Gene birth contributes to structural disorder encoded by overlapping genes

S. Willis; J. Masel

Within each overlapping pair, we provisionally classified the gene with the higher CAI value as ancestral and the gene with lower CAI value as novel. We then compared the two sets of relative adaptedness valuesusing the wilcox.test function in R to perform a Mann-Whitney U Test. We chose a p-value cutoff of 0.035 after analyzing a receiver operating characteristic (ROC) plot ( Figure 2A ). The combined effects of our length threshold and p-value cutoff are illustrated in Figure 2B . In total, 27 gene pairs were determined to have statistically-significant CAI values, 19 of which had also been classified via phylogenetic analysis. Of the gene pairs whose ancestral vs. novel classification were obtained both by statistically significant CAI differences and by phylogenetics, there was one for which the CAI classification contradicted the phylogenetics. That exception was the p104/p130 overlap in the Providence virus. This overlap is notable because the ancestral member of the pair was acquired through horizontal gene transfer, which renders codon usage an unreliable predictor of relative gene ages [35] . We therefore used the phylogenetic classification and disregarded the CAI results. In total, we were able to classify ancestral vs. novel status for 47 overlapping gene pairs (Figure 3 ).