The Shape of Life: A New 4D Atlas Reveals How the Genome Folds and Functions

Abstract

For over two decades, the Human Genome Project has provided the linear sequence of life—a string of three billion letters that encodes the instructions for a human being. Yet, within the nucleus of a living cell, this code is far from linear. It is folded, looped, and compacted into a complex three-dimensional structure that shifts dynamically over time. This spatiotemporal organization, known as the "4D nucleome," is the physical operating system that regulates gene expression. In a landmark study published in Nature in late 2025, the 4D Nucleome Consortium unveiled the most comprehensive atlas of this architecture to date. By integrating data from diverse genomic assays and employing advanced artificial intelligence, the researchers have not only mapped the static structure of chromosomes but also revealed how this structure rewires itself to sustain life. This report explores the consortium's findings, from the discovery of over 140,000 chromatin loops to the development of predictive models that link DNA folding to human disease.

Introduction: The "Parts List" Meets the "Instruction Manual"

When the human genome was first sequenced, it was hailed as the "book of life." However, having the letters of a book does not equate to understanding the story, especially if the pages are crumpled into a tight ball. In human cells, approximately two meters of DNA are packed into a nucleus roughly 10 micrometers in diameter.¹ This packaging is not random; it is a highly ordered, hierarchical system that determines accessibility. Genes buried deep within the fold are silenced, while those on the surface or in specific loops are accessible to the cellular machinery that reads them.

This structure is dynamic, changing as cells divide, differentiate, and respond to stress. Thus, the genome exists in four dimensions: the three spatial dimensions plus time. Understanding this 4D nucleome is critical because errors in folding are increasingly linked to developmental disorders, cancers, and other complex diseases.²

In December 2025, the 4D Nucleome (4DN) Consortium, a massive collaborative effort funded by the National Institutes of Health, published a flagship paper titled "An integrated view of the structure and function of the human 4D nucleome".² Led by principal investigators Job Dekker (UMass Chan Medical School) and Feng Yue (Northwestern University), the team performed an exhaustive analysis of two widely used cell lines: H1 human embryonic stem cells (hESCs) and HFFc6 immortalized foreskin fibroblasts.⁵ Their goal was to move beyond simple maps and create a predictive understanding of how the genome folds and functions.

The Multimodal Toolkit: Beyond Hi-C

For years, the gold standard for mapping the 3D genome was Hi-C (High-throughput Chromosome Conformation Capture). This technique works by chemically "freezing" (crosslinking) interactions between DNA segments, cutting the DNA, and then gluing the interacting pieces together to sequence them. While revolutionary, Hi-C has limitations, particularly in resolution and the ability to detect simultaneous interactions among multiple DNA strands.¹

To build a complete atlas, the 4DN Consortium did not rely on a single method. Instead, they applied a battery of complementary assays to the same cell lines, allowing for a rigorous benchmarking of genomic technologies. This "multimodal" approach revealed that different assays perceive the genome through different lenses, each capturing unique features of the nuclear architecture.¹

The Hierarchy of Assays

The researchers categorized these methods based on their specific strengths. Micro-C, a higher-resolution variant of Hi-C that uses an enzyme (micrococcal nuclease) rather than chemicals to fragment DNA, emerged as the superior tool for detecting fine-scale structures. The study found that Micro-C could identify chromatin loops and the boundaries of Topologically Associating Domains (TADs) with significantly greater precision than traditional Hi-C.¹

Conversely, methods like SPRITE (Split-Pool Recognition of Interactions by Tag Extension) and GAM (Genome Architecture Mapping) excelled at a different scale. Unlike capture-based methods that see pairwise contacts (DNA point A touches DNA point B), SPRITE and GAM can detect clusters of interactions where multiple DNA segments come together simultaneously. The study confirmed that these methods are most effective for mapping large-scale compartmentalization—the segregation of the genome into active "A" and inactive "B" zones—and for detecting long-range interactions that span vast genomic distances.¹

Table 1: Comparative Strengths of Genomic Assays

Assay Type	Key Methods Analyzed	Primary Strengths Identified	Best Use Case
Pairwise Capture	Micro-C	Ultra-high resolution; precise detection of loop anchors and domain boundaries.	Mapping specific chromatin loops and TADs.
Standard Capture	Hi-C	Robust, genome-wide coverage; industry standard for general topology.	General mapping of A/B compartments and TADs.
Enrichment-Based	ChIA-PET, PLAC-seq	Targets specific proteins (e.g., RNA Pol II, CTCF); high signal-to-noise ratio for regulatory interactions.	Identifying enhancer-promoter loops linked to gene expression.
Multi-way / Ligation-Free	SPRITE, GAM	Detects multi-way clusters and long-range inter-chromosomal interactions; avoids ligation bias.	Mapping large-scale compartments and nuclear bodies (e.g., speckles).
Nuclear Positioning	TSA-seq, DamID	Maps DNA distance relative to fixed nuclear landmarks (nuclear lamina, speckles).	Determining the radial position of genes within the nucleus.

