New tool Dory analyzes differences in image-based chromatin tracing data

Researchers at the University of Virginia and Yale School of Medicine have introduced a new computational method designed to detect structural differences in the three-dimensional organization of the genome between groups of cells, addressing one of the most pressing analytical gaps in modern genomics. The tool, called Dory, is described in a study published in Genome Biology and offers researchers a statistically rigorous, user-friendly framework for analyzing image-based chromatin tracing data—one of the fastest-growing areas in the study of how genomes fold inside the cell nucleus.

The spatial organization of the genome is now widely recognized as a fundamental driver of cell identity and gene regulation. Within every nucleus, the roughly two meters of DNA packaged into each human cell folds into complex, hierarchical structures that bring distant regulatory elements, such as enhancers, into physical contact with the genes they control. These contacts can determine which genes are active and which remain silent, and disruption of this folding architecture has been implicated in developmental disorders, aging, and diseases such as cancer. Understanding how these structures change between cell types, states, or disease conditions is therefore a central goal of contemporary genomics.

Chromatin tracing has emerged as a powerful technology for visualizing these structures directly. Unlike traditional genome-wide assays such as Hi-C, which infer chromatin contacts from populations of millions of cells, chromatin tracing uses sequential fluorescence imaging to resolve the physical positions of specific genomic regions inside individual cells. Each imaged cell yields a set of coordinates—traces—that describe how chromatin is folded in that particular nucleus. This single-cell resolution is invaluable, because it allows researchers to see the genuine diversity of genome conformations within a tissue rather than an averaged picture.

However, the rapid adoption of chromatin tracing has outpaced the development of computational tools for analyzing it. Most existing analytical approaches focus on summarizing the structures within a single group of cells, such as computing average distances between genomic loci. Far fewer methods can robustly answer the question that biologists most often ask: how does the spatial organization of a given genomic region differ between two conditions—healthy versus diseased cells, for example, or two distinct developmental stages? Without principled statistical methods for this differential analysis, researchers have had to rely on ad hoc comparisons that are vulnerable to technical artifacts and small sample sizes.

Dory was developed specifically to fill this gap. At its core, the method quantifies pairwise spatial distances among the genomic regions being traced. For each pair of traced loci, Dory builds a statistical description of the distances measured across all cells in one group and compares it to the equivalent distribution in the other group. By doing so, it captures differences that simple averages miss: a pair of loci may show the same mean distance in both conditions while exhibiting dramatically different variability, or a contact that is frequent and stable in one group may be rare and diffuse in another. Dory detects these significant structural differences and condenses them into a differential score matrix—a comprehensive table of scores for every pair of regions, indicating which spatial relationships have changed and by how much.

A critical strength of the approach is its attention to statistical validity. Measurements from chromatin tracing are inherently noisy, subject to imaging limitations, and often derived from limited numbers of cells. Dory’s statistical framework is designed to distinguish true biological differences in chromatin architecture from variation that arises from sampling and technical noise. The authors demonstrated the robustness of their method through extensive validation analyses, including comparisons in which the whole set of traces was subsampled to test whether the identified differences remained stable, and comparisons against bootstrapping-based assessments of differential region pairs. These analyses showed that Dory’s differential scores are reproducible and reliable even when data volume varies.

The researchers applied Dory to multiple chromatin tracing datasets, and the results provide compelling evidence for its biological utility. In one set of analyses, the method identified chromatin structural changes associated with alterations in A/B compartments—large-scale genome domains that divide chromosomes into transcriptionally active (A) and inactive (B) regions. The ability to connect localized changes in spatial distances with these broad compartmental shifts gives researchers a new way to link large-scale genome architecture to fine-grained structural rearrangements visible at the single-cell level. The team’s supplementary analyses included a schematic interpretation of an “ABchange” score and permutation tests establishing the statistical correlation between Dory’s differential scores and compartment alterations, further grounding the method’s outputs in well-established concepts of genome organization.

Perhaps most strikingly, Dory uncovered changes in promoter–enhancer interactions linked to differential gene expression. Promoter–enhancer contacts are among the most functionally consequential spatial relationships in the genome, as they directly connect regulatory DNA elements to the genes whose activity they modulate. In analyses involving mouse fetal liver, the method traced the spatial relationship between the promoters of candidate cancer progression drivers and their surrounding genomic regions, identifying regulatory candidates whose expression changes could explain the structural shifts. The datasets examined included promoters of genes such as Myc, Anxa8, Oit1, and Fhit—genes with well-known roles in cancer and cellular differentiation. By identifying “CloserRegions,” genomic neighborhoods that move closer to a promoter’s region in one condition compared with another, and cross-referencing these with the binding sites of transcriptional regulators, the framework points directly to candidate mechanisms that could drive the observed architectural and expression changes.

The implications for disease research, particularly cancer, are substantial. The team behind Dory has a long-standing focus on mapping the 3D genome in cancer, and the new method integrates naturally with chromatin tracing workflows developed to study how tumor genomes reorganize as cells progress toward malignancy. By providing a quantitative differential score for every region pair, Dory allows researchers to systematically compare chromatin architecture across tumor stages, between tumor and normal cells, or between cells responding to therapy and those that do not. Such comparisons could reveal how specific structural rearrangements enable oncogenes to escape regulation or bring enhancers into contact with genes they would normally never touch.

Dory is also designed with usability in mind. The authors describe it as a robust and user-friendly tool for quantitative analysis of image-based 3D genome data, an important consideration given that many research groups generating chromatin tracing data lack the computational expertise to develop bespoke statistical pipelines. By packaging the differential analysis into an accessible method, the team lowers the barrier for laboratories across genomics, cell biology, and cancer research to extract meaningful insights from their imaging experiments.

The study was conducted by Zhaoxia Ma, Shengyuan Wang, and Chongzhi Zang at the University of Virginia’s Department of Genome Sciences and Department of Biochemistry and Molecular Genetics, together with Miao Liu and Siyuan Wang of Yale School of Medicine’s Departments of Genetics and Cell Biology. The work received support from the National Institutes of Health, including grants R35GM133712, R01CA292936, UH3CA268202, and R01HG013503, as well as funding from the University of Virginia Comprehensive Cancer Center.

As chromatin tracing and related imaging-based genomics technologies continue to spread through the research community, the demand for principled statistical tools will only grow. Dory arrives at a moment when single-cell 3D genome analysis is poised to become a standard component of studies on gene regulation, development, and disease. By turning stacks of fluorescence images into statistically validated maps of structural change, it transforms a qualitative impression—that the genome “looks different” between conditions—into precise, quantitative, and interpretable biology. The method’s ability to connect changes in physical DNA distances to compartments, enhancer–promoter contacts, and gene expression promises to make image-based 3D genome analysis an increasingly central tool in the effort to understand how the architecture of our genomes shapes who our cells become.