The abstract is read at the publisher; the summary is JClub's.
KORUS infers ancestry from sequencing-derived k-mer Bloom-filter representations without variant calling, recovering continental-scale structure with supervised accuracy up to 81.8% on 22 WGS samples.
- Why it matters: Existing ancestry inference often depends on explicit variant calls and fixed population labels, limiting flexible use of sequencing data and probabilistic ancestry estimates.
- What they did: KORUS builds supervised and unsupervised ancestry models from 643 1000 Genomes and Simons Genome Diversity Project genomes, then decodes ancestry from k-mer features using Bloom filters.
- The result: On an independent cohort of 22 samples, it achieved 72.7–81.8% classification accuracy and ancestry-fraction concordance R2 = 0.844–0.993 versus ADMIXTURE and R2 = 0.665–0.983 versus ntRoot, and supports probabilistic affinity projections for ancient and archaic genomes.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.