Spatial chromatin architecture and accessibility co-profiling of mammalian tissues

Ping Wang; Juan Wang; Qixuan Wang; Mark W Youngblood; Yang Cheng; Bo Tao; Josiah Hiu-Yuen Wong; Yu Luan; Shaojun Yu; Alok Swaroop; Khizar R Nandoliya; Hinda Najem; Yihao Fu; Kathleen McCortney; Di Zhang; Yarui Diao; Amy B Heimberger; Adam M Sonabend; Craig M Horbinski; Rong Fan; Feng Yue

doi:10.1038/s41592-026-03217-4

. 2026 Sep 1;23(10):2015–2028. doi: 10.1038/s41592-026-03217-4

Spatial chromatin architecture and accessibility co-profiling of mammalian tissues

Ping Wang ^1,^#, Juan Wang ^1,^#, Qixuan Wang ^1,^#, Mark W Youngblood ^1,^2,^#, Yang Cheng ¹, Bo Tao ³, Josiah Hiu-Yuen Wong ¹, Yu Luan ⁴, Shaojun Yu ¹, Alok Swaroop ^1,^5,⁶, Khizar R Nandoliya ², Hinda Najem ², Yihao Fu ¹, Kathleen McCortney ², Di Zhang ³, Yarui Diao ⁷, Amy B Heimberger ², Adam M Sonabend ², Craig M Horbinski ^2,⁸, Rong Fan ^3,^✉, Feng Yue ^1,^6,^✉

PMCID: PMC13645645 PMID: 42680843

Abstract

Cellular function depends on the spatial organization of cells and biomolecules within the tissue microenvironment. Advances in spatial omics have enabled profiling of molecular features such as transcriptome, proteome and epigenome, and there has been rapid progress of imaging-based approaches to study spatial three-dimensional (3D) genome organization. Here we present Spatial-ATAC-Hi-C, a microfluidic‑based platform for genome-wide, spatially resolved joint-profiling of 3D genome organization and chromatin accessibility on tissue slides. Applied to mouse and human brains, Spatial-ATAC-Hi-C revealed distinct chromatin architecture and gene regulatory programs in neuronal and non-neuronal populations in their native tissue context. In glioblastoma and astrocytoma samples, we detected spatially resolved 3D genome alterations, copy number variations and structural variations across tumor regions, revealing clinically relevant oncogenic events and clonal heterogeneity. By co-profiling of genome architecture and chromatin accessibility while preserving tissue architecture, Spatial-ATAC-Hi-C provides a powerful tool for studying spatial gene regulation in human biology and disease.

Subject terms: Genomics, Genomic analysis, Sequencing, Chromatin, Cancer genomics

Spatial-ATAC-Hi-C enables the profiling of both chromatin accessibility and organization in a spatially resolved manner, as demonstrated on mouse brain and human glioblastoma and astrocytoma samples.

Main

Nearly all complex biological processes are shaped by spatial context. The precise location and neighboring interactions of individual cells influence their proper functions and development in health and disease^1,2. Local tissue microenvironments, consisting of neighboring cells, extracellular matrix and paracrine signaling molecules, can impact diverse biomolecular processes, including protein localization, chromatin accessibility and histone modifications.

To understand cellular biology in its native environment, progress has been made recently in spatial genome techniques. Based on the target molecules, they can be categorized into spatial transcriptomics, epigenomics, genomics, proteomics and multi-omics approaches. Spatial transcriptomics (ST) technologies, including ST³, Slide-seq^4,5, DBiT-seq⁶, Seq-scope⁷, pixel-seq⁸ and stereo-seq⁹, enable high-resolution measurement of genome-wide gene expression through solid-phase-based spatial approaches or microfluidics-based barcoding strategies. These platforms have achieved varying spatial resolutions, ranging from 55 μm to 2 μm, with some techniques reaching single-cell or subcellular resolution. Spatial epigenomic techniques, such as spatial-ATAC-seq¹⁰ and spatial-CUT&Tag¹¹, allow users to spatially profile chromatin accessibility and histone modifications (H3K27ac and H3K27me3), and study gene regulatory events within local tissue regions. In particular, Slide-DNA-seq¹² provides an opportunity to interrogate the copy number alterations of tumor clones within their spatial tissue context. Spatial proteomics, including spatial-CITE-seq¹³ and PLATO¹⁴, facilitate high-resolution mapping of hundreds to thousands of proteins using a microfluidics-based device. Additionally, spatial multi-omics platforms, such as spatial-Mux-seq¹⁵, spatial ARP-seq and spatial CTRP-seq¹⁶ have emerged to enable simultaneous spatial co-profiling of the epigenome, transcriptome, and proteome, enhancing our ability to understand the complex interplay among different molecular layers within the tissue microenvironment.

The 3D organization of the genome is a critical layer of gene regulation, playing an important role in key biological processes, including development^17,18 and immune response^19–21, as well as disease initiation and progression^22–25. Historically, Hi-C²⁶ and other 3C-based techniques (such as HiChIP²⁷, ChIA-PET²⁸, Micro-C²⁹, PLAC-seq³⁰, Trac-looping³¹ and HiCAR³²), have been frequently used to reveal hierarchical features of the nuclear architecture, including compartments²⁶, topologically associating domains (TADs)^33,34 and chromatin loops³⁵. Building on this, recent single-cell technologies (for example, HiRes³⁶, scCARE-seq³⁷, GAGE-seq³⁸, LiMCA³⁹, droplet Hi-C⁴⁰, snm3c-seq⁴¹, scMicro-C⁴² and scSPRITE⁴³) have provided insights into cell-to-cell heterogeneity in higher-order genome structure. In addition, Ma et al. developed the Spatial-SPRITE⁴⁴, which is based on microfluid equipment and potentially can generate spatially mapped SPRITE data.

We have also seen rapid progress in imaging-based approaches, such as DNA-MERFISH^45,46, DNA seqFISH+^47–49, chromatin tracing^50,51, MINA⁵², ORCA⁵³ and Hi-M⁵⁴ and in situ genome sequencing techniques (IGS⁵⁵ and ExIGS⁵⁶), which have made it possible to visualize subcellular chromatin folding within individual nuclei. While the aforementioned imaging-based techniques provide high-resolution interactions for genome folding and provide insightful discoveries, most of them are focused on selected loci or regions. Therefore, developing sequencing-based technologies for genome-wide, high-resolution, spatially resolved mapping of 3D genome organization will greatly complement the imaging-based approaches, and is essential for elucidating how genome folding and transcriptomic regulation are orchestrated within the native spatial microenvironment.

In this study, we developed Spatial-ATAC-Hi-C, a technology that can jointly profile 3D genome organization and chromatin accessibility within native tissue context. By applying Spatial-ATAC-Hi-C in adult mouse brains, human cerebellum tissues, we delineated cell-type- and region-specific 3D genome organization landscapes. To gain clinical insights, we performed Spatial-ATAC-Hi-C in glioblastoma (GBM) and astrocytoma patient samples and uncovered aberrant and region-specific chromatin interactions that shape glioma gene regulation. Finally, we showed that Spatial-ATAC-Hi-C could be used to robustly identify large-scale genome abnormalities, such as structural variants (SVs) and copy number variants (CNVs), and spatially resolved distinct tumor clones harboring these genomic events, demonstrating its value in the study of tumor heterogeneity and evolution.

