Identifying signatures of natural selection through the genome using mixed models

Identifying signatures of selection has traditionally relied on detecting traces in modern day genomes. In particular, the length of linkage disequilibrium (LD) blocks in modern day genomes has often been used as an indicator of selection. However, in recent years, the emergence of ancient DNA has enabled new approaches to infer selection that directly use genetic data from the past and reconstruct the evolutionary history of genomes.

In this presentation, I will introduce a methodological framework that was recently proposed to identify variants under selection across the genome: mixed models. Mixed models have long been applied in the Genome-Wide Association Study (GWAS) literature, as they effectively account for population structure and easily integrate confounders. In this context, I will present the framework and outline our plans to further improve current approaches.