Preprint: long-read sequencing of 1,490 Hispanic/Latino participants in All of Us yields 10 million novel variants

A preprint posted to bioRxiv on 2 November 2026 reports findings from long-read whole-genome sequencing of 1,490 self-identified Hispanic/Latino participants enrolled in the National Institutes of Health All of Us Research Program.

The authors used long-read sequencing — which resolves complex structural variants and repetitive regions that short-read technologies often miss — alongside ancestry-aware analytical frameworks to characterise genetic variation, local ancestry, and DNA methylation patterns across this admixed cohort. They identify approximately 10 million novel variants not previously catalogued in existing reference databases, along with more than 100 candidate phenotype associations that were not detectable using conventional short-read approaches or population reference panels that do not adequately represent admixed Latino ancestry.

The study speaks directly to the persistent underrepresentation of Hispanic/Latino populations in large-scale human genetic studies, a gap that limits both the discovery of medically relevant variants and the accuracy of polygenic models when applied to these populations. The authors emphasise the importance of ancestry-aware interpretation and representative reference data for variant classification in admixed individuals.

The work is of interest to population genetics researchers, those working on health equity in genomics, and educators discussing the limitations and future directions of genome-wide association and sequencing studies. This is a preprint and has not yet undergone peer review.