Impact of SNP selection on admixture signals

less than 1 minute read


While SNP ascertainment bias is a well known issue in SNP array-based studies (Clark et al (2005), Albrechtsen et al (2010), Lachance et al (2013)), I have not been able to find a published study or discussion in the Internet about the impact that SNP selection bias can have on amixture analysis using the software admixture (Alexander et al (2009)).

Here I explore this topic by 1) genotyping 150 high-coverage samples from Africa, Europe and East Asia (see Polaris Diversity Panel), 2) sub-setting the snps to a) all snps found, b) snps in highly reliable and accessible regions of the human genome, c) snps present in the Illumina Infinium OmniExpress array, and d) snps present in the Human Origins array.