Results

Development of Spatial-ATAC-Hi-C technology

The overall design of the Spatial-ATAC-Hi-C technology is described in Fig. 1a. In brief, the spatial information is preserved through barcoded oligonucleotides and microfluidic-based indexing, utilizing the same device as previously established for spatial multi-omics assays^{6,10,11,13,57}. Fresh frozen tissue sections are immobilized onto microarray grade slides and crosslinked with formaldehyde to preserve native chromatin structure. After nuclei permeabilization, enzyme digestion is performed followed by in situ proximity ligation of chromatin DNA fragments on the tissue slide. Next, Tn5 transposition is performed directly on tissue sections using an in-house assembled Tn5 transposome complex to fragment the proximity-ligated chromatin DNA. Horizonal barcodes A (A1–A50) and vertical barcodes B (B1–B50) are sequentially delivered across the tissue surface using microchannels, and ligated to the universal ligation linker (Supplementary Tables 1 and 2). This creates a two-dimensional grid of spatially barcoded tissue pixels (n = 2,500). After reverse crosslinking and purification, the biotin-ligated chromatin fragments are enriched for sequencing library preparation. Finally, based on the barcode information, we can build the chromatin interaction maps for each pixel from the original tissue slides (Fig. 1b).

Quality assessment and benchmarking of Spatial-ATAC-Hi-C

To assess whether Spatial-ATAC-Hi-C accurately captures 3D genome architecture and ATAC-seq profiles, and whether barcoding induces biases in the Hi-C signals, we performed the following control experiments: bulk in situ Hi-C, bulk ATAC-seq and Spatial-ATAC-Hi-C without barcoding (Spatial-no-barcoding) using adjacent mouse brain tissue sections.

We first compared the 3D genomic features between Spatial-ATAC-Hi-C, in situ Hi-C and Spatial-no-barcoding (Supplementary Table 3). We observed that the Hi-C contact maps, PC1 values and insulation scores were highly consistent across the three datasets (Fig. 1c and Extended Data Fig. 1a–e). These results demonstrated that Spatial-ATAC-Hi-C can accurately capture the 3D genome architecture and is not affected by the barcoding step.

Extended Data Fig. 1 — **a-b**. Scatterplots comparing the PC1 values (a) and insulation scores (b) computed from Spatial-ATAC-Hi-C versus in situ Hi-C on adjacent slides of mouse brain tissue section R8. PCC, Pearson correlation coefficient. **c-d**. Scatterplots comparing the PC1 values (c) and insulation scores (d) computed from Spatial-ATAC-Hi-C versus Spatial-no-barcoding on adjacent slides of mouse brain tissue section R8. PCC, Pearson correlation coefficient. e. Hi-C maps (upper panel) and insulation score tracks (lower panel) of Spatial-ATAC-Hi-C in mouse brain R8 sample, in situ Hi-C and Spatial-no-barcoding from adjacent mouse R8 tissues at genomic region chr2: 106 – 118 Mb. **f-g**. Scatterplots comparing the PC1 values (f) and insulation scores (g) computed from Spatial-ATAC-Hi-C of mouse brain tissue section R6 versus Dip-C. PCC, Pearson correlation coefficient. h. Chromatin contact matrices of Spatial-ATAC-Hi-C (upper triangle) and Dip-C (lower triangle) at 1 Mb resolution. Bottom tracks: the PC1 tracks from Spatial-ATAC-Hi-C (upper) and Dip-C (lower). i. Chromatin contact matrices of Spatial-ATAC-Hi-C (upper triangle) and Dip-C (lower triangle) at 100Kb resolution. Bottom tracks: the insulation score tracks from Spatial-ATAC-Hi-C (upper) and Dip-C (lower).

The tagmentation step in Spatial-ATAC-Hi-C is performed in situ on fixed tissue sections, which differs from the standard Hi-C protocol, where library preparation is performed on extracted genomic DNA. Therefore, to assess whether this step affects the contact map of Spatial-ATAC-Hi-C, we compared the Spatial-ATAC-Hi-C data with a Dip-C mouse brain dataset⁵⁸, representing a comparison between in situ tagmentation and tagmentation performed on extracted genomic DNA following cell lysis. These two datasets also showed high consistency in Hi-C contact maps, PC1 and insulation scores (Extended Data Fig. 1f–i).

Furthermore, in the Spatial-ATAC-Hi-C protocol, as Hi-C is performed directly on the tissue sections, we cannot sonicate proximity-ligated DNAs on the slide to introduce spatial barcodes. Therefore, we use Tn5 transposomes to fragment the proximity-ligated DNAs and then incorporate spatial barcodes during library preparation. Consequently, the chromatin accessibility information is captured as an inherent feature of the Hi-C library construction. Therefore, we evaluated the ATAC-seq data from the Spatial-ATAC-Hi-C data by comparing it with bulk ATAC-seq data. From the genome browser views, we observed that the profile of pseudobulk ATAC-seq data was highly consistent with the bulk ATAC-seq data, as well as with ATAC-seq datasets from the ENCODE Consortium⁵⁹ (Fig. 1d and Extended Data Fig. 2a). Genome-wide, the two datasets demonstrated high correlation (Spearman correlation (SCC) = 0.895) (Extended Data Fig. 2b). In addition, both datasets showed strong enrichment at the union peak regions (Extended Data Fig. 2c). Taken together, we conclude that in general, the patterns of open chromatin accessibility are largely consistent between Spatial-ATAC-Hi-C and bulk ATAC-seq.

Extended Data Fig. 2 — a. Comparison of aggregated chromatin accessibility profiles of Spatial-ATAC-Hi-C in mouse brain R6 sample with bulk ATAC-seq from adjacent mouse brain R6 tissue and the ATAC-seq profiles in the forebrain, midbrain and hindbrain of E16.5 mouse embryos from ENCODE consortium⁵⁹ at genomic region chr7: 45.11 – 45.21 Mb. b. Density plot of log1p(Spatial-ATAC-Hi-C signal) in mouse brain R6 sample versus log1p(bulk ATAC-seq signal) from adjacent mouse brain R6 tissue. Each dot represents the ATAC-seq signal from a 10 kb genomic bin. c. The signal of bulk ATAC-seq from adjacent mouse brain R6 tissue and Spatial-ATAC-Hi-C in mouse brain R6 sample at their union ATAC-seq peak regions.

