Detecting genomic signatures of natural selection with principal component analysis: application to the 1000 Genomes data

摘要

To characterize natural selection, various analytical methods for detectingcandidate genomic regions have been developed. We propose to performgenome-wide scans of natural selection using principal component analysis. Weshow that the common Fst index of genetic differentiation between populationscan be viewed as a proportion of variance explained by the principalcomponents. Considering the correlations between genetic variants and eachprincipal component provides a conceptual framework to detect genetic variantsinvolved in local adaptation without any prior definition of populations. Tovalidate the PCA-based approach, we consider the 1000 Genomes data (phase 1)after removal of recently admixed individuals resulting in 850 individualscoming from Africa, Asia, and Europe. The number of genetic variants is of theorder of 36 millions obtained with a low-coverage sequencing depth (3X). Thecorrelations between genetic variation and each principal component providewell-known targets for positive selection (EDAR, SLC24A5, SLC45A2, DARC), andalso new candidate genes (APPBPP2, TP1A1, RTTN, KCNMA, MYO5C) and non-codingRNAs. In addition to identifying genes involved in biological adaptation, weidentify two biological pathways involved in polygenic adaptation that arerelated to the innate immune system (beta defensins) and to lipid metabolism(fatty acid omega oxidation). An additional analysis of European data showsthat a genome scan based on PCA retrieves classical examples of localadaptation even when there are no well-defined populations. PCA-basedstatistics, implemented in the PCAdapt R package and the PCAdapt open-sourcesoftware, retrieve well-known signals of human adaptation, which is encouragingfor future whole-genome sequencing project, especially when definingpopulations is difficult.