scEvoSel maps evolutionary selection signals onto single-cell transcriptomic atlases to reveal selection-relevant cellular contexts

Abstract

Selection is a core driving force of evolution. Although genome-wide scanning can identify numerous genomic regions affected by selection, it remains challenging to further reveal the specific cellular contexts in which these selection signals act. To address this issue, we developed scEvoSel, an analytical framework that integrates population-level selection signals with single-cell RNA sequencing data to identify selection-relevant cell types. We calculated DCMS scores based on the population variation data from wild boars and domestic pigs and mapped them to the single-cell atlas of skeletal muscle from wild boars and Duroc pigs. The results showed that scEvoSel accurately identified fibro-adipogenic progenitors (FAPs) and myocytes, which exhibited significant changes in cell proportions, as well as lymphoid cells, which showed breed-associated transcriptional divergence. Moreover, we also calculated DCMS scores derived from population variation data of red junglefowl and Leghorn chickens and mapped them to the single-cell atlas of the liver of Leghorn chicken. Hepatocytes, endothelial cells, Kupffer cells and B cells were identified by scEvoSel as the primary selection-relevant cell types. Additionally, scEvoSel enables the analysis of key genes and functional pathways affected by selection within the corresponding cellular contexts. Overall, scEvoSel provides a framework for resolving population selection signals at single-cell resolution, expands the research dimensions for functional interpretation of selection signals, and offers a methodological tool for investigating cellular contexts associated with selection-driven evolutionary processes and complex trait formation.