We applied our method to an African-American population in this study. Application to more complex admixed populations such as Latinos will have to account for the reduced accuracy in local ancestry inference21 to avoid downward bias. Restricting to two ancestry categories (e.g. Native American vs. non-Native American ancestry)36 is one approach to handle multi-way admixture, but it may be possible to extend our derivation to multi-way admixture. There is evidence that African Americans have a small proportion of admixture from Native American populations (0.5%)24, but this very small proportion is unlikely to significantly change our results. Substantial errors in the assumed population genetic structure would perturb the values of FSTC and θ, and resulting h2 estimates would be biased in proportion to these errors. Application to sex chromosomes can be adapted from the approach taken in[31], but must be analyzed separately due to the differences in admixture proportion of European ancestry on autosomes and sex chromosomes.
