The challenge that the team set out to overcome was technical rather than conceptual. For decades, scientists have known that G4-forming sequences exist throughout the genome, but the most repetitive parts of human DNA, such as centromeres and the long arrays of ribosomal DNA, resisted assembly with standard short-read sequencing technologies. These regions are composed of long, near-identical repeats that short reads cannot be uniquely placed, leaving gaps in even the most celebrated reference assemblies. The recent arrival of long-read sequencing and the resulting gap-less, telomere-to-telomere human genome assemblies changed the picture entirely. For the first time, researchers could scan the entirety of a complete human genome, from end to end, for sequences with the potential to form G-quadruplexes, and then compare those findings across many individuals.

In the new study, the researchers systematically identified potential G4-forming sequences in the gap-less reference human genome assembly and extended their analysis to 88 haplotypes, or individual genome copies, drawn from people of diverse ancestries. This breadth mattered. By looking across many genomes rather than a single reference, the team could ask not only where G4s sit in the genome but also how variable they are between people, a direct window into their evolutionary and mutational behavior. The scale of the analysis allowed them to move beyond anecdotal observations and build a population-level picture of G4 landscape and instability.

One of the most striking findings concerns where G4s concentrate. The team reports that G4-forming sequences are highly enriched in specific repetitive regions of the genome, including certain centromeric and pericentromeric repeat types, the specialized DNA that surrounds and structures the centromere, the point where chromosomes attach to the segregation machinery during cell division. They also found marked enrichment in ribosomal DNA arrays, the tandem repeats that encode the RNA components of ribosomes, the molecular machines that build proteins in every cell. These are precisely the regions that had been invisible to earlier G4 surveys, which means that a substantial fraction of the human G4 landscape had effectively been missing from the scientific record until now.

Importantly, the researchers did not rely on sequence patterns alone. Computational predictions of G4 formation are useful, but they can be misleading, because not every guanine-rich motif actually folds into a quadruplex under physiological conditions. To address this, the team experimentally validated the most prevalent G4 structures they detected, confirming that the sequences identified in these repetitive regions can indeed form the characteristic four-stranded architecture. This biophysical validation anchors the computational survey in laboratory reality and strengthens the case that the newly mapped G4s are genuine structural features of human DNA rather than artifacts of pattern matching.

The study also examined the chemical modification landscape around G4s, with intriguing results. DNA methylation, the addition of a methyl group to cytosine bases, is a key epigenetic mark that influences whether genes are active or silent. The researchers found that G4-forming regions tend to show lower methylation than would be expected by chance. This observation fits with a growing body of evidence that alternative DNA structures can influence the epigenetic state of the sequences that carry them, potentially affecting how nearby genes are regulated. While the study establishes the correlation across the complete genome, it also raises new questions about cause and effect: whether G4 formation actively discourages methylation, or whether unmethylated regions are simply more permissive to quadruplex folding, remains an open avenue for future work.

Perhaps the most consequential finding relates to genome stability. By comparing G4 loci across the 88 haplotypes, the researchers could measure how often these sequences change from one generation to the next. The answer was unambiguous: G4-forming regions are genomically unstable, showing an excess of insertions and deletions relative to the surrounding genome. This makes mechanistic sense. A four-stranded structure folded out of one DNA strand creates an awkward substrate for the enzymes that copy and repair DNA, and stalled replication forks at stable secondary structures are a classic source of small insertions and deletions. The new data suggest that this mutagenic tendency is a genome-wide property of G4s, measurable at the population scale, and that these structures act as ongoing engines of genetic variation in repetitive DNA.

The team then connected G4s to one of the most dramatic events in the life of a genome: meiosis, the specialized cell division that produces sperm and eggs. During meiosis, programmed double-strand breaks are deliberately introduced at thousands of hotspots throughout the genome to initiate recombination, the shuffling of genetic material between chromosomes. Which sites are cut is largely determined by PRDM9, a protein whose DNA-binding domain varies between individuals and whose genotype therefore reshapes the recombination landscape from person to person. Remarkably, the researchers found that G4-forming motifs are consistently enriched at meiotic double-strand break hotspots across individuals with different PRDM9 genotypes. This consistency suggests that the physical propensity of DNA to fold into quadruplexes may contribute to the targeting or processing of meiotic breaks in a way that transcends the specific PRDM9 variant an individual carries, hinting at a structural layer of regulation overlaid on the well-known genetic one.

