Assessing the limits of local ancestry inference from small reference panels.

Area of Science:

• Genomics
• Population Genetics
• Bioinformatics

Background:

• Admixture is a common biological process with significant implications for evolutionary and demographic studies.
• Existing local ancestry inference (LAI) methods often require large reference panels, limiting their application in non-model organisms and ancient DNA studies.
• The reliability of LAI under various demographic conditions and with limited reference data remains underexplored.

Purpose of the Study:

• To identify demographic conditions favoring accurate local ancestry estimation with minimal reference panels.
• To compare the performance of existing LAI tools (RFMix, MOSAIC) against a novel method (simpLAI) using single individuals as references.
• To establish guidelines for LAI tool usage and propose post-processing steps to enhance accuracy.

Main Methods:

• Simulations of diverse demographic models to test LAI performance.
• Comparative analysis of RFMix, MOSAIC, and simpLAI.
• Development and evaluation of post-painting filtering strategies.

Main Results:

• Demographic conditions conducive to accurate LAI with small reference panels were identified.
• simpLAI demonstrated efficacy even with single diploid genomes per reference population.
• Proposed filtering steps significantly improved the precision and accuracy of inferred admixed tracts.

Conclusions:

• Accurate local ancestry inference is achievable with minimal reference data under specific demographic scenarios.
• The simpLAI method offers a valuable alternative for populations lacking extensive genomic resources.
• This work provides practical guidance for applying LAI tools and improving the reliability of admixture tract identification.