An Icelandic pangenome reference.

Pathogenic variants located in low-mappability regions, including a SNP in GBA1 and a missense SNP in CBS, missed when short reads are mapped to a linear reference.

Reference bias affects most genomic studies that map short reads to a linear reference genome, and can be mitigated by mapping to the multiple haplotypes represented in a pangenome. The authors introduce two methods: Emblask, a hybrid long- and short-read haplotype-resolved dual assembly pipeline for parent-offspring trio data, and Weaver, for mapping to pangenomes at scale. They assembled 698 Icelandic haplotypes and added them to the Human Pangenome Reference Consortium pangenome to build the HPRC-ICE reference, which includes 51.41 million small variants. Mapping the short reads of 57,630 Icelanders to HPRC-ICE called 98.96 million variants, a 6.17% increase over a linear reference, and uncovered variants in low-mappability regions, including a pathogenic SNP in GBA1 associated with early-onset Parkinson's disease and a missense SNP in CBS pathogenic for homocystinuria. The GBA1 association was replicated in the UK Biobank through targeted remapping of 429,193 British and Irish participants.

Synthesis written by Geno'X. For the full original abstract, please refer to the source publication.

The 6.17% gain in variants concerns regions that were poorly covered until now, and the GBA1 example shows that pathogenic variants are genuinely lost by linear-reference pipelines, which bears on annotation and reanalysis of negative exomes and genomes far more than on panel composition. Two caveats: the reference is enriched in Icelandic haplotypes, so its benefit in other genetic backgrounds remains to be quantified, and moving to a pangenome means revisiting coordinates, filters and annotation across the whole diagnostic chain. What would settle the matter for a laboratory is an additional diagnostic yield measured in unsolved patients, not a variant count.

Analysis by Dr Thibaut Benquey