V- and VL-scores unveil viral signatures and origins of protein families

V-score and VL-score are annotation-free metrics that quantify the "virus-likeness" of protein families and genomes, released with an open-access searchable database, V-Score-Search. Applied to public protein databases they link 19-59% of protein families with viruses (a 5-8x increase over current estimates) and detect viral genomes, prophages and host-derived auxiliary viral genes in fragmented sequences more efficiently than existing approaches.