Taken together, the findings reframe G-quadruplexes not as curiosities of guanine-rich sequence but as dynamic, widespread, and functionally meaningful elements of the human genome. Their enrichment in centromeric and ribosomal DNA places them at the heart of chromosome architecture and protein synthesis; their reduced methylation ties them to epigenetic regulation; their excess of insertions and deletions marks them as drivers of mutation; and their presence at recombination hotspots links them to the machinery of heredity itself. Each of these threads also points toward disease. Genome instability at repetitive regions is a recurring theme in cancer and in a range of neurological and developmental disorders, and a complete map of where G4s form, and where they mutate, gives researchers a new tool for interpreting disease-associated variation in parts of the genome that were previously unreadable.

The study also illustrates how technological progress reshapes entire fields. Just as the telomere-to-telomere assemblies revealed previously hidden genes and structural variants, they have now revealed a hidden structural genome, one written not in the order of bases alone but in the capacity of those bases to fold. With 88 diverse haplotypes already surveyed, the stage is set for broader population studies that could trace how G4 landscapes differ across ancestries, how they evolve under selection, and how their instability contributes to both human evolution and human disease. For a structure first noticed as an oddity of guanine chemistry, the G-quadruplex has come a long way, and the complete human genome, it turns out, is full of it.

Subject of Research: Genomic landscape and mutational dynamics of G-quadruplex DNA structures in the complete human genome

Article Title: Landscape and mutational dynamics of G-quadruplexes across the complete human genome

Article References: Landscape and mutational dynamics of G-quadruplexes across the complete human genome. (n.d.). https://doi.org/10.1186/s13059-026-04255-z

Image Credits: AI Generated

DOI: 10.1186/s13059-026-04255-z

Keywords: G-quadruplexes, human genome, telomere-to-telomere assembly, centromeres, ribosomal DNA, genomic instability, DNA methylation, meiotic recombination, PRDM9, long-read sequencing, insertions and deletions, genome biology

The core problem is one of identity. Short-read sequencing technologies, which dominate modern transcriptomics, produce fragments of RNA sequence that are typically only 100 to 150 nucleotides long. When those fragments originate from a HERV locus, they often match dozens or hundreds of nearly identical copies scattered across the genome with equal fidelity. Standard alignment algorithms, faced with equally good candidates, either discard the read as ambiguous or assign it arbitrarily, which means that expression signals from individual proviral loci are blurred together or lost entirely. Long-read sequencing platforms can span entire proviruses and resolve their origins, but they remain more expensive, lower throughput, and less commonly available, which limits their use in large clinical cohorts. The result has been a persistent blind spot: researchers know that HERV transcripts are present in a sample, but they frequently cannot say which specific retroviral elements are active, where they sit in the genome, or what regulatory logic controls them.

ERVmancer attacks this ambiguity by borrowing an idea from evolutionary biology. Rather than treating each HERV copy as an isolated entity, the framework organizes the 3,043 HERV loci it tracks into a phylogenetic structure, grouping related proviruses into clades that reflect their shared ancestry. When a short sequencing read cannot be uniquely assigned to a single locus, the method uses the phylogeny to make a principled probabilistic assignment: reads that are ambiguous among closely related copies within a clade contribute to the quantification of that clade, while information that distinguishes individual loci is preserved wherever the sequence allows it. In this way, ERVmancer can report expression at multiple resolutions simultaneously, from the individual proviral locus up through families and entire clades, giving researchers a hierarchical view of retroviral activity that mirrors the evolutionary reality of these elements rather than flattening it.

