A Simple Method to Detect Candidate Overlapping Genes in Viruses Using Single Genome Sequences

Schlub, Timothy E; Buchmann, Jan P; Holmes, Edward C

The detection of 17 antisense ORFs is notable. Antisense overlaps have been shown to exist in a number viruses that use DNA as a replication intermediate including those in the Herpesviridae (Ward et al. 1996) , REP/ORF3 in Porcine Schlub et al. . doi:10.1093/molbev/msy155 MBE circovirus 2 (He et al. 2012 ) and HBZ/p12 and HBZ/p30 in Human T lymphotropic virus 1 (Arnold et al. 2006) . They are also suspected to occur in many more viruses with DNA intermediaries, including a long suspected antisense protein (asp) in HIV-1 (Torresilla et al. 2015; Cassan et al. 2016 ). In addition, they have been infrequently suggested to occur in RNA viruses that do not use DNA intermediates, such as a more than 100 amino acid (a) overlapping antisense hypothetical protein in Rice black streaked dwarf virus (dsRNA) (Zhang et al. 2001 ), a 96 aa overlapping antisense hypothetical protein in Lymphocytic Choriomeningitis Mammarenavirus (-ssRNA) (Salvato et al. 1989) , and a possible 167 aa overlapping antisense ORF called "NEG8" in human influenza A virus (Clifford et al. 2009; Sabath et al. 2011) . Our method can be used to investigate these further. For example, in the case of NEG8 we find that a 167 codon ORF on in the NEG8 reading frame (Àc2) is highly statistically unlikely by both the codon permutation and synonymous mutation methods, providing further evidence for a functional benefit of this ORF. Interestingly, however, a further frameshift of 1 nucleotide (frame Àc1) would make ORFs of such lengths much more likely (P ¼ 0.02 and 0.03 for codon permutation and synonymous mutation methods respectively), demonstrating the importance of the expected ORF lengths on every individual reading frame, rather than just the sense direction. Furthermore, when applying our method to HIV-1, we find that the possible antisense ORF (asp) is not substantially longer than expected by chance alone (P ¼ 0.06 and P ¼ 0.04 for codon permutation and synonymous mutation methods, respectively) in that reading frame.