Source data

Finally, we evaluated whether the Hi-C data from Spatial-ATAC-Hi-C is biased toward the open chromatin regions, as we used the Tn5 transposomes for fragmentation. To perform this benchmark, we first converted the Hi-C read pairs into one-dimensional tracks for both the Spatial-ATAC-Hi-C and in situ Hi-C datasets. These one-dimensional tracks were then compared to the bulk ATAC-seq data performed in adjacent tissue section. We observed that the Spatial-ATAC-Hi-C reads showed a stronger enrichment at open chromatin regions comparing to the in situ Hi-C datasets (Extended Data Fig. 3a,b). We further quantified the proportion of reads overlapping open chromatin regions and found that 4.5% (8,086,063 of 181,231,943) of the Spatial-ATAC-Hi-C reads and 2.6% (7,359,754 of 287,122,769) of the in situ Hi-C reads had at least one read end overlapping these regions (Extended Data Fig. 3c). Next, we compared the chromatin loops between these two datasets and found that 6,492 (62%) loops were shared (Extended Data Fig. 3d,e). Overall, 52% of Spatial-ATAC-Hi-C-specific loops had at least one anchor overlapping ATAC-seq peaks, in contrast to 44% of in situ Hi-C-specific loops (Extended Data Fig. 3f). This analysis indicates that, at chromatin loop level, Spatial-ATAC-Hi-C is largely consistent with in situ Hi-C, with slightly more enrichment at chromatin accessible regions.

Extended Data Fig. 3 — a. Comparison of Spatial-ATAC-Hi-C 1 d track in mouse brain R6 sample, in situ Hi-C 1 d track and bulk ATAC-seq track from adjacent mouse brain R6 tissues at genomic region chr2: 25,004,450 – 25,802,210. b. The coverage of Spatial-ATAC-Hi-C in mouse brain R6 sample (red) and in situ Hi-C from adjacent mouse brain R6 tissue (gray) at bulk ATAC-seq peak regions from adjacent mouse brain R6 tissue. c. Barplot showing the fraction of reads overlapping with bulk ATAC-seq peaks from adjacent mouse brain R6 tissue in Spatial-ATAC-Hi-C in mouse brain R6 sample (red) and in situ Hi-C from adjacent mouse brain R6 tissue (gray). Reads are stratified based on whether both ends or either end overlap ATAC-seq peaks. d. Venn diagram showing the overlap between chromatin loops identified in Spatial-ATAC-Hi-C in mouse brain R6 sample and in situ Hi-C from adjacent mouse brain R6 tissue. Numbers indicate the number of chromatin loops unique to each sample or shared between both samples. e. Aggregate Peak Analysis (APA) plots of Spatial-ATAC-Hi-C specific chromatin loops, in situ Hi-C specific chromatin loops and chromatin loops shared by Spatial-ATAC-Hi-C and in situ Hi-C. f. Barplot showing the fraction of chromatin loops overlapping with bulk ATAC-seq peaks from adjacent mouse brain R6 tissue for the three loop sets shown in (e). Chromatin loops are stratified based on whether both ends or either end overlap ATAC-seq peaks.

Source data

In this study, we performed Spatial-ATAC-Hi-C in a variety of tissue samples, including mouse adult brains, human adult cerebellum, as well as GBM patient samples. Across all sample types, we achieved medians of 25,343–58,403 total contacts per pixel (Fig. 1e, Extended Data Fig. 4a and Supplementary Table 4). Each pixel contained on average ~3–21 cells, with only modest variation across tissues and specimens (Extended Data Fig. 4b). Among the total contacts, 88.1–90.3% were intra-chromosomal (Extended Data Fig. 4c), and 24–33.3% were long-range interactions (>10 kb; Fig. 1f). Compared to recently published single-cell Hi-C methods^40,60,61, Spatial-ATAC-Hi-C yielded comparable total and long-range (>10 kb) contact reads relative to Droplet Hi-C⁴⁰, single-cell Hi-C⁶⁰ and sciHi-C⁶¹ (Extended Data Fig. 4d,e). Notably, Spatial-ATAC-Hi-C yielded a higher long-range ratio than both Dip-C⁵⁸ and Droplet Hi-C (Extended Data Fig. 4f). For the ATAC-seq data modality from Spatial-ATAC-Hi-C, the median unique fragments per pixel ranged from 34,596 to 90,465 (Extended Data Fig. 4g and Supplementary Table 5). The transcription start site (TSS) enrichment of ATAC-seq signal profiles demonstrated a clear peak in the center and displayed nucleosome periodicity patterns (Fig. 1g), confirming the quality of ATAC-seq data. Both the total fragments and TSS enrichment were comparable to previously published spatial-ATAC-seq data (Extended Data Fig. 4h,i).

Extended Data Fig. 4 — a. Histogram showing the Spatial-ATAC-Hi-C reads distribution in each sample. Median value of unique read pairs in each sample is labeled in the histogram. b. Boxplot showing cell counts across selected pixels in each sample. c. Violin plot showing the fraction of cis contacts over the unique total contacts in each sample. **d-f**. Violin plot showing the comparison of Spatial-ATAC-Hi-C unique total contacts at log10 scale (d), long-range contacts ( > 10 kb) at log10 scale (e) and long-range ratio ( > 10 kb/total cis) (f) with published single cell Hi-C methods^40,58,60,61. g. Density plot showing the TSS enrichment score versus unique fragments per cell in each sample. **h-i**. Violin plot showing the comparison of Spatial-ATAC-Hi-C unique fragments at log10 scale (h) and TSS enrichment scores (i) with published spatial-ATAC-seq methods^10,101. Box plots overlaid on violin plots in **b-f**, h and i show the median (center line), the 25th and 75th percentiles (box limits), and whiskers extending to the most extreme data points within 1.5 × IQR of the lower and upper quartiles. Outliers beyond the whiskers are not shown in **c-f**, h and i. The number of pixels in each sample is shown in c, h, i. The number of pixels/cells in each method is shown in **d-f**.

Source data

Spatial-ATAC-Hi-C captures complex cell identities in adult mouse brain

We first performed Spatial-ATAC-Hi-C using a coronal adult mouse brain section that encompassed the anterior hypothalamus (R6 region). We conducted clustering analysis using either chromatin interaction (Fig. 2a) or chromatin accessibility data (Fig. 2b). Specifically, we calculated the gene-associating domain (GAD) score⁶² with the Hi-C data, which measures the strength of chromatin contacts across each gene body. We then computed the gene activity score (GAS)⁶³ with the ATAC-seq data, which quantifies DNA fragments within gene bodies and promoter regions for each gene. We identified eight clusters with the GAD scores and seven clusters with the GAS, respectively (Fig. 2a,b). The two clustering results showed strong concordance and their spatial distributions were also highly consistent with tissue anatomical and class annotations⁶⁴ (Fig. 2c,d). For example, both the ATAC-seq cluster A2 and Hi-C cluster H2 were mapped to the striatum region, while Hi-C cluster H0 and ATAC-seq cluster A1 were mainly mapped to the isocortex region, indicating that Spatial-ATAC-Hi-C data faithfully recapitulated the spatial divisions in mouse brain.