To find out whether this phylogeny-guided strategy actually works better than existing approaches, the team benchmarked ERVmancer using matched datasets in which the same biological samples had been sequenced with both short-read and long-read technologies. The long-read data served as a ground truth of sorts, since it can resolve individual loci directly. Across these comparisons, ERVmancer outperformed existing methods in both sensitivity, meaning its ability to detect genuine HERV expression, and specificity, meaning its ability to avoid false assignments. The evaluation extended across HERV families, and supplementary analyses using simulated sequencing data further probed how reads were distributed among internal phylogenetic clades, providing a detailed picture of where the method succeeds and where uncertainty remains. The tool has been released as a Bioconda package, a distribution channel widely used in the bioinformatics community, which should make adoption straightforward for laboratories analyzing RNA sequencing data.

The practical value of the framework was demonstrated in two disease contexts with very different biology. The first was multiple sclerosis, an autoimmune disease of the central nervous system in which HERV activity has long been suspected of playing a role, though pinning down which elements are involved has been difficult. The researchers applied ERVmancer to RNA sequencing data from spontaneous lymphoblastoid cell lines derived from patients with multiple sclerosis and from healthy controls, all of whom provided written informed consent under protocols reviewed by institutional review boards at the National Institutes of Health and The Wistar Institute. The analysis reproduced expression patterns that had previously been observed only with long-read sequencing, validating that the short-read approach could recover biologically meaningful signal, and it identified HERV expression patterns specific to the disease context, including variability within the ERV3 clade that the team confirmed independently using quantitative reverse-transcription PCR.

The second application turned to cancer, where the interplay between tumor suppressors and endogenous retroviruses has become an increasingly active frontier. Analyzing breast cancer cell lines from the Cancer Cell Line Encyclopedia, ERVmancer uncovered a striking regulatory relationship: the transcription factor p53, famous as the guardian of the genome for its role in suppressing tumors, appears to actively suppress the expression of the HERVH-LTR7 clade. This finding is significant because HERVH and its LTR7 long terminal repeat are known to be active in embryonic stem cells and in certain tumors, where they help maintain a stem-like, proliferative state. The observation that p53 exerts suppressive control over this clade suggests that one way the cell’s primary tumor suppressor keeps cells from reverting to a primitive, uncontrolled state is by silencing the viral remnants that normally fuel such programs in early development. Genome-wide analysis of p53 binding sites across diverse HERV clades and families, presented in the supplementary data, supports the breadth of this regulatory reach.

These findings illustrate why the ability to resolve HERV expression at the clade and locus level matters. A bulk measurement that lumps all HERVH activity together might obscure the fact that a specific subset of proviruses, defined by their evolutionary lineage and their regulatory sequences, is being selectively repressed or activated. In multiple sclerosis, distinguishing disease-associated expression within particular clades could eventually clarify whether endogenous retroviral transcripts act as drivers of pathology, as biomarkers of immune dysregulation, or as byproducts of inflammation. In oncology, understanding which retroviral elements a tumor reactivates, and which tumor suppressors keep them in check, could inform strategies to exploit viral-like antigens on cancer cells for immunotherapy, an approach that has already attracted considerable interest because HERV expression in tumors can generate targets that distinguish malignant cells from healthy tissue.

The methodological advance also speaks to a broader trend in genomics: the recognition that repetitive regions of the genome, long excluded from analysis or collapsed into consensus sequences, contain biology worth recovering in full. As short-read sequencing remains the workhorse of large-scale studies, including biobank-scale transcriptomics and clinical cohorts, tools that can extract locus-resolved information from inherently ambiguous data multiply the value of existing datasets without requiring new sequencing campaigns. ERVmancer’s hierarchical output means that researchers with different questions can query the same analysis at different resolutions, whether they care about a single provirus with a known regulatory function or about the aggregate behavior of an entire retroviral lineage across hundreds of samples.

