Application of partial least squares in exploring the genome selection signatures between populations.

Natural and artificial selection have led to substantial variation in the phenotypic traits of different populations. Therefore, there is a need to develop methods that are based on cross-population comparisons to discover loci related to specific traits. Here, we suggested a strategy to detect the genome selection signatures between populations based on the partial least squares (PLS) theory. Using the binary population indicator as the response variable in the PLS analysis, alleles under selection between populations were identified from the first PLS component. We explored the theory behind the PLS analysis to reveal its usefulness in detecting the loci under selection. Through the simulation study, the results showed that the PLS method had a better performance than the FST and EigenGWAS methods. In addition, by using the real data hapmap3, we found that rs11150606 in PRSS53 gene and rs1800414 in OCA2 gene were under selection between East Asian populations and three other populations, including African, American, and European populations. We concluded that this strategy was easily carried out and might supplement for the deficiency of the EigenGWAS method in some cases. To facilitate the application of this method, we developed an R script that is freely accessible at http://klab.sjtu.edu.cn/PLS/ .