Fig. 2 — a, An unsupervised clustering analysis was performed based on the gene-associated domain score derived from spatial Hi-C data. An overlay of clusters with tissue image reveals the spatial distribution of the clusters (left). Uniform manifold approximation and projection (UMAP) embedding of the clustering analysis from spatial Hi-C data (right). b, An unsupervised clustering analysis was performed based on the GAS derived from spatial-ATAC-seq data. An overlay of clusters with tissue image reveals the spatial distribution of the clusters (left). UMAP embedding of the clustering analysis from spatial-ATAC-seq data (right). c, Mouse brain region anatomy annotation from Allen Brain Institute (https://mouse.brain-map.org/static/atlas). d, Mouse brain class annotation from Allen Brain Institute (https://mouse.brain-map.org/static/atlas). e, Projection of mouse brain R6 spatial-ATAC-seq data (right) to single-cell ATAC-seq data from mouse brain (left). f, Spatial distribution of each cell type from e in mouse brain R6 sample. g, Spatial distribution of GAD scores (top), GAS (middle) and MERFISH (bottom) for selected marker genes. h, Heatmap of genes with differential activity scores in excitatory neuron, inhibitory neuron and non-neuron cell types. i, GO term enrichment of excitatory neuron marker genes. The statistical significance of GO term enrichment was assessed using Fisher’s exact test.

Source data

To annotate cell identities, we projected the ATAC-seq data from each pixel to the published scATAC-seq data in the mouse brain atlas⁶⁵ from the similar coronal sections (Fig. 2e). After the projection, we acquired the predicted cell-type annotation for each pixel (Fig. 2f). The spatial projection of these cell-type annotations was highly consistent with regions and cell-type classes previously established using MERFISH⁶⁴ (Fig. 2d). Next, we examined the spatial patterns of the cell-type-specific marker genes. As shown in Fig. 2g, both Adora2a, a marker gene of cerebral nuclei (CNU), lateral ganglionic eminence (LGE) GABA class⁶⁶, and Penk (Proenkephalin), a known medium spiny neuron marker⁶⁷, showed high GAD and GAS scores in the striatum (STR) (anatomy annotation) and CNU-LGE GABAergic neuronal layers (cell-type annotation). This is highly consistent with the spatial patterns from the MERFISH data (Fig. 2g). Similarly, the activity patterns of the oligodendrocyte (Oligo) marker gene Cldn11 and the intratelencephalic-extratelencephalic glutamatergic (IT-ET Glut) marker gene Satb2 (ref. ⁶⁶) also demonstrated consistent results between Spatial-ATAC-Hi-C and MERFISH (Fig. 2g). Taken together, these results demonstrated the reliability of our Spatial-ATAC-Hi-C data in recapitulating the cellular identity.

Next, we grouped these clusters into three major cell types: excitatory neurons, inhibitory neurons and non-neuronal cells. We identified 915 major cell-type-specific genes, including Satb2 and Neurod6 for excitatory neurons, Adora2a and Gad2 for inhibitory neurons, and Mpo and Nfia for non-neuronal cells (Fig. 2h). Exemplary pseudobulk tracks were demonstrated for the Gad2 and Neurod6 gene regions (Supplementary Fig. 1a,b). We identified 1,006 major cell-type-specific open chromatin regions (Supplementary Fig. 1c and Supplementary Table 6). Gene Ontology (GO) enrichment analysis of marker genes revealed distinct functional signatures across major cell types. Excitatory neuron markers were enriched for pathways related to nervous system development and glutamatergic synaptic signaling, including synapse assembly and ionotropic glutamate receptor signaling (Fig. 2i). Inhibitory neuron markers were enriched for processes associated with chemical synaptic transmission, G protein-coupled receptor signaling, and ion transport related to neuronal activity (Supplementary Fig. 1d). Meanwhile, non-neuronal markers were enriched for pathways involved in cell differentiation, epithelial cell adhesion and calcium ion regulation (Supplementary Fig. 1e). Together, these results support the functional specificity of the identified marker genes across excitatory neurons, inhibitory neurons and non-neuronal cell populations.

We further applied Spatial-ATAC-Hi-C to a different coronal section containing the anterior hippocampus and thalamus (R8 region). Again, we observed high concordance between clustering results from the Hi-C and ATAC-seq data modalities (Extended Data Fig. 5a,b), both of which aligned with cell supertype annotations from MERFISH (Extended Data Fig. 5c). Integration of our data with the scATAC-seq data from the Mouse Brain Atlas⁶⁵ enabled the annotation of 16 major cell-type identities (Extended Data Fig. 5d). These annotations were in general spatially consistent with the cell-type annotations from the MERFISH data (Extended Data Fig. 5c,e).

Next, we performed further projection with subtype annotations for pixels in the IT-ET glutamatergic neuron cell type (Extended Data Fig. 5f). This allowed us to identify distinct neuronal subtypes, including: L2/3 IT CTX Glut (layer 2/3 intratelencephalic cortical glutamatergic) and L4/5 IT CTX Glut (layer 4/5 intratelencephalic cortical glutamatergic). These subtypes exhibited a characteristic laminar distribution pattern, extending from the outer to inner layer of the isocortex and were consistent with subclass annotations from the MERFISH results (Extended Data Fig. 5g).

The spatial distribution patterns of GAD and GAS for known marker genes (Extended Data Fig. 5h) recapitulated the cell-type specificity of these genes. For example, IT-ET Glut marker gene Satb2 (ref. ⁶⁶) showed high accessibility and chromatin interactions selectively in cortical and hippocampal neurons. The post-mitotic neuron marker Rbfox3 showed high GAS and GAD scores in the isocortex and thalamus regions. Olig2, a marker gene for oligodendrocytes and their precursor cells (OPCs)⁶⁸, mainly exhibited strong gene activity at the ventral posteromedial (VPM) thalamus. Gbx2, as an essential gene to maintain the thalamic neuron identity⁶⁹, showed high gene activity in the thalamus glutamatergic neurons. Genome-wide, we identified 431 major cell-type-specific genes (Extended Data Fig. 5i) and 1,731 major cell-type-specific open chromatin regions (Extended Data Fig. 5j).

Taken together, our results demonstrated that the Spatial-ATAC-Hi-C data can spatially capture complex cell identities in adult mouse brain tissue sections.

Spatial-ATAC-Hi-C reveals spatially resolved, cell-type-specific 3D chromatin architecture

First, we investigated the spatial variability of A/B compartments across adult mouse brain tissue sections. We computed the A/B compartment scores (PC1 value) for each cell type (Supplementary Tables 7 and 8) and each pixel at 100-kb resolution. The A/B compartment tracks were largely conserved across different cell types (Extended Data Fig. 6a). Of the 1,683 identified cell-type-specific A/B compartment regions, 765 exhibited high correlations (Pearson correlation coefficient (PCC) > 0.3) between PC1 values and the GAS scores of genes overlapping these regions (Fig. 3a).

