Gene birth contributes to structural disorder encoded by overlapping genes

S. Willis; J. Masel

We used a multiple regression approach to determine which factors predict ISD values [44] . Gene whose ages could be classified, from non-overlapping genes in the species in which those 47 overlapping gene pairs were found, and from the 27 available pre-overlapping ancestral homologs. Some numbers in the main text are for the full set of 92 overlapping gene pairs and might not match exactly. (B) While frame significantly impacts disorder content (green is consistently higher than blue), it does not drive the high ISD of overlapping genes. Even controlling for frame, novel ISD > ancestral (left), supporting birth facilitation. Supporting conflict resolution, ancestral > non-overlapping either in-frame (yellow) or matched frameshifted control. The preadaptation version of the constraint hypothesis is supported by the fact that non-overlapping ISD < pre-overlapping (compare the two yellow bars). Means and 66% confidence intervals were calculated from the Box-Cox transformed (with λ = 0.4) means and their standard errors, and are shown here following back-transformation. designation (ancestral vs. novel vs. one or more nongenic controls) and relative reading frame (+1 vs. +2) were used as fixed effects. Homologous sequences are not independent; we accounted for this by using a linear mixed model [34] , with homology group as a random effect. Species is a stand-in for a number of confounding factors, e.g. %GC content, and so was included as a second random effect. The data were normalized using a Box-Cox transformation prior to analysis. Pairwise comparisons discussed throughout the Results were performed using contrast statements applied to the linear model, using the minimum number of gene designations necessary to make the comparison in question.