120 Using Simulated Data to Evaluate the Accuracy of Selection Signature Detection Methods in Varying Scenarios

• Published in: Journal of animal science 2023-11, Vol.101 (Supplement_3), p.14-15
• Main Authors: Retief, Anel, Brito, Luiz F F, Visser, Carina
• Format: Article
• Language: English
• Subjects: Accuracy; Animal populations; Animals; Demographics; Demography; Genetic diversity; Genomes; Genotypes; Genotyping; Population; Population genetics; Population number; Populations; Quantitative trait loci; Signatures; Statistical methods
• ISSN: 0021-8812; 1525-3163
• DOI: 10.1093/jas/skad281.018

Abstract The identification of molecular signatures of selection has become increasingly important in recent years. The detection of selection signatures can aid in both population characterization as well as the detection and identification of genes and advantageous mutations within populations. Various statistical methodologies have been developed to detect directional selection signatures based on different demographic or selection models. The detection of definite evidence of selection on genomes remains a challenge, as other events in the demographic history of a population under investigation may imitate the effects of selection on the genetic variation in the genome. To accurately detect selection signatures, researchers need to consider the large number of methodologies available as well as the factors influencing the detection of selection signatures. Using simulated data, it is possible to evaluate several methods and factors on a larger number of genotypes without the expense of genotyping several animals. The objective of this study was to simulate several sheep populations that have undergone different selection pressures, with varying parameters that may possibly influence the accuracy of detecting selection signatures and testing multiple methods of selection signature detection to find the most reliable method for each scenario. Sheep populations were simulated with varying numbers of QTL (5, 100, 2000) and h2 (0.15, 0.4). The populations were then expanded through either selection for high, low, or random genetic values. Thereafter the populations were selected for sample sizes of 20, 200 and 1000 individuals, resulting in a total of 54 populations for analysis. The populations were then subjected to ROH, iHS, Fst and XP-EHH analysis. After the most effective method for detection of selection signatures for each scenario was identified, the influence of effective population size, generations of animals included in the sample, relatedness, and sex of individuals on the accuracy of each method was tested. Preliminary results indicated that the accuracy of XP-EHH as well as ROH increased with larger population sizes, as well as greater numbers of QTL. Further results of this study will be presented at the conference.