Mapping the genomic architecture of adaptive traits with interspecific introgressive origin: a coalescent-based approach

Abstract

Recent studies of eukaryotes including human and Neandertal, mice, and butterflies have highlighted the major role that interspecific introgression has played in adaptive trait evolution. A common question arises in each case: what is the genomic architecture of the introgressed traits? One common approach that can be used to address this question is association mapping, which looks for genotypic markers that have significant statistical association with a trait. It is well understood that sample relatedness can be a confounding factor in association mapping studies if not properly accounted for. Introgression and other evolutionary processes (e.g., incomplete lineage sorting) typically introduce variation among local genealogies, which can also differ from global sample structure measured across all genomic loci. In contrast, state-of-the-art association mapping methods assume fixed sample relatedness across the genome, which can lead to spurious inference. We therefore propose a new association mapping method called Coal-Map, which uses coalescent-based models to capture local genealogical variation alongside global sample structure. Using simulated and empirical data reflecting a range of evolutionary scenarios, we compare the performance of Coal-Map against EIGENSTRAT, a leading association mapping method in terms of its popularity, power, and type I error control. Our empirical data makes use of hundreds of mouse genomes for which adaptive interspecific introgression has recently been described. We found that Coal-Map's performance is comparable or better than EIGENSTRAT in terms of statistical power and false positive rate. Coal-Map's performance advantage was greatest on model conditions that most closely resembled empirically observed scenarios of adaptive introgression. These conditions had: (1) causal SNPs contained in one or a few introgressed genomic loci and (2) varying rates of gene flow - from high rates to very low rates where incomplete lineage sorting dominated as a primary cause of local genealogical variation.

Fig. 1

Local genealogical variation: a sequence level view. The illustration shows an example outcome of evolution under the species phylogeny depicted in Fig. 2, where individual a is sampled from population A, individual b is sampled from population B, and individual h is sampled from population H. a A haploid genome sequence is shown for each of the three individuals. Different genealogies are observed for different genomic loci, depending on the specific coalescent history of each locus. Each locus is colored green or blue based on the topology of the genealogy at that locus. The ancestral and derived alleles are represented as 0 and 1, respectively. In our example, the locus marked with a dashed red box contains a causal SNP that contributes to the observed phenotype shown in (b). c Sample structure (or the evolutionary relationships between samples) in green loci differs from sample structure in blue loci. d In our example, global sample structure (i.e., sample structure measured across all sites) takes the form of a star tree. Notice that global sample structure differs from local sample structure in any single locus

Fig. 2

Local genealogical variation: a species phylogeny view. The illustration shows two different pairs of incongruent local genealogies evolving within a species phylogeny: one pair involving incomplete lineage sorting, and the other involving hybrid origin from two different parental populations. The species phylogeny involves three populations A, B, and H. Populations A and B diverged at time t ₁. At time t ₂, a hybridization event between the ancestral populations of A and B occurred, giving rise to a hybrid population H. a The genealogies of two different loci (green and blue) are shown. A lineage in H originated from the ancestral population of B with probability γ (blue locus) or the ancestral population of A with probability 1−γ (green locus). b The genealogies of two different loci (green and red) are shown. The H alleles at both loci originated from the ancestral population of A. For the green locus, the H lineage and A lineage coalesce between time t ₂ and t ₁. For the red locus, tracing backwards in time we find that no coalescence events occur until after time t ₁, resulting in ancestral polymorphism and incomplete lineage sorting. Note that local genealogical variation can involve both topological differences (as shown here) and branch length differences

Fig. 3

Genomic tracts in Mus musculus samples that were inferred to originate via introgression with M. spretus. PhyloNet-HMM [11] was used to infer genomic signatures of interspecific introgression using the procedure described by Liu et al. [5]. Only western European and north African samples are shown. Samples are labeled by country, city/region (if multiple samples originated from the same country), and identifier (if multiple samples originated from the same city/region). The haploid chromosomes are shown for each sample in a separate track and colored a non-white color to denote introgressive origin or white otherwise; each sample is assigned a different color for visibility. The chart shows a subset of the introgressed genomic regions that were inferred to originate due to the recent selective sweep associated with anticoagulant pesticide use. The location of the Vkorc1 gene is highlighted, which is known to contribute to anticoagulant resistance in both mice and humans [4, 12]. Panel adapted from [5]

Fig. 4

Compared to EIGENSTRAT, Coal-Map has comparable or better power and false positive rate on model conditions involving adaptive gene flow (hybridization frequency γ=0.5 and selection coefficient s=0.56). True positive rate and false positive rate are shown for both methods using receiver operating characteristic (ROC) curves. a Results are shown for the model condition where causal SNPs are drawn from a single locus. Coal-Map and EIGENSTRAT have AUROC of 0.947 and 0.889, respectively. b Results are shown for the two-causal-loci model condition. Coal-Map has an AUROC of 0.897 and EIGENSTRAT has an AUROC of 0.859. c Results are shown for the all-causal-loci model condition, where Coal-Map and EIGENSTRAT have AUROC of 0.845 and 0.816, respectively

Fig. 5

Compared to EIGENSTRAT, Coal-Map has comparable or better power and false positive rate on model conditions involving neutral gene flow (hybridization frequency γ=0.5). True positive rate and false positive rate are shown for both methods using receiver operating characteristic (ROC) curves. a Results are shown for the model condition where causal SNPs are drawn from a single locus. Coal-Map and EIGENSTRAT have area-under-ROC-curve (AUROC) of 0.938 and 0.870, respectively. b Results are shown for the two-causal-loci model condition. Coal-Map has an AUROC of 0.898 and EIGENSTRAT has an AUROC of 0.860. c Results are shown for the all-causal-loci model condition, where Coal-Map and EIGENSTRAT have AUROC of 0.837 and 0.827, respectively

Fig. 6

The cumulative histogram of p-values reported by Coal-Map and EIGENSTRAT at causal SNPs is shown on model conditions involving neutral gene flow (hybridization frequency γ=0.5). Results are shown for the a single-causal-locus, b two-causal-loci, and c all-causal-loci model conditions, respectively. Cumulative frequency is reported over all replicates from a model condition

Fig. 7

In the performance study utilizing genomic data from mouse chromosome 7, Coal-Map has similar or typically better power and type I error control compared to EIGENSTRAT. Figure layout and description are otherwise identical to Fig. 4. For the single-causal-locus model condition, Coal-Map and EIGENSTRAT have AUROC of 0.965 and 0.929, respectively; for the two-causal-loci model condition, Coal-Map and EIGENSTRAT have AUROC of 0.942 and 0.923, respectively

Fig. 8

In the performance study utilizing genomic data from mouse chromosome 17, Coal-Map has similar or typically better power and type I error control compared to EIGENSTRAT. Figure layout and description are otherwise identical to Fig. 7. For the single-causal-locus model condition, Coal-Map and EIGENSTRAT have AUROC of 0.968 and 0.914, respectively; for the two-causal-loci model condition, Coal-Map and EIGENSTRAT have AUROC of 0.943 and 0.904, respectively

Fig. 9

The cumulative histogram of p-values reported by Coal-Map and EIGENSTRAT at causal SNPs is shown for the performance study utilizing genomic data from mouse chromosome 7

Fig. 10

The cumulative histogram of p-values reported by Coal-Map and EIGENSTRAT at causal SNPs is shown for the performance study utilizing genomic data from mouse chromosome 17. Figure layout and description are otherwise identical to Fig. 9