For decades, efforts to clock the speed of human evolution and trace the roots of genetic disease have been missing a crucial piece: the messiest, most repetitive stretches of our DNA. A sweeping new analysis, published in Nature, has finally filled that gap. By mapping the complete genomes of 28 individuals spanning four generations of a single family, researchers discovered that the true rate of genetic change is far richer and more complex than textbooks suggest. Short, unstable tandem repeats emerged as the unrivaled champions of mutation, and the number of large-scale structural variants that appear each generation may be a full order of magnitude higher than previously believed.
A Multi-Generation Puzzle
The study hinged on a remarkable resource: a pedigree from the Centre d'Etude du Polymorphisme Humain (CEPH) that has been followed since the 1990s. Scientists assembled high-quality diploid genomes for individuals across four generations, allowing them to trace new, or de novo, mutations that appeared in each child but were absent from both parents. By looking at grandparents, parents, and children together, the team could confidently separate genuine new mutations from sequencing errors and rare inherited variants.
Most prior work relied on “trio” designs (mother, father and child) sequenced with short-read technology. Short reads excel at finding single-letter changes (single nucleotide variants, or SNVs) but stumble when confronted with repetitive DNA. Tandem repeats, segments where a short DNA motif is repeated many times in a row, are notoriously difficult to read accurately and are therefore often excluded. Yet these regions are known to be genomically hyperactive, and ignoring them left a gap in our understanding.
Most prior work relied on “trio” designs (mother, father and child) sequenced with short-read technology. Short reads excel at finding single-letter changes (single nucleotide variants, or SNVs) but stumble when confronted with repetitive DNA. Tandem repeats, segments where a short DNA motif is repeated many times in a row, are notoriously difficult to read accurately and are therefore often excluded. Yet these regions are known to be genomically hyperactive, and ignoring them left a gap in our understanding.
An Arsenal of Sequencing Technologies
To capture the full landscape of de novo change, the team threw five complementary sequencing methods at the problem. They combined the extreme accuracy of PacBio HiFi and the ultra-long reads of Oxford Nanopore with the short-read precision of Illumina, plus Strand-seq and Element AVITI for additional structural insights. More than 95% of each person’s diploid genome was assembled using two hybrid pipelines, Verkko and hifiasm. The result was a detailed, family-wide catalog of spontaneous mutations spanning SNVs, small insertions and deletions (indels), and large structural rearrangements.
The first three generations were used to build the genomic baseline; the fourth and youngest generation provided the de novo landscape. In total, the researchers identified between 98 and 206 new mutations per child, with an average of approximately 152 de novo events per generation. That figure integrates everything from single base swaps to sweeping structural changes.
The first three generations were used to build the genomic baseline; the fourth and youngest generation provided the de novo landscape. In total, the researchers identified between 98 and 206 new mutations per child, with an average of approximately 152 de novo events per generation. That figure integrates everything from single base swaps to sweeping structural changes.
Tandem Repeats: The Mutation Accelerators
Breaking the average down revealed stark asymmetries. An average of 74.5 SNVs appeared per generation, but the number varied by more than an order of magnitude depending on the surrounding genomic context — consistent with recent population-scale analyses. The most dramatic contribution, however, came from tandem repeats. Indels and structural variants rooted in these repetitive stretches averaged 65.3 per generation, dwarfing the 7.4 indels found outside repeat regions. Short, highly variable tandem repeats were the undisputed hotspots, and 32 specific loci mutated recurrently across generations, acting like molecular clocks that tick with unusual speed.
Centromeres, the densely packed hubs of chromosomes, were not immune. The study documented an average of 4.4 new centromeric mutations per generation, a figure seldom captured by standard methods.
Centromeres, the densely packed hubs of chromosomes, were not immune. The study documented an average of 4.4 new centromeric mutations per generation, a figure seldom captured by standard methods.
Structural Variants Rewritten
Perhaps the most striking revision concerns structural variants (SVs) — large-scale rearrangements such as deletions, duplications, and inversions. The widely accepted rate for germline SVs hovered around 0.2–0.3 events per generation. The new data tell a different story: the team uncovered 3–4 de novo structural variants in each new generation, a rate more than ten times higher than past estimates. This leap stems directly from the ability to resolve repetitive regions where many SVs originate. If confirmed in broader cohorts, it would significantly recalibrate how scientists think about genome instability and the origins of structural birth defects.
Fathers, Age, and Post-Zygotic Events
The paternal lineage dominated the mutational load. Between 70% and 80% of all detected de novo variants originated in the father’s germline, and the proportion increased with paternal age — an observation that reinforces the well-known link between older fathers and a higher number of new point mutations in offspring. Meanwhile, not every de novo SNV was a true germline event. The researchers identified that 16% of the apparent new mutations were post-zygotic, arising shortly after fertilization and leading to mosaicism, where some cells in the body carry the variant and others do not.
A focused assembly of six Y chromosomes from male family members yielded an average of 12.4 de novo mutations per generation on the male-specific sex chromosome, adding a dedicated view of paternal-line evolution.
A focused assembly of six Y chromosomes from male family members yielded an average of 12.4 de novo mutations per generation on the male-specific sex chromosome, adding a dedicated view of paternal-line evolution.
A Foundation for the Future
By weaving together diverse sequencing technologies and a four-generation family, the study re-frames the pace of human mutagenesis. It shows that the genome’s most variable regions, especially short tandem repeats and structural variants, are far more dynamic than widely appreciated. The average of 152 new mutations per generation is not just a number; it is a new reference point for understanding how heritable disease risk and human diversity arise.
The authors caution that one family, however deeply analyzed, cannot speak for all of humanity. The next step is to scale the approach to many families and populations, testing whether these elevated rates hold universally. If they do, the implications stretch from evolutionary biology to clinical genetics, prompting a closer look at repetitive regions as potential drivers of complex disease. The findings also underscore a methodological lesson: when it comes to capturing the full mutational spectrum, there is no substitute for long reads, multi-platform data, and multi-generational depth.
The authors caution that one family, however deeply analyzed, cannot speak for all of humanity. The next step is to scale the approach to many families and populations, testing whether these elevated rates hold universally. If they do, the implications stretch from evolutionary biology to clinical genetics, prompting a closer look at repetitive regions as potential drivers of complex disease. The findings also underscore a methodological lesson: when it comes to capturing the full mutational spectrum, there is no substitute for long reads, multi-platform data, and multi-generational depth.
Publication date: 2026.05.08
Source:
Porubsky D. et al. Human de novo mutation rates from a four-generation pedigree reference. Nature. 2025 Jul;643(8071):427-436. doi: 10.1038/s41586-025-08922-2. Epub 2025 Apr 23. PMID: 40269156; PMCID: PMC12240836.
https://www.nature.com/articles/s41586-025-08922-2
Source:
Porubsky D. et al. Human de novo mutation rates from a four-generation pedigree reference. Nature. 2025 Jul;643(8071):427-436. doi: 10.1038/s41586-025-08922-2. Epub 2025 Apr 23. PMID: 40269156; PMCID: PMC12240836.
https://www.nature.com/articles/s41586-025-08922-2