Genomic scans for adaptive introgression, using a new method, that authors call “VolcanoFinder”

These GEITP pages have often discussed the topic of evolution of humans, including “adaptive introgression” (i.e. the process by which beneficial alleles are introduced into a species from a closely-related species). [Recall that ‘allele’ is a gene on one chromosome; the chromosome pair has one allele from the mother, the other allele on the other chromosome from the father).] Authors [see attached article] present an analytically-manageable model for studying the effects of adaptive introgression on non-adaptive genetic variation in the genomic region surrounding the beneficial allele. The result described is a characteristic volcano-shaped pattern of increased variability that arises around the positively-selected site; thus, authors present an open-source method, VolcanoFinder, to detect this signal in genomic data. Notably, VolcanoFinder is a population-genetic likelihood-based approach, rather than a comparative-genomic approach; therefore, one can probe genomic variation data from a single population for footprints of adaptive introgression — even from an unknown and possibly extinct donor species.

Whereas classic species concepts imply genetic isolation, research over the last 3 decades shows that hybridization between closely related species (or diverged subspecies) is widespread. For adaptation research, this offers the intriguing perspective of an exchange of key adaptations between related species, with potentially important implications for our view of the adaptive process. Recent studies have brought solid evidence of cross-species introgression of advantageous alleles: well-documented examples cover a wide range of taxa — including transfer of wing-pattern mimicry genes in Heliconius butterflies, herbivore resistance and abiotic (i.e. non-living part of an ecosystem that shapes its environment, e.g. temperature, light, and water) tolerance genes in wild sunflowers, pesticide resistance in mice and mosquitoes, and new mating and vegetative incompatibility types in an invasive fungus. Such adaptive introgressions have also been found to have occurred in modern humans: [a] local adaptation to hypoxia (low atmospheric oxygen levels) at high-altitude was shown to be associated with selection for a Denisovan-related haplotype at the EPAS1 (hypoxia pathway) gene in Tibetan populations; [b] positive selection has been characterized for three archaic haplotypes, independently introgressed from Denisovans or Neanderthals in a cluster of genes involved in the innate immune response, and [c] immunity-related genes show evidence of selection for Neanderthal and Denisovan haplotypes (these GEITP pages have previously discussed these papers; recall ‘haplotype’ refers to a segment of genes on same chromosome — coming from one parent).

In all examples above, evidence of adaptive introgression rests on a comparative analysis of DNA from both donor and recipient species. In particular, human studies often rely on maps of introgressed Neanderthal or Denisovan fragments in the modern human genome. The tell-tale signature of adaptive introgression is a segment of mutations from the donor population that is present in strong linkage disequilibirum (LD; non-random association of alleles at different loci in a given population) and in high frequency in the recipient population. Unfortunately, solid data from a potential donor species may not always be available — especially in the case of an extinct donor. In absence of a donor, introgression can sometimes be inferred from haplotype statistics in the recipient species, the most recent methods making use of machine-learning algorithms based on several statistical methods.

However, there is currently no framework for a joint inference of admixture and selection (e.g. adaptive introgression), and selection is usually inferred from an unexpectedly high frequency of introgressed haplotypes. A recent article on adaptive introgression in plants identified four different types of studies in this field, focusing on [a] introgression, [b] genomic signatures of selection, [c] adaptively relevant phenotypic variations, and [d] fitness. The attached paper attempts to bridge the gap between classes [a] and [b] — and to detect specific genomic signatures of an introgression sweep.

Using coalescent theory, authors [see attached article] derived analytical predictions for these patterns. Based on these results, authors developed a composite-likelihood test to detect signatures of adaptive introgression, relative to the genomic background. Simulation results showed that VolcanoFinder has high statistical power to detect these signatures, even for older sweeps and for soft sweeps initiated by multiple migrant haplotypes. Lastly, authors implement VolcanoFinder to detect archaic introgression in European and sub-Saharan African human populations; they uncovered interesting candidates in both populations — such as TSHR (throid-stimulating hormone receptor) in Europeans and TCHH— RPTN (epidermal differentiation complex (EDC) on chromosome 1) in Africans. These biological implications provide guidelines for identifying and circumventing artifactual signals during empirical applications of VolcanoFinder. 😊


PLoS Genet Jun 2020; 16: e1008867

This entry was posted in Center for Environmental Genetics. Bookmark the permalink.