¹

The New Atlas: 140,000 Loops and Nested Compartments

By integrating these diverse datasets, the consortium generated the most detailed catalog of 3D genomic features ever produced for human cells.

The Loop Catalog

One of the headline findings was the identification of approximately 140,000 chromatin loops in each of the cell types studied.⁸ These loops are fundamental functional units of the genome. They often extrude outward to bring a gene promoter into physical contact with an enhancer—a regulatory switch that might be located millions of base pairs away on the linear strand.

The integration of data revealed distinct classes of loops. "Structural" loops, anchored by the protein CTCF and the ring-shaped cohesin complex, tend to be stable and are best detected by Micro-C and Hi-C. These loops form the scaffold of the genome. In contrast, "functional" loops, which are associated with active transcription and RNA Polymerase II, were highlighted by enrichment assays like ChIA-PET and PLAC-seq.¹ The study showed that while structural loops often persist across cell types, functional loops are highly specific, forming only when a particular gene needs to be activated.

Microcompartments and Phase Separation

The classic view of the genome divides chromatin into two massive compartments: Compartment A (active, open, gene-rich) and Compartment B (inactive, closed, gene-poor). The 4DN study added a new layer of complexity to this model: microcompartments.¹

Using high-resolution Micro-C data, the researchers observed that within the larger A/B compartments, chromatin further segregates into smaller, nested interactions. These microcompartments likely represent the aggregation of specific regulatory elements or gene clusters that share similar biochemical properties. This finding supports the hypothesis that phase separation—the physical phenomenon where liquids of different densities separate (like oil and vinegar)—plays a role in organizing the genome, creating concentrated "droplets" of transcriptional activity.¹

Structure Meets Function: The Housekeeping Paradox

A central question of the 4D Nucleome Project is how physical folding relates to biological function, specifically DNA replication and gene transcription. The comparison between embryonic stem cells (hESCs) and fibroblasts (HFFc6) yielded a surprising insight regarding "housekeeping genes."

Rewiring for Stability

Housekeeping genes are essential for basic cellular survival and are active in virtually every cell type. Because their expression is constant, it was previously assumed that their regulatory environment would be static. However, the consortium's maps revealed the exact opposite.

The researchers found that while the expression of housekeeping genes remains steady, the physical connections sustaining that expression change dramatically. The same housekeeping gene often interacts with a completely different set of enhancers in a stem cell compared to a fibroblast.⁸ This suggests that the genome actively "rewires" itself to maintain stability. As the cell differentiates and the landscape of available enhancers changes, housekeeping genes adapt by shifting their 3D contacts to find new sources of regulatory activation. This demonstrates a key principle of the 4D nucleome: functional stability is achieved through structural flexibility.

Replication Timing and Nuclear Position

The study also solidified the link between a gene's location in the nucleus and the timing of its replication. The genome does not replicate all at once; some parts are copied early in the cell cycle (S-phase), while others are copied late. The integrated models showed that regions located in the active A compartment, particularly those near nuclear speckles (hubs of RNA processing), replicate early. Conversely, regions in the B compartment, often tethered to the nuclear lamina (the rigid edge of the nucleus), replicate late.¹¹ This confirms that the 3D spatial arrangement of the genome is intimately tied to the temporal control of DNA replication.

The Predictive Turn: AI and the "Dark Genome"

Perhaps the most transformative aspect of the study is the shift from describing the genome to predicting it. The consortium demonstrated that by training deep learning models (such as architectures similar to Akita) on their high-resolution Micro-C data, they could predict the 3D folding of a genome based solely on its DNA sequence.¹

Decoding Non-Coding Variants

This predictive capability addresses a major bottleneck in modern genetics. Genome-Wide Association Studies (GWAS) have identified thousands of genetic variants (mutations) associated with diseases ranging from diabetes to schizophrenia. However, over 90% of these variants lie in the "non-coding" regions of the genome—the vast stretches of DNA that do not code for proteins. Determining how a mutation in "junk DNA" causes disease has been notoriously difficult.

The 4DN researchers showed that many of these non-coding variants function by altering the 3D folding of the genome. A single letter change (SNP) can disrupt a binding site for CTCF, causing a loop to collapse or boundaries to shift. This structural error can cause a gene to lose contact with its enhancer or, more dangerously, to "hijack" a distant enhancer, leading to aberrant expression.⁵

By using their AI models, the team performed in silico mutagenesis—digitally inserting mutations into the sequence and asking the computer to predict the resulting change in folding. They successfully predicted how specific variants associated with diseases would alter chromatin loops, providing a mechanism for pathogenicity that would be invisible to standard sequencing.¹ This creates a new pipeline for "genomic pathology," where a patient's unique genetic sequence can be analyzed to reveal structural defects in their nucleome.

