Comparison of PCA and DAPC methods for analysis of Iranian Buffalo population structure using SNPchip90k data

Understanding of population genetic structure is valuable for better implementation of breeding programs and most importantly, preservation of genetic resources. Genomic data provide an opportunity to consider complex evolutionary history of populations and reconstruct rare historical events. In this research, the structure of Iranian buffalo populations was studied by using principal component analysis and discriminant analysis principal component methods. For this purpose, the number of 404 buffalos from three breeds including North, Azari and Khozestani were sampled and genotyped by SNPChip 90k from Padano Company in Italy. The results of principal component analysis and discriminant analysis principal component showed a clear picture of the genetic structure of the studied populations. Assessing the optimal number of clusters with criteria BIC, K = 3 by the DAPC method showed the best results. The result of cross-validation for retaining principal components was optimized to 50 first components that showed the lowest MSE. In this study, DAPC predicted assignment of individuals to clusters and membership probabilities with 100% accuracy. PCA method was not able to provide a group assessment and DAPC method outperformed than PCA in achieving a clear variance difference between populations. DAPC method can be applied in quality control and stratification population correction of GWAS as an alternative to the PCA because of summarizing the genetic differentiation between groups and overlooking within-group variation and providing better population structure.