Pangenome graph method improves trait mapping and genomic prediction beyond single-reference approaches
A preprint from the Human Pangenome Reference Consortium describes EdgeDepth, a graph-based method that outperforms linear-reference approaches when mapping genetic variants associated with gene expression.
Researchers working with data generated by the Human Pangenome Reference Consortium (HPRC) have posted a preprint to bioRxiv describing EdgeDepth, a graph-based analytical method designed to associate sequence variation with traits using short-read genomic data mapped to a pangenome variation graph rather than a single linear reference genome.
The HPRC has made available 462 open-access reference genomes alongside a variation graph that captures differences among them. The longstanding limitation of aligning all genomic data to a single linear reference — typically GRCh38 — is well documented: variants in genomic regions that diverge from the reference are systematically under-represented, which can reduce the power of genome-wide association studies (GWAS) and polygenic prediction models. The preprint uses the genetics of gene expression variation (expression quantitative trait loci, eQTL mapping) as a model system to evaluate whether pangenome-based methods address this limitation in practice.
The authors report that EdgeDepth improves trait association signals and genomic prediction accuracy compared with conventional single-reference pipelines, with gains concentrated in regions of high sequence diversity and in variant classes — such as structural variants and insertions — that are poorly captured by standard short-read alignment to a linear reference.
The work is a preprint and has not yet been peer-reviewed. It represents a methodological advance relevant to the design of future GWAS and polygenic risk score research, and to ongoing efforts to ensure that genomic studies reflect the full breadth of human sequence diversity rather than artefacts of reference-genome choice.
Plain-language version
For patients, families, and general readers. Educational only — not medical advice.
Scientists studying human genetics have long compared everyone's DNA to a single "reference" sequence — essentially one template genome. A new approach described in a preprint (a study not yet formally reviewed by other scientists) uses a richer map called a pangenome, built from 462 different people's genomes, to look for links between DNA variation and biological traits more accurately.
The researchers developed a tool called EdgeDepth and tested it by looking at how genetic differences affect the activity of genes. They found their approach picked up signals that the older single-reference method missed, particularly in regions of DNA that vary a lot between people.
This kind of work could help future studies better capture genetic diversity across different populations, ultimately making genetic research more representative. The study is at an early stage and has not yet been peer-reviewed.
This is an educational summary, not medical advice. If anything here raises questions for you, please speak with your GP or a clinical professional.
Sources
Read the original reporting — these are the public sources this summary draws from.
-
Primary sourcePreprint bioRxiv (Cold Spring Harbor Laboratory) · 2026-07-03Pangenome-based human genome analysis improves trait association and genomic prediction