Neutral mutation disequilibrium is common in prokaryotes and animals

Widely used models of sequence evolution assume DNA sequences are at equilibrium with respect to the processes that drive divergence, mutation and natural selection. Violations of this assumption have been shown to affect the conclusions from analyses of genetic variation. However, there is currently a lack of statistical methods to assess whether this phenomenon occurs with sufficient frequency and magnitude to warrant concern. Here, we introduce such techniques and apply them to cases where genetically encoded changes to mutation are expected to induce neutral mutation disequilibrium (NMD) across an entire genome (loss of the hypermutable base 5-methyl-cytosine in Drosophila melanogaster), or in a limited genomic region (translocation of the Fxy gene to the pseudo-autosomal region in Mus musculus). We find that the proportion of 9,237 genes in NMD in the D. melanogaster genome ([~]0.81) is markedly elevated compared to its sister taxa D. simulans ([~]0.31) which retains 5-methyl-cytosine. We further establish that the translocation of Fxy has resulted in a [~] 3x increase in the magnitude of NMD. These empirical case studies validate the robustness of our methods. Applying the methods to human evolution, from 613 genes, we estimated the proportion of NMD for the exons as [~]0.64 and introns as [~]0.22. These results establish NMD as a common phenomenon and suggest that conclusions from studies that assumed equilibrium in order to understand biological mechanisms may misattribute to biology what should be credited to measurement error. SignificanceIt has been famously said that all models are wrong, but some are useful. Statistical models of sequence divergence provide the fundamental quantities from which we derive meaning about the causative mechanisms responsible for patterns in molecular evolution. Published work has established current models are not just wrong, but misleading - incorrectly assuming neutral mutation equilibrium leads to errors in the inference of the mode of natural selection and the existence of the molecular clock. In this work, we demonstrate that neutral mutation disequilibrium is common, suggesting that these errors may be widespread. Our results raise the strong likelihood that detected patterns in molecular evolution are being misattributed to biological mechanisms when, in fact, they reflect systematic measurement error.