The Soifua Manuia reference panel with 2,570 Samoan haplotypes improves genotype imputation quality among Samoans

Abstract

Whole-genome sequencing of 1,285 Samoan individuals combined with 1000 Genomes Project data to build a Samoan-specific imputation reference panel (2,570 haplotypes), since Pacific Islanders are absent from standard panels like TOPMed, 1KGP and HRC. The panel yields up to twice as many well-imputed variants as TOPMed-R3 and 1KGP, with the largest accuracy gains for low-frequency, population-specific variants and those on smaller haplotypes. Just 24-48 Samoan genomes were enough to match or beat TOPMed-R3 performance. Uniquely well-imputed variants were enriched for moderate- and high-impact functional consequences, consistent with founder and bottleneck effects in Samoan population history. Meta-imputation with TOPMed-R3 performed worse than the Samoan panel alone.