A central part of population genomics consists of finding genomic regionsimplicated in local adaptation. Population genomic analyses are based ongenotyping numerous molecular markers and looking for outlier loci in terms ofpatterns of genetic differentiation. One of the most common approach forselection scan is based on statistics that measure population differentiationsuch as $F_{ST}$. However they are important caveats with approaches related to$F_{ST}$ because they require grouping individuals into populations and theyadditionally assume a particular model of population structure. Here weimplement a more flexible individual-based approach based on Bayesian factormodels. Factor models capture population structure with latent variables calledfactors, which can describe clustering of individuals into populations orisolation-by-distance patterns. Using hierarchical Bayesian modeling, we bothinfer population structure and identify outlier loci that are candidates forlocal adaptation. As outlier loci, the hierarchical factor model searches forloci that are atypically related to population structure as measured by thelatent factors. In a model of population divergence, we show that the factormodel can achieve a 2-fold or more reduction of false discovery rate comparedto the software BayeScan or compared to a $F_{ST}$ approach. We analyze thedata of the Human Genome Diversity Panel to provide an example of how factormodels can be used to detect local adaptation with a large number of SNPs. TheBayesian factor model is implemented in the open-source PCAdapt software.
展开▼
