By Diagnostics World Staff
November 3, 2026 | As genome sequencing moves from diagnosing rare disease toward population screening, clinicians face a basic question: what does a normal genome look like? Researchers at the University of Exeter, UK, set out to answer it by mining sequencing data from more than 900,000 people in the US All of Us program and the UK Biobank. Their findings, published this month in Genome Medicine (DOI: 10.1186/s13073-026-01792-x), suggest that “normal” can look alarming on paper.
The median individual differs from the GRCh38 reference genome at about 4.4 million positions, rising to about 5.2 million in people of African ancestry. Most of that variation is common and likely neutral. But the typical genome also carries at least 216 variants predicted to damage protein function, one to three rare pathogenic or likely pathogenic variants (mostly carrier status for recessive conditions), and, by Wright’s estimate, 100 to 200 variants of uncertain significance in disease-associated genes.
“Everybody does have rare, deleterious, pathogenic, disease-causing variation in their genomes,” said corresponding author Caroline Wright, professor at the University of Exeter. Because the cohorts are largely healthy, she argues most VUS are likely benign, and reporting them broadly would generate “a huge amount of work, probably for limited benefit.”
Polygenic scores carry a similar warning. Across 28 common diseases, the average person lands in the top 10% of risk for about three. Non-European individuals are flagged at the extreme end more often, an artifact of scores built largely on European-ancestry data. Wright cautions against calculating scores in everyone without considering symptoms, family history, and other clinical context.
The team also estimated about 60 de novo mutations per genome, roughly 80% from the father, and called for more diverse sequencing and pangenome references to improve interpretation for underrepresented populations.
For the full story, including more on population diversity, de novo mutations, and future multiomic directions, read Kyle Proffitt’s article on Bio-IT World.