Extended Data Fig. 6 — a. Pseudobulk (upper panel) and single-pixel (lower panel) A/B compartment tracks on chr8 for each cell type of mouse brain R6 sample. b. Heatmap showing cell-type-specific chromatin loops in mouse brain R6 sample. c. Excitatory neuron-specific chromatin loop involving *Neurod6* gene in mouse R8 sample. The loop anchors were highlighted in blue and orange shades. The upper panel shows the pseudobulk Hi-C map of inhibitory neuron (upper triangle) and excitatory neuron (lower triangle) at genomic region chr6: 55.2 – 56.2 Mb. The lower panel shows the virtual 4 C interaction track anchored at *Neurod6* region. d. Spatial loop signal at single-pixel resolution for the differential loop involving *Neurod6* showed in (c). e. Heatmap showing cell-type-specific chromatin loops in mouse brain R8 sample. f. APA plots of cell-type-specific loops in mouse R8 sample. g. GO term enrichment of genes located in the excitatory neuron-specific loop anchors shown in (f). h. GO term enrichment of genes located in the inhibitory neuron-specific loop anchors shown in (f). The statistical significance of GO term enrichment in g and h was assessed using Fisher’s exact test.

Source data

Fig. 3 — a, Heatmap of differential compartment regions (left) and GAS (right) of compartment region correlated genes (overlapped with the compartment regions and PCC of compartment score and GAS ≥ 0.3). b, GO term enrichment of compartment region correlated genes shown in a. IL, interleukin. c, Excitatory neuron-specific chromatin loop involving *Satb2* gene. The loop anchors are highlighted in blue and orange shades. Pseudobulk Hi-C map of inhibitory neuron (top triangle) and excitatory neuron (bottom triangle) at genomic region chr1: 55.5–58.5 Mb (top). Virtual 4C interaction track anchored at *Satb2* region (bottom). d, Excitatory neuron-specific chromatin loop involving *Satb2* gene in snm3C dataset. The loop anchors are highlighted in blue and orange shades. Pseudobulk Hi-C map of CNU-LGE GABA (top triangle) and IT-ET Glut (bottom triangle) at genomic region chr1: 55.5–58.5 Mb (top). Virtual 4C interaction track anchored at *Satb2* region (bottom). e, Spatial loop signal at single-pixel resolution for the differential loop involving *Satb2* showed in c. f, Inhibitory neuron-specific chromatin loop involving *Gpr88* gene. The loop anchors are highlighted in blue and orange shades. Pseudobulk Hi-C map of inhibitory neuron (top triangle) and excitatory neuron (bottom triangle) at genomic region chr3: 115.3–117.3 Mb (top). Virtual 4C interaction track anchored at *Gpr88* region (bottom). g, Inhibitory neuron-specific chromatin loop involving the *Gpr88* gene in the snm3C dataset. The loop anchors are highlighted in blue and orange shades. Pseudobulk Hi-C map of CNU-LGE GABA (top triangle) and IT-ET Glut (bottom triangle) at genomic region chr3: 115.3–117.3 Mb (top). Virtual 4C interaction track anchored at *Gpr88* region (bottom). h, Spatial loop signal at single-pixel resolution for the differential loop involving *Gpr88* showed in f. i, Aggregate peak analysis (APA) plots of cell-type-specific loops in mouse R6 sample (left) and GO term enrichment of genes located in the corresponding cell-type-specific loop anchors (right). j, Bubble plot of transcription factor motif enrichment at chromatin accessible regions located in cell-type-specific loops in i. Bubble size and color represent the significance of enrichment (−log₁₀ P value). Significance of motif enrichment was assessed using hypergeometric test. The statistical significances of GO term enrichment in b and i were assessed using Fisher’s exact test.

Source data

Ontology enrichment analysis of the genes in cell-type-specific A/B compartments revealed biologically coherent functional signatures for each cell type (Fig. 3b). Excitatory neuron-specific compartments were enriched for terms related to synaptic function, including neuron cell–cell adhesion, neurotransmitter transport and positive regulation of excitatory postsynaptic potential. Inhibitory neuron-specific compartments were enriched for telencephalon cell migration, GPCR signaling and glutamate receptor signaling. Non-neuronal-specific compartments were dominated by immune-related terms, including defense response to Gram-positive bacteria, regulation of interleukin-1 production, and macrophage chemotaxis and migration, consistent with the microglial component of non-neuronal brain cells. Together, these results suggest that integration of chromatin compartmentalization and accessibility reveals the distinct functional identities of major brain cell classes.

To investigate the 3D genome at a higher resolution, we performed chromatin loop analysis. We observed an excitatory neuron-specific loop which links the Satb2 gene promoter region to a distal chromatin accessible region located ~1 Mb away (Fig. 3c). The chromatin interaction signals from excitatory neurons were significantly higher than those of inhibitory neurons (one-way analysis of variance (ANOVA), P= 6.87 × 10⁻⁶), which was quantitatively demonstrated using virtual 4C tracks. To further validate this observation, we leveraged snm3C data, which provides single-nuclei 3D genome information across all cell types from adult mouse brain⁷⁰. Pseudobulk Hi-C maps and virtual 4C tracks derived from the snm3C data in the representative excitatory (IT-ET Glut) and inhibitory (CNU-LGE GABA) neurons were consistent with the Spatial-ATAC-Hi-C (Fig. 3d). At the spatial single-pixel level, we observed that the signals of this loop were specifically high in cells from the isocortex region (Fig. 3e). Similarly, we showed inhibitory neuron-specific loops for the Gpr88 gene (Fig. 3f), which were further supported by snm3C data (Fig. 3g) and showed specific high loop signals in cells from the striatum region (Fig. 3h). These results demonstrated that Spatial-ATAC-Hi-C enables the identification of spatially resolved cell-type-specific chromatin loops.

Overall, we identified 268 excitatory neuron-, 556 inhibitory- and 41 non-neuronal-specific loops in the mouse brain R6 sample (Fig. 3i, Extended Data Fig. 6b and Supplementary Table 9). GO analysis of the genes in these cell-type-specific loop anchors revealed pathway enrichments consistent with established cellular functions (Fig. 3i). Next, we performed motif enrichment analysis for the open chromatin regions located in these cell-type-specific loop anchors (Fig. 3j). In excitatory neurons, enrichment of Jun-AP1 and Fosl2 suggests involvement of activity-dependent transcriptional programs, while Atf3 enrichment is consistent with regulatory pathways linked to axon regeneration. In inhibitory neurons, Meis1 enrichment aligns with its known role in the differentiation of striatal medium spiny neurons.

We performed a similar analysis for the Spatial-ATAC-Hi-C data in the mouse brain R8 sample. We demonstrated an excitatory neuron-specific loop associating with the Neurod6 gene (Extended Data Fig. 6c), which shows specific high loop signals in cells from the isocortex and hippocampal regions (Extended Data Fig. 6d). Overall, we identified 1,688 excitatory neuronal-specific, 243 inhibitory neuronal-specific and 17 non-neuronal-specific loops (Extended Data Fig. 6e,f and Supplementary Table 9). Again, excitatory neuron-specific loop associated genes are enriched for pathways related to synaptic signaling and neuronal projection, consistent with the functional roles of excitatory neurons (Extended Data Fig. 6g). Inhibitory neuron-specific loop associated genes were enriched for pathways associated with neuronal projection development, including axonogenesis and axon guidance, consistent with the extensive axonal arborization and circuit wiring characteristic of inhibitory interneurons (Extended Data Fig. 6h).