Conclusion: A Blueprint for Future Discovery

The publication of "An integrated view of the structure and function of the human 4D nucleome" represents a maturation of the field of nuclear architecture. We have moved from low-resolution sketches of "chromosome territories" to high-definition, dynamic blueprints of the nuclear interior.

The 4DN Consortium has provided the scientific community with three indispensable resources:

1. A Benchmark: A clear, evidence-based guide on which genomic assays to use for specific biological questions.

2. An Atlas: A massive, integrated database of loops, domains, and compartments that serves as a reference for the healthy human genome.

3. A Crystal Ball: Computational tools that allow researchers to predict how genetic alterations will ripple through the 3D structure of the nucleus.

As we look to the future, these insights are poised to revolutionize the diagnosis and treatment of genetic diseases. We now understand that the genome is not merely a static library of code but a dynamic, folding machine. When that machine misfolds, disease can follow—and for the first time, we have the maps necessary to see it happening.

Glossary of Key Terms

• 4D Nucleome: The organization of the genome in three-dimensional space (3D) and how it changes over time (the 4th dimension).

• Hi-C / Micro-C: Genomic assays used to map 3D contacts between DNA segments. Micro-C offers higher resolution by using enzymes rather than chemicals to fragment DNA.

• Chromatin Loop: A structure where two distant pieces of DNA are held together, often bringing enhancers and promoters into proximity.

• TAD (Topologically Associating Domain): A "neighborhood" of DNA that interacts frequently with itself and is insulated from neighboring regions.

• Compartment A/B: The large-scale segregation of the genome into active (A) and inactive (B) chromatin regions.

• CTCF & Cohesin: The primary protein architects of the 3D genome. Cohesin extrudes loops, and CTCF acts as the "stop sign" or anchor.

• GWAS (Genome-Wide Association Study): An observational study of a genome-wide set of genetic variants in different individuals to see if any variant is associated with a trait.

Works cited

1. An integrated view of the structure and function of the human 4D nucleome - PMC - NIH, accessed January 16, 2026, https://pmc.ncbi.nlm.nih.gov/articles/PMC11526861/

2. Scientists map the human genome in 4D - EurekAlert!, accessed January 16, 2026, https://www.eurekalert.org/news-releases/1110886

3. Simpson Querrey Institute for Epigenetics: Feinberg School of Medicine - Northwestern University, accessed January 16, 2026, https://www.feinberg.northwestern.edu/sites/epigenetics/index.html

4. 今週号 - Nature Asia, accessed January 16, 2026, https://go.nature.com/2HFKE2t

5. A hidden world inside DNA is finally revealed | ScienceDaily, accessed January 16, 2026, https://www.sciencedaily.com/releases/2026/01/260107225541.htm

6. An integrated view of the structure and function of the human 4D nucleome, accessed January 16, 2026, https://edoc.mdc-berlin.de/id/eprint/26014/

7. An Integrated View of The Structure and Function of The Human 4D Nucleome | PDF | Gene, accessed January 16, 2026, https://www.scribd.com/document/973867774/s41586-025-09890-3

8. The Ultimate Genetic Map: The 4D Nucleome Program Reveals DNA's 4D Secrets, accessed January 16, 2026, https://commonfund.nih.gov/4DNucleome/highlights/ultimate-genetic-map-4d-nucleome-program-reveals-dnas-4d-secrets

9. An integrated view of the structure and function of the human 4D nucleome - PubMed, accessed January 16, 2026, https://pubmed.ncbi.nlm.nih.gov/41407856/

10. 4D Nucleome - biorXiv Feeds, accessed January 16, 2026, https://connect.biorxiv.org/relate/feed/66

11. An integrated view of the structure and function of the human 4D nucleome - ResearchGate, accessed January 16, 2026, https://www.researchgate.net/publication/398781905_An_integrated_view_of_the_structure_and_function_of_the_human_4D_nucleome

12. An integrated view of the structure and function of the human 4D nucleome, accessed January 16, 2026, https://researchprofiles.ku.dk/en/publications/an-integrated-view-of-the-structure-and-function-of-the-human-4d-/

13. Scientists Uncover Hidden 3D Architecture of Human DNA - GreekReporter.com, accessed January 16, 2026, https://greekreporter.com/2025/12/24/3d-human-dna/

14. An integrated view of the structure and function of the human 4D, accessed January 16, 2026, https://www.semanticscholar.org/paper/An-integrated-view-of-the-structure-and-function-of-Dekker-Oksuz/376c7aa7da9195c4ec11c0015d608d47cf9af797

15. Systematic maps reveal how human chromosomes are organized - ResearchGate, accessed January 16, 2026, https://www.researchgate.net/publication/398791927_Systematic_maps_reveal_how_human_chromosomes_are_organized

16. Volume 649 Issue 8097, 15 January 2026, accessed January 16, 2026, https://evol.mcmaster.ca/brian/journals//Nature_WedJan14.html

17. Department of Systems Biology - UMass Chan Medical School, accessed January 16, 2026, https://www.umassmed.edu/dsb/