The study, which was published open access on 25 September 2026 and led by corresponding authors Andrew Patterson and Noam Auslander, brought together expertise spanning virology, neuroimmunology, and computational biology, including contributions from Steven Jacobson of the NIH Neuroimmunology Branch and Paul Lieberman and Samantha Soldan of The Wistar Institute. It was supported by funding from the W. W. Smith Charitable Trust, the Michelson Medical Research Foundation, the United States Department of Defense, and multiple institutes of the National Institutes of Health. As with any computational method, its conclusions will need to be tested and extended by independent groups, and the authors note that the published version followed full peer review at Genome Biology.

For a field that has spent decades peering at the ghostly traces of ancient viruses through a fog of repetitive sequence, ERVmancer offers something close to a lens correction. By letting the evolutionary relationships among HERVs guide the interpretation of ambiguous reads, the framework turns a longstanding liability of short-read data into a tractable inference problem, and it delivers early biological dividends: confirmation that long-read-quality insight can be recovered from ordinary transcriptomic data, disease-specific expression signatures in multiple sclerosis, and evidence that p53, the cell’s most celebrated tumor suppressor, doubles as a warden of the viral archive within our DNA. As researchers begin applying the tool to other cancers, autoimmune conditions, and developmental systems, the regulatory grammar of the genome’s viral fossils may finally come into focus, revealing how infections from the deep past continue to script human biology today.

Subject of Research: A phylogeny-guided computational method for quantifying human endogenous retrovirus expression and regulation in health and disease

Article Title: ERVmancer: a phylogeny-guided framework for decoding human endogenous retrovirus regulatory mechanisms in health and disease

Article References: Patterson, A., Duong, B., Yoon, L., Foster, M., MacMullen, L., Wickramasinghe, J., Lucas, A., Srivastava, A., Jacobson, S., Murphy, M. E., Soldan, S. S., Lieberman, P. M., & Auslander, N. (2026). ERVmancer: a phylogeny-guided framework for decoding human endogenous retrovirus regulatory mechanisms in health and disease. Genome Biology. https://doi.org/10.1186/s13059-026-04287-5

Image Credits: AI Generated

DOI: 10.1186/s13059-026-04287-5

Keywords: human endogenous retroviruses, HERVs, ERVmancer, phylogenetics, RNA sequencing, short-read mapping, multiple sclerosis, breast cancer, p53, HERVH-LTR7, transposable elements, computational genomics

The research, led by Alexandra Zidovska, an associate professor in NYU’s Department of Physics, addresses a gap that has persisted since the completion of the Human Genome Project. That monumental effort, launched in 1990 after decades of genetic breakthroughs, delivered a sequence of the human genome, but the physical rules governing how that genome is organized inside the cell nucleus remain an active and unsettled area of research. Among the unanswered questions is the role of gravity, a constant mechanical stress on everything that lives. To probe it, Zidovska and her colleagues reasoned, gravity must be removed as a factor, which is precisely what their custom-built laboratory apparatus was designed to do.

The stakes of the question are considerable. The human genome is a one-dimensional sequence encoded in roughly two meters of DNA molecules, packed with extraordinary efficiency into three dimensions inside a cell nucleus barely 10 micrometers across, about the width of a silk fiber. This compact hierarchical structure is directly linked to genomic function, and deviations from it can contribute to human diseases, including cancer, as well as developmental afflictions. But the physical principles that maintain this organization are not well understood, and it has remained unknown whether gravity contributes to preserving it, or whether its absence might induce genomic aberrations.

To find out, the NYU team designed and built a random positioning machine tailored for imaging the human genome in live cells. The device rotates dishes of living cells along two independent axes, following a three-dimensional rotational path that averages out the direction of the gravitational vector, producing a simulated microgravity, a condition of weightlessness created on Earth. The principle resembles that of the larger machines astronauts use in their training. A crucial preparatory step was the elimination of air bubbles from the cell cultures, since even small bubbles can interfere with delicate measurements during rotation.

