Gene birth contributes to structural disorder encoded by overlapping genes

S. Willis; J. Masel

We artificially frameshifted 150 non-overlapping viral genes in those 80 species, and found higher ISD in the +2 reading frame (0.35 ± 0.02) than in the +1 reading frame (0.19 ± 0.01) for non-ORF controls, and an even more extreme difference for ORF controls (0.47 ± 0.02 vs. 0.18 ± 0.01). The artifact hypothesis predicts that the +1 and +2 members of the 92 verified overlapping gene pairs will follow suit. In agreement with this, the overlapping regions of genes in the +2 reading frame had higher mean ISD (0.48 ± 0.03) than those in the +1 reading frame (0.31 ± 0.02). While this provides strong evidence that frame shapes ISD as an artifact of the genetic code, average ISD across both ways of frameshifting nonoverlapping genes (0.27±0.01 and 0.22±0.02 for non-ORF and ORF frameshifted non-overlapping genes, respectively) is significantly lower than the ISD of all overlapping sequences (0.39 ± 0.02), showing that the artifact hypothesis cannot fully explain elevated ISD in the latter.