Poster #8 - Christopher Sottolano

Revealing the hidden landscape of selection: detecting divergent, soft, and background signatures in diverse human genomes

Christopher J. Sottolano, PhD [1], Grace Tzun-Wen Shaw, PhD [1], Timothy L. Mosbruger, PhD [1], Yuncheng Duan, PhD [3], Andrew S. Allen [3], Dimitri S. Monos, PhD [1,2], Tristan J. Hayeck, PhD [1,2] 1. Division of Genomic Diagnostics, Department of Pathology and Laboratory Medicine, Children's Hospital of Philadelphia 2. Department of Pathology and Laboratory Medicine, Perelman School of Medicine, University of Pennsylvania [3] Department of Biostatistics and Bioinformatics, Duke University

Throughout evolution, environmental pressures have left discernible patterns of genetic variation across human populations. These signatures can be used to uncover disease-linked variants and inform diagnosis and treatment. Existing methods of detecting these patterns excel in settings of hard sweeps, where selective pressure is strong enough to drive specific haplotypes to fixation. However, they are less effective in detecting background selection or soft sweeps, which are more subtle but thought to be frequent and potentially associated with many diverse phenotypes. We developed a new test statistic for detecting divergent positive and balancing selection, including soft sweeps. The test detects regional differentiation in the site frequency spectrum, linkage disequilibrium, and density of polymorphisms across populations to characterize signatures of selection. Sensitivity analysis was performed using simulations based on an out-of-Africa demographic model to generate realistic genetic patterns. Our method improved detection of early incomplete sweeps (positive selection but not fixed) achieving 30% higher AUC and 21% higher precision at FDR 0.05 compared to leading methods. This study leverages the latest release of 1000 Genomes Project (1KGP) to characterize genetic diversity and evolutionary selection among world populations. This ties potential divergent evolution to corresponding genes and diseases in specific populations. Divergent selection spanning the BTNL2 gene was seen in East Asians at variants tied to both liver and lung function. Whereas in Africans, strong divergent selection signal is observed across ALG10B at variants related to cardiomyopathy, along with signal in HLA-C and SPOCK3 related to immune response. Interestingly, consistent pattern enrichment of selection signal across populations were observed in both unprocessed pseudogenes and polymorphic pseudogenes pointing to potential biological functions favoring diversification and development of novel genes. Our results reveal both unique and shared patterns of evolutionary selection across global populations, highlighting their implications of genetic variation on disease phenotypes.