Strand-specific compositional asymmetries in double-stranded DNA viruses

Analysis of 22 complete sequences of double-stranded DNA viruses reveals striking compositional asymmetries between leading and lagging, and between transcribed and non-transcribed strands. In all bi-directionally replicated genomes analyzed, the observed leading strand GC skew (measuring relative excess of guanines versus cytosines) is different from that in the lagging strand. In most of these genomes GC skew switches polarity close to replication origins. GC skew changes linearly across adenovirus linear genomes, which replicate from one end. In papillomavirus, GC skew is positive in one half of the genome where transcription and replication proceed in the same direction, and is close to zero in the other half with divergent transcription and replication. Possible contributions of these two processes (and associated repair mechanisms) as well as other potential sources of strand bias in the observed asymmetries are discussed. Use of cumulative skew plots for genome comparisons is demonstrated on the example of herpes simplex virus.