Spatial co-profiling of chromatin structure and accessibility in human brain

To investigate the spatial 3D genome organization in human brains, we performed Spatial-ATAC-Hi-C in adult human cerebellum sample. In the cerebellum, hematoxylin and eosin (H&E) staining of adjacent tissue sections revealed distinct anatomical patterns, with darker regions corresponding to the granule cell layer and lighter regions to the molecular layer (Supplementary Fig. 2a). The clustering results of both ATAC and Hi-C data modalities from the Spatial-ATAC-Hi-C data demonstrated good agreement with these patterns (Supplementary Fig. 2b). The spatial distributions of GAS and GAD scores of the cerebellar granule neuron (CGN) marker genes were also consistent with the clustering results. As representative examples, we highlighted the following genes in Supplementary Fig. 2cd: ETV1, a member of the ETS family of transcription factors that exhibits restricted expression exclusively in CGNs and has been functionally associated with CGN maturation⁷¹; NEUROD1, a basic helix-loop-helix (bHLH) transcription factor required for proper differentiation of cerebellar granule neurons precursors⁷²; and NFIA, a critical transcriptional regulator that specifically controls temporal gene expression required for dendritogenesis and synaptogenesis during the developmental maturation of CGNs⁷¹.

To examine the chromatin accessible regions in this sample, we compared the ATAC-seq signals from the Spatial-ATACHi-C data with the published scATAC-seq data from the granular cells in the cerebellum (CBGRC)⁷³. As shown in Supplementary Fig. 2e, the two datasets were highly similar near the ETV1 and NEUROD1 promoter regions, at both pseudobulk level and single-pixel level. Examining the ATAC-seq peaks from CBGRC, we also observed strong consistency between the ATAC-seq data from Spatial-ATAC-Hi-C and the ATAC-seq from CBGRC (Supplementary Fig. 2f).

Next, we investigated the chromatin interactions surrounding these marker genes. We observed that for both ETV1 and NFIA, cluster H0 demonstrated higher chromatin interactions between the gene promoter and a distal chromatin accessible region than cluster H1 (Supplementary Fig. 2g,h). In addition, we observed obviously higher enrichment of single-pixel ATAC-seq signals in H0 than in H1.

Collectively, these results highlight the ability of Spatial-ATAC-Hi-C in spatially profiling 3D genome architectures and chromatin accessibilities and revealing cell-type-specific regulatory events in human brain.

Spatial-ATAC-Hi-C reveals spatial distribution of CNVs

Recent work, including ours, such as HiCBreakFinder⁷⁴, HiNT⁷⁵ and EagleC⁷⁶, has shown that Hi-C is a powerful tool for profiling CNVs and SVs genome-wide. It has been used to identify SVs in different types of cancers, such as leukemia²², GBM^77,78, liposarcoma⁷⁹, ependymoma⁸⁰ and prostate cancer⁸¹. To demonstrate that Spatial-ATAC-Hi-C can measure the spatial distribution of CNVs, we designed the following experiment: we mounted two tumor samples, one from a patient with GBM (NU04741) and the other one from a patient with astrocytoma (NU04707), onto the same sample slide and performed a Spatial-ATAC-Hi-C experiment. We also performed whole-genome sequencing (WGS) for each tumor using adjacent tissue sections for validation (Fig. 4a,b).

Fig. 4 — a, Schematic workflow of the experimental design. b, Spatial location of the two tumor samples NU04707 (in blue) and NU04741 (in orange), which were stitched together onto one slide. c, Copy number profiles for NU04707 and NU04741 from Spatial-ATAC-Hi-C data and WGS data (top) and zoomed-in profiles on chr9 and chr12 (bottom). The copy number profiles for Spatial-ATAC-Hi-C data were generated based on the pseudobulk of the pixels from NU04707 and NU04741 samples on each side of the stitched slide. The copy number profiles for WGS experiments were generated from the adjacent tissue section of these two patients separately. d, UMAP embedding of the clustering analysis based on single-pixel copy number profiles at 5-Mb resolution. e, Spatial distribution of the clusters in d. f, Heatmap showing the whole-genome copy number in each pixel. Panel a created in BioRender; Wang, P. https://biorender.com/e8huh87 (2026).

First, we examined the CNVs identified from Spatial-ATAC-Hi-C. We generated pseudobulk Hi-C datasets for NU04707 (astrocytoma) and NU04741 (GBM) from the Spatial-ATAC-Hi-C data and performed CNV calling at 50-kb resolution using NeoLoopFinder⁸². Overall, the CNV profiles inferred from Spatial-ATAC-Hi-C were highly concordant with those derived from WGS data (Fig. 4c). We demonstrated two specific examples: (1) copy number losses on chr9 containing the CDKN2A and CDKN2B genes in the astrocytoma sample; (2) copy number gains on chr12 containing the CDK4 gene in the GBM sample (Fig. 4c). Both examples showed high consistencies between the Spatial-ATAC-Hi-C and WGS datasets.

Next, we validated the SVs identified from Spatial-ATAC-Hi-C. From the pseudobulk Spatial-ATAC-Hi-C contact maps, we observed several SVs (Extended Data Fig. 7a). Specifically, in NU04707, we demonstrated a translocation event between chr2 and chr5 identified from the Spatial-ATAC-Hi-C contact maps (Extended Data Fig. 7b). This SV event, which was further validated by the WGS data, leads to a potential ADAM23::CTNND2 fusion event (Extended Data Fig. 7c). In NU04741, we demonstrated a translocation event between chr7 and chrX, leading to SDK1::PAK3 fusion (Extended Data Fig. 7d), which was also further supported by the WGS data (Extended Data Fig. 7e). Using these pseudobulk Hi-C datasets, we systematically identified 73 and 12 SV events in NU04707 and NU04741, respectively using EagleC⁷⁶ (Extended Data Fig. 7f). For WGS-based SV detection, we used both Delly⁸³ and Lumpy⁸⁴. SVs identified from the Hi-C data were considered supported by WGS only if they were supported by both Delly and Lumpy. A total of 53 (72.6%) SVs in NU04707 and 7 (58%) SVs in NU04741 can be further supported by the WGS SV results (Extended Data Fig. 7f). We and others have previously demonstrated that Hi-C can serve as a complementary method for large-scale SV detection^74–76. These results demonstrate that Spatial-ATAC-Hi-C can faithfully capture SV profiles at pseudobulk level.

Extended Data Fig. 7 — a. Pseudobulk Hi-C maps for NU04707 (upper triangle) and NU04741 (lower triangle) at whole-genome level. Examples of structural variations (SVs) were highlighted in dashed rectangles. b. Chromatin interaction between chr2 and chr5 regions in pseudobulk Hi-C maps for NU04707. SV breakpoint was highlighted in black circle. Fusion genes were labeled near the SV breakpoint. c. WGS data provide supporting evidence for SV identified by Hi-C data in (b). Each line in the WGS panel represents a read pair. WGS reads that support the breakpoint site are marked as purple (forward strand) and red (reverse strand). Overlapping portions of read pairs are in green. d. Chromatin interaction between chr7 and chrX regions in pseudobulk Hi-C maps for NU04741. SV breakpoint was highlighted in black circle. Fusion genes were labeled near the SV breakpoint. e. WGS data provide supporting evidence for SV identified from Hi-C data in (d). f. Barplot showing the number of SVs identified from pseudobulk Spatial-ATAC-Hi-C data from NU04707 and NU04741. Number of SVs supported by WGS data were labeled in purple.

