A potential source of bias in our estimates is from errors in the ancestry inference algorithm. To show that our estimates are not the result of Ancestry Composition errors or biases, we validated the estimates of low levels of African ancestry in European Americans comparing to f4 statistics,64 1000 Genomes Project consensus estimates,78 and ADMIXTURE estimates.66 Another line of evidence supporting our estimates of non-European ancestry in European Americans in the US is that we observe a substantially lower occurrence of Native American and African ancestry in individuals who self-report four grandparents born in the same European country. The inferred segments of African and Native American are uniformly distributed across the genome. Although we expect that some of the inferred ancestry might arise from difficulties in assigning ancestry in complex regions of the genome, only a small fraction of the estimated African and Native American ancestry in European Americans can be explained through such biases and is not expected to give rise to any substantial (more than 1%) ancestry from any population.