 parvovirus might encode a second protein, called "X", in an overlapping reading frame. Since then, experimental approaches failed to detect it. In contrast, sequence analyses can reliably predict whether a protein is expressed from an overlapping frame, provided that it is beneficial to the virus and thus under selection pressure. We used a dedicated software, Synplot2, to identify regions of VP1 likely to encode functional proteins in overlapping frames. Synplot2 detected the X open reading frame and confirmed it is under highly significant selection pressure. We discovered that the X protein is homologous to the ARF1 protein of human parvovirus 4, another suspected protein encoded in a frame overlapping VP1. These findings provide compelling evidence that the X protein must be expressed and functional. We predict that it contains a predicted transmembrane region. We found that the X frame contains a potential AUG start codon in parvovirus B19 and in all related species. Yet no currently known viral transcript has the potential to encode the X protein in a monocistronic fashion. Therefore, the X protein is probably expressed either from an unmapped monocistronic mRNA, or translated by a non-canonical mechanism from the VP1 mRNA or from a short transcript, R3, which has no currently known function. Finally, Synplot2 also detected proteins likely to be expressed from a frame overlapping VP1 in species distantly related to parvovirus B19: porcine parvovirus 2 and bovine parvovirus 3.