There are several SNPs in the HLA region on chromosome 6, in the lactase locus on chromosome 2, and in the inversion regions on 8p23 and 17q21.31 common in populations of European ancestry [38] that are sources of stratification that will often appear in the top principal components. While one may exclude these SNPs from such an analysis, it is unknown if any similar inversions exist at appreciable frequency in non-European population. Thus it may be preferable to detect and correctly interpret the analysis including these regions rather than avoiding specific regions. The Eigensoft analysis will result in the computation of 10 principal components. If any of these eigenvectors are significantly associated with the phenotype under study, it is recommended that these eigenvectors be adjusted for in any downstream analysis to correct for any bias due to population stratification. Alternatively, if it is expected that only a very small number of samples represent ethnic outliers in the study population, using Eigenstrat with iterative outlier removal, and reconciling these individuals with other ancestry information such as self-report could be used to identify a coherent set of ethnic outliers (which are identified as outliers and self-report as outliers) to potentially exclude from
