Putting RFMix and ADMIXTURE to the test in a complex admixed population

Area of Science:

• Population Genetics
• Computational Biology
• Genomics

Background:

• Admixed human populations present challenges for ancestry inference tools.
• Existing computational tools have been validated primarily on simple admixture histories.
• Performance of ancestry inference tools in complex admixture scenarios remains largely unknown.

Purpose of the Study:

• To evaluate the accuracy of computational tools for global and local ancestry inference in complex admixed populations.
• To compare the performance of RFMix against ADMIXTURE in intricate admixture scenarios.
• To assess the utility of accurate ancestry estimation for downstream genetic analyses.

Main Methods:

• Utilized simulations to model complex admixture scenarios.
• Employed the RFMix algorithm for ancestry inference.
• Compared RFMix performance with ADMIXTURE for global ancestry proportions.
• Assessed local ancestry assignment accuracy.

Main Results:

• RFMix demonstrated superior performance in determining global ancestry proportions compared to ADMIXTURE.
• RFMix achieved 89% accuracy in assigning local ancestry in a 5-way admixed population.
• The tool accurately estimates both global and local ancestry in complex admixed groups.

Conclusions:

• RFMix is a highly accurate tool for global and local ancestry estimation in complex, worldwide admixture scenarios.
• The study underscores the need for computational tools compatible with genetically structured populations.
• Accurate ancestry estimates from RFMix can enhance genetic association studies in increasingly admixed populations.