Gene birth contributes to structural disorder encoded by overlapping genes

S. Willis; J. Masel

We find stronger support for the birth-facilitation hypothesis. Of the 92 verified overlapping viral gene pairs, we were able to classify the relative ages of the component genes as ancestral vs. novel for 47 pairs (Table I ). In agreement with the predictions of the birthfacilitation process, and controlling for frame, novel genes have higher ISD than either ancestral members of the same gene pairs or artificially-frameshifted controls ( Figure 4B ). We confirmed this using a linear mixed model, with frame (+1 vs. +2) as a fixed effect, gene type (novel vs. ancestral) as a fixed effect, species (to control for %GC content and other subtle sequence biases) as a random effect, and homology group (to control for phylogenetic confounding) as a random effect. Within this linear model, the prediction unique to the birth-facilitation hypothesis, namely that ISD in the overlapping regions of novel genes is higher than that in ancestral genes, is supported with p = 0.03. Inspection of Figure 4B suggests that elevation of novel gene ISD above ancestral is specific to the +2 frame; this is confirmed in the analysis of Figure 5B . Running separate statistical models for the two frames, the +2 frame difference is supported with p = 0.006.