Source data

Finally, we investigated whether Spatial-ATAC-Hi-C could capture the spatial heterogeneity in CNV profiles. We inferred the CNV profiles at 5-Mb resolution for each single-pixel and performed clustering analysis based on the CNV matrix (Fig. 4d,e). In total, there were five clusters: C0 mostly from the astrocytoma patient NU04707 and C1, C2 and C3 mostly from the GBM patient NU04741 (Fig. 4d,e). We plotted the CNV profiles for each pixel in a heatmap (Fig. 4f), and we observed copy number differences among different clusters, such as chr12p and chr11q. These results demonstrate that Spatial-ATAC-Hi-C can capture the spatial heterogeneity in CNV profiles.

Spatial 3D genome structure and chromatin accessibility co-profiling of glioblastoma

Finally, to demonstrate the clinical potential of Spatial-ATAC-Hi-C, we performed Spatial-ATAC-Hi-C on freshly-resected glioblastoma samples from human patients, which are known to carry complex genomic landscape and extensive CNVs and SVs⁷⁷.

We first performed Spatial-ATAC-Hi-C in GBM sample NU03747 (Fig. 5a) and generated a median of 45,287 Hi-C reads and 90,466 ATAC-seq reads per pixel. In total, we generated ~141 million unique Hi-C read pairs and ~264 million ATAC-seq reads. We profiled genome-wide CNVs using the pseudobulk Hi-C data at 100-kb resolution, and then computed CNVs for each pixel at 5-Mb resolution (Fig. 5b) using NeoLoopFinder⁸². We observed known genomic alterations of GBM, including chr7 gain and chr10 loss, as well as two prominent focal amplifications on chr7 and chr12 (marked by red circles in Fig. 5b). Both the chromosomal level gain/loss and local amplifications were observed in the bulk data and in the single-pixel data.

Previous studies have shown that CNVs represent a major source of tumor heterogeneity^77,78,85. Based on the CNV profiles, we identified three clusters (Supplementary Fig. 3a), and among them, C0 and C1 showed clear spatial separation (Fig. 5c). To explore the features that contribute to the heterogeneity, we first focused on the focal amplification regions as they were shown to be potentially major drivers in GBM⁷⁷. We observed that there were two regions located in chr7 and chr12 on the Hi-C map that showed interaction with the rest of the whole genome and formed a stripe-like pattern (Fig. 5d). Such pattern indicated high copy numbers, and is consistent with what we observed from the CNV profiles presented in Fig. 5b. When zooming in on the Hi-C map, we observed that the focal amplifications contained known oncogenes such as EGFR, MDM2 and CDK4 (Fig. 5e,f). There are potentially ~80–100 copies at these amplification regions. The C0 cluster showed lower copies at EGFR and MDM2 regions, whereas it showed higher copies at the CDK4 region (Fig. 5e,f). This difference can also be consistently observed at a single-pixel level. For example, at the EGFR amplification region, pixels with higher copies were predominantly enriched in C1, whereas those with relatively lower copy numbers were more common in other clusters (Fig. 5g). The spatial distribution of the copy numbers at these gene regions further supported the differences between the two clusters (Fig. 5h). These results indicated that EGFR, CDK4 and MDM2 amplifications could be a strong feature and marker for tumor heterogeneity. Furthermore, the stripe patterns across all chromosomes on the Hi-C map and the high copy numbers at these regions suggest that they might be located on extrachromosomal (ec) DNAs⁸⁶, which has also been shown to contribute to GBM heterogeneity⁸⁷. Using EagleC software, we identified many SVs in this region and the rest of the genome (Supplementary Table 10). These observations are noteworthy and warrant in-depth studies in GBM to further elucidate the underlying mechanisms.

We further explored the clustering based on epigenetic features. We identified two clusters based on Hi-C and ATAC-seq data modalities, both of which exhibited spatial distribution patterns consistent with results based on copy numbers (Supplementary Fig. 3b,c). To avoid the influence from focal amplifications, we have excluded genes that are in the focal amplification regions. We have identified 1,841 differential open chromatin regions across different clusters (Supplementary Fig. 3d). These results demonstrated that the Spatial-ATAC-Hi-C enabled the spatial characterization of genetically distinct tumor clones.

To investigate distinct 3D genome features across different clusters, we identified differential chromatin interactions. For example, we observed higher chromatin interactions in cluster C0 than in C1 for RNFT2, consistent with its higher GAS in the cluster C0 (Fig. 5i). We identified 440 C0-specific chromatin loops and 516 C1-specific chromatin loops genome-wide (Fig. 5j). Functional enrichment analysis revealed that C0 cluster-specific chromatin loops were predominantly involved in natural-killer-cell-mediated immunity and leukocyte-mediated cytotoxicity pathways, whereas C1 cluster-specific chromatin loops were significantly enriched in the Wnt signaling pathway (Fig. 5k).

To further validate our spatial chromatin profiling approach for revealing GBM heterogeneity, we performed Spatial-ATAC-Hi-C on another sample from a patient with GBM (NU03766) (Supplementary Fig. 4a). Similar to the analysis we presented above for NU03747, we identified CNVs for the pseudobulk Hi-C data at 100 kb and for the single-pixel Hi-C data at 5-Mb resolution, revealing chromosome 6q loss and chr7 amplification (Supplementary Fig. 4b). Using genome-wide single-pixel CNV profiles, we identified three spatially segregated clusters (Supplementary Fig. 4c,d). Next, we showed the GAD and GAS scores for the tumor suppressor gene-CDKN2B, oncogene-EGFR and SLAMF1 (Supplementary Fig. 4e). We observed that CDKN2B, EGFR and SLAMF1 exhibited strong region-specific and cluster-specific activities (CDKN2B in C2, EGFR in C0, and SLAMF1 in C1).

To explore the functional implications of the identified tumor clusters, we performed differential gene activity analysis based on GAS scores across clusters (Supplementary Fig. 4f). GO analysis of differentially active gene sets revealed that genes in cluster C1 were specifically enriched for immune response-related pathways, indicating potential immune cell infiltration and suggesting a distinct tumor microenvironment (Supplementary Fig. 4g). We also identified 1,781 cluster-specific open chromatin regions (Supplementary Fig. 4h)

In summary, we highlighted the clinical potential of Spatial-ATAC-Hi-C for delineating tumor heterogeneity and uncovering oncogenic features associated with CNVs and 3D genome architecture in GBM.

