Extended FLARE method improves local ancestry inference when reference panels are poorly matched

Sharon R. Browning, Seth D. Temple, and Brian L. Browning of the University of Washington, writing in PLOS Genetics, describe an extension of their FLARE local ancestry inference method designed to address a key practical limitation: the original algorithm required a closely matched reference panel for each ancestry component, which is not available for all populations.

The updated method incorporates a haplotype clustering algorithm that enables accurate local ancestry inference even when one or more ancestries in an admixed individual lack a well-matched reference. The authors report that the extension retains the computational efficiency of the original FLARE implementation while substantially broadening the range of admixed populations to which it can be applied.

The work is demonstrated on real admixed datasets and the method is publicly available as an R package. Accurate local ancestry inference underpins a range of downstream genomic analyses, including admixture mapping, polygenic score portability assessments, and the identification of ancestry-specific variant associations. Improving performance in under-resourced populations has implications for equity in genomic research.

The paper is peer-reviewed and published in PLOS Genetics. It is primarily of interest to statistical and computational geneticists, population geneticists, and researchers working with admixed cohorts.