Rotation alone, however, introduces a confounding problem. Turning a dish of cells through three dimensions inevitably stirs the surrounding fluid, generating flows that do not exist in genuine zero gravity outside Earth. Previous studies of simulated microgravity have often been obscured by such flows and by the cell aggregates they can induce. The researchers therefore developed novel 3D rotational algorithms that not only simulated microgravity but minimized fluid flow generation, and they created additional algorithms to investigate, separately, what flows themselves do to cells and their genomes. Together, these advances sharply reduced both fluid motions and cell clumping, allowing the team to isolate the effects of weightlessness from the artifacts of the simulation.

The experimental design then became a matter of careful comparison. The scientists exposed one set of cells to simulated microgravity with minimized flows, another to strong fluid flows, and a control group to neither condition, each exposure lasting 24 hours. They then deployed sophisticated imaging techniques, recording streams of images for detailed physical analysis. The measurements tracked changes in cell shape and volume, the shape and volume of the cell nucleus, the nuclear envelope, the genome itself, and the nucleolus, the largest liquid condensate within the nucleus. The team specifically examined changes in the genome’s organization and dynamics, as well as any DNA damage induced by the simulated conditions.

The results paint a picture of remarkable resilience, with telling exceptions. Cells exposed to flows became elongated in shape, but cells in simulated microgravity kept their form. The volume of the cell nucleus increased after exposure to weightlessness, indicating that on Earth gravity normally acts to diminish it. Despite that change in volume, the thickness and structure of the nuclear envelope, the membrane that encases the genome, remained unchanged, suggesting gravity has minimal influence on those architectural traits. Most strikingly, the genome itself maintained its physiological organization and motions throughout the day-long exposure to simulated microgravity, and the technique produced no detectable DNA damage, whereas exposure to flows did damage DNA. The nucleolus became smoother after exposure to either condition.

Our data show that the genome, its organization, and dynamics are incredibly robust and seem unaffected by gravity, or lack thereof, after 24 hours, Zidovska observed. In the same way, she noted, the results suggest that the physical organization of the human genome may undergo minimal changes in outer space during comparable timescales. The subtlety of the changes, however, comes with a caveat that the researchers themselves emphasize: small perturbations could amplify over time and eventually affect cell physiology, and longer exposures, particularly in the space environment where actual DNA damage occurs, could produce effects not seen in a single day of simulation.

The findings arrive at a moment of renewed ambition in human spaceflight, with plans for extended missions to the Moon and Mars raising urgent questions about how the body copes with weightlessness. How will the human genome be affected in outer space, Zidovska asked in describing the motivation for the work. By demonstrating that the genome’s organization and dynamics withstand a day of weightlessness on Earth, the study offers reassurance for short-duration exposures while framing the longer-term questions that future experiments, and future astronauts, will need to answer. It also provides the scientific community with a validated toolkit, combining imaging-compatible microgravity simulation with flow-suppressing algorithms, for disentangling gravitational effects from mechanical artifacts in cell biology.

Beyond its implications for space travel, the work speaks to a more fundamental mystery: the physics of the genome. Understanding which external forces shape, or fail to shape, the two meters of DNA coiled inside each of our cells touches on questions of disease, development, and the basic mechanics of life. The paper’s co-authors included Nikitas Kanellakopoulos, a doctoral student at NYU; undergraduates Manav Patel, Brandon Sato, and Melaina Lawrence; and Leif Ristroph, an associate professor at NYU’s Courant Institute School of Mathematics, Computing, and Data Science. The research was supported in part by grants from the National Science Foundation and the National Institutes of Health. For now, the message is one of quiet stability: the architecture of our genetic material, refined over billions of years, appears built to hold its shape whether it rests on Earth or floats among the stars, at least for a day.

Subject of Research: The effects of simulated microgravity and fluid flows on the organization, dynamics, and integrity of the human genome in living cells

Article Title: Scientists uncover gravity’s impact on the human genome

Article References: Scientists uncover gravity’s impact on the human genome. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: microgravity, human genome, cell nucleus, chromatin, DNA damage, nuclear envelope, nucleolus, random positioning machine, space biology, biophysics, Science Advances, New York University