The landscape of nucleotide diversity in Drosophila melanogaster is shaped by mutation rate variation

Abstract

What shapes the distribution of nucleotide diversity along the genome? Attempts to answer this question have sparked debate about the roles of neutral stochastic processes and natural selection in molecular evolution. However, the mechanisms of evolution do not act in isolation, and integrative models that simultaneously consider the influence of multiple factors on diversity are lacking; without them, confounding factors lurk in the estimates. Here we present a new statistical method that jointly infers the genomic landscapes of genealogies, recombination rates and mutation rates. In doing so, our model captures the effects of genetic drift, linked selection and local mutation rates on patterns of genomic variation. We then formalize a causal model of how these microevolutionary mechanisms interact, and cast it as a linear regression to estimate their individual contributions to levels of diversity along the genome. Our analyses reclaim the well-established signature of linked selection in Drosophila melanogaster, but we estimate that the mutation landscape is the major driver of the genome-wide distribution of diversity in this species. Furthermore, our simulation results suggest that in many evolutionary scenarios the mutation landscape will be a crucial factor shaping diversity, depending notably on the genomic window size. We argue that incorporating mutation rate variation into the null model of molecular evolution will lead to more realistic inferences in population genomics.

Improved inference of population histories by integrating genomic and epigenomic data

“…Additionally, the potential complexity resulting by integrating multiple independent markers could be tackled by the use of continuous time Markov chains for the emission matrix. We predict that our results pave the way to improve the inference of 1) biological traits or recombination rate through time [17, 68], 2) multiple merger events [37], and 3) recombination and mutation rate maps [5, 4]. Our method also should help to dissect the effect of evolutionary forces on genomic diversity [32, 31], and to improve the simultaneous detection, quantification and dating of selection events [1, 8, 30].…”

“…and 3) recombination and mutation rate maps [5,4]. Our method also should help to dissect the effect of evolutionary forces on genomic diversity [32,31], and to improve the simultaneous detection, quantification and dating of selection events [1,8,30].…”

“…We additionally set the region level methylation and demethylation rate to µ RM = 1.6 × 10 − 4 and µ RU = 9.5 × 10 − 4 per generation per bp according to [18]. Because we do not account for the effect of variable mutation or recombination rate along the genome, we cut the five chromosome of A. thaliana into eight smaller scaffolds [4, 5]. By doing this we remove centromeric regions and limit the effect the variation of mutation and recombination rate along the genome.…”

Simultaneous Inference of Past Demography and Selection from the Ancestral Recombination Graph under the Beta Coalescent

“…subgraphs) during the forward-pass step. To illustrate the GNN strategy, we visualize the compression-like process, from the coalescent trees (1) being processed by GNNcoal (2,3) to the inferred variable of interest (4,5) in Figure 1. the analogues compression of node embeddings (or feature vectors) as in Fig.…”

“…Sub-graphs are generated by a pooling network with updated feature vectors and a final compression step is performed until ideally one node per graph remains (2)(3). Lastly, the column-wise mean is taken after applying a time mask (blue -based on number of coalescent events), so that single feature vector remains (4)(5). Detailed description of the graph convolution, feature vector initialization, pooling methodology, coalescent time mask construction, and dataset generation can be found in Supplementary Text S2 or [102].…”