A draft Korean pangenome graph from 264 assemblies reveals intrapopulation structural variation and improves ancestry-matched variant analysis

Authors:

Abstract
Human pangenome studies have improved representation across populations, but intrapopulation variation in homogeneous populations remains undercharacterized. Here we present a draft Korean pangenome from 264 Korean assemblies, including 8 near-T2T-level assemblies from 4 deeply sequenced individuals, as a pilot for the Korean Pangenome Project to generate 1,000 T2T-level genomes. We identified 20,194,050 SNVs, 5,406,444 small indels, 172,368 structural variants, and 1,521,240 assembly-based variants absent from major public resources. Segmental-duplication analysis defined 6,282 loci spanning 249.0 Mb and revealed Korean-enriched copy-number patterns at ARHGEF5 , AMY1/AMY2A , and C4A/C4B . The Korean pangenome graph contained 25,785,548 graph-derived variants, with 399.6 Mb of Korean private-to-low-shared sequence in the merged graph. In benchmarking, the KPP graph markedly improved SV and InDel detection over HPRC and CPC, especially in medically relevant genes and immune loci including KIR and MHC. These results establish KPP as a foundation for population-specific graph genomics in East Asia.

This publication has 56 references indexed in Scilit: