Gene birth contributes to structural disorder encoded by overlapping genes

S. Willis; J. Masel

Non-overlapping gene ISD is not statistically different from the mean of the +1 and +2 artificiallyframeshifted control versions of the same nonoverlapping nucleotide sequences (p = 0.88; contrast statement applied to a linear model with fixed effect of actual non-overlapping gene sequence vs. +1 artificially-frameshifted version vs +2 artificiallyframeshifted version, and gene identity as a random effect). In other words, despite the enormous effect of +1 vs. +2 reading frame, we find no support for the artifact hypothesis in explaining the elevated ISD of overlapping regions. In each overlapping gene pair, there is always exactly one gene in each of the two reading frames, such that the large effects of each of the two frames cancel each other out when all overlapping genes are considered together. It is nevertheless important to control for the large effect of frame while testing and quantifying other hypotheses.