Ectopic gene conversion shown to generate recurrent variants across 534 million gnomAD sites
A preprint using gnomAD v4.1 data argues that the standard single-origin assumption for variant interpretation fails for a substantial fraction of human genetic variation, with ectopic gene conversion reintroducing the same nucleotide changes predictably across generations.
A central assumption in population genetics — that a given variant arose once and was subsequently shaped by selection, genetic drift, and demographic history — may be wrong for a substantial proportion of human variation. A preprint posted to bioRxiv on 28 August 2026 presents evidence from gnomAD v4.1 (534 million variants) that ectopic gene conversion (EGC) — a process in which DNA sequence is copied from one genomic location to a non-allelic homologous site — reintroduces the same nucleotide changes generation after generation.
The authors show that EGC rates are strongly structured by two fixed properties of genome architecture: the length of the homologous template sequence and the distance between the donor and acceptor sites. This makes the process, in principle, predictable from sequence data alone. The finding challenges standard population-genetic interpretations of allele frequencies and has direct implications for how researchers model variant origins and infer evolutionary history.
For geneticists working on variant interpretation, the results suggest that some variants currently treated as rare or unique in origin may instead represent recurrently generated alleles — a distinction that could matter for understanding their functional context. The preprint has not yet been peer-reviewed.
Sources
Read the original reporting — these are the public sources this summary draws from.
-
Primary sourcePreprint bioRxiv (Cold Spring Harbor Laboratory) · 2026-08-28Sequence architecture shapes human allele frequencies through ectopic gene conversion