Discussion

In this study, we present Spatial-ATAC-Hi-C, a method for simultaneous spatial profiling of chromatin accessibility and 3D genome organization in tissue sections. As an emerging technology, Spatial-ATAC-Hi-C currently has several limitations. First, Spatial-ATAC-Hi-C operates at a spatial resolution of 50 × 50 μm per pixel, which does not achieve single-cell resolution, whereas platforms such as Xenium and Stereo-seq can reach subcellular resolution, and spatial-ATAC-seq methods can achieve resolutions as fine as ~10 × 10 μm. Second, Spatial-ATAC-Hi-C currently covers a relatively limited tissue area (~5 × 5 mm), compared to the larger capture areas provided by Visium (11 × 11 mm), Xenium (up to 10.45 × 22.45 mm) and Atera (up to 5 cm²). In addition, Spatial-ATAC-Hi-C has not yet been optimized for formalin-fixed paraffin-embedded (FFPE) samples, whereas many ST platforms, such as Visium HD, Atera, Stereo-seq (V2)⁸⁸ and Patho-DBiT⁸⁹, and spatial FFPE-ATAC-seq⁹⁰ have been adapted for FFPE-compatible workflows. Further optimizations are needed to overcome these limitations in future studies to facilitate the wide adoption of Spatial-ATAC-Hi-C.

The current spatial resolution of Spatial-ATAC-Hi-C is insufficient to resolve fine cellular heterogeneity. For example, the current protocol does not capture Purkinje cell layers in human cerebellum samples, which were detected by near single-cell resolution methods such as Slide-seq⁴ (~10 μm). Further optimization of Spatial-ATAC-Hi-C, for example by reducing pixel size to ~10 × 10 μm, could enable higher-resolution mapping of chromatin architecture and more precise annotation of distinct cell population.

Compared to single-cell-based 3D genome techniques, such as single-cell Hi-C or Dip-C, Spatial-ATAC-Hi-C enables spatially resolved mapping of region-specific chromatin loops and structural alterations across tissue compartments, although it identifies fewer cell clusters than these single-cell approaches. In contrast, single-cell Hi-C/Dip-C approaches offer finer resolution of chromatin interactions but lack spatial context. Together, these complementary approaches offer a more comprehensive understanding of how 3D genome organization contributes to cell identity and tissue architecture.

Compared to current imaging-based methods for 3D genome mapping, we believe that sequencing-based approaches are complementary and provide unique contributions. While imaging-based work provides high-resolution 3D genome architecture mapping in situ at the subcellular level, Spatial-ATAC-Hi-C offers (1) genome-wide coverage of chromatin organization in intact tissue; (2) in situ mapping of cell type-dependent spatial distributions with both chromatin accessibility and folding information; and (3) in situ detection of CNVs and SVs, revealing genomic heterogeneity and clonal evolution.

In sum, spatially resolved co-profiling of 3D chromatin architecture and chromatin accessibility represents a substantial advancement in spatial omics technique as well as in 3D genomics. This integrated approach provides a powerful tool to systematically explore the dynamic interplay between 3D chromatin structure, accessibility and spatiotemporal gene regulation within a complex biological context while preserving native spatial organization. This methodology provides a spatial approach for advancing research in cancer biology and developmental biology.

Methods

Preparation of tissue slides

Mouse C57 Anterior Hypothalamus Coronal Frozen Sections Reg. 6 (cat. no. MF-201-06-C57), Mouse C57 Hippocamp/thalamus/hypothal Coronal Frozen Section Reg. 8 (cat. no. MF-201-08-C57) and human brain cerebellum frozen sections (cat. no. HF-202) were purchased from Zyagen. Brains of C57BL/6 mice were snap-frozen in OCT without any fixative and sectioned in coronal orientation at a thickness of 7–10 μm. Tissue sections were collected on poly-L-lysine-coated glass slides (Electron Microscopy Sciences, cat. no. 63478-AS).

Glioma brain tumor specimens were collected fresh from the operating room at Northwestern Memorial Hospital using a protocol approved by the Institutional Review Board (STU00095863). All patients underwent informed consent for molecular analysis and research studies. After resection, specimens were flash-frozen in Optimal Cutting Temperature (Tissue-Tek) before storage in −80 °C until further analysis. An accompanying block from each tumor underwent H&E staining, and adequate tumor content was verified by a board-certified neuropathologist. In preparation for spatial profiling, poly-L-lysine- coated slides (Electron Microscopy Sciences; cat. no. 63478-AS) were chilled, and the samples sectioned into 10-μm thick slices using a cryostat, with temperature set at −20 °C (Leica). Excess OCT was manually trimmed before placement on each slide. Several adjacent slides were prepared for each sample to enable profiling with other modalities, including H&E stains on either side of the characterized slide.

Preparation of Tn5 transposome

Unloaded Tn5 transposase (cat. no. C01070010) were purchased from Diagenode, and the transposome was assembled following the manufacturer’s guidelines. The oligonucleotides (Supplementary Table 1) applied for transposome assembly were:

Tn5ME-A, 5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAG-3′

Tn5MErev, 5′-/5Phos/CTGTCTCTTATACACATCT-3′

Tn5ME-B, 5′-/5Phos/CATCGGCGTACGACTAGATGTGTATAAGAGA CAG-3′

DNA barcode sequencing, annealing and DNA oligonucleotides

The DNA oligonucleotides were synthesized by Integrated DNA Technologies with HPLC purification. Barcode A1–50 or B1–50 was first annealed with ligation linker 1 or 2, 10 µl of 100 µM ligation linker1 or 2, 10 µl of 100 µM each barcode A or B, and 20 µl of 2× annealing buffer (20 mM Tris, pH 7.5–8.0, 100 mM NaCl and 2 mM EDTA) and mixed well. DNA oligonucleotides used for PCR and library construction are shown in Supplementary Table 1. All DNA barcode sequences are provided in Supplementary Table 2.

Photomask fabrication

The design of the chrome photomask with a 50-μm channel width has been described previously⁹¹ and the actual photomasks were ordered from the company Front Range Photomasks.

Polydimethylsiloxane mold fabrication

The silicone molds were fabricated in the microfabrication and nanofabrication facility (NUFAB) core at Northwestern University, following clean room guidelines. In brief, a thin layer of SU-8 3025 (Kayaku Advanced Materials) was spin coated on a plasma-treated prime-grade silicon wafer at 2,500 rpm for 40 s. The spin-coated wafer was then heat-treated and exposed to a total of 400 mJ cm⁻² UV light under the photomask. The resulting wafer was developed and hard baked. The final thickness of the mold was ~35 μm.

PDMS microfluidic chip fabrication

Polydimethylsiloxane (PDMS) chips were fabricated as described previously⁹¹. In brief, Part A and Part B of the PDMS precursor (SYLGARD 184 Silicone Elastomer, Dow Corning) was mixed in a 10:1 ratio and poured onto the silicone mold mentioned above. The mixture was degassed completely, and heat cured at 65 °C for ~2.5 h or overnight. The hardened PDMS piece was extracted and trimmed to match the dimension of the glass slide. Inlet and outlet holes were punched at 2 mm in diameter, using Titanium-coated Follicle Unit Extraction punches (Robbins Instruments). Visual inspection was carried out to ensure the integrity of channels, inlet and outlet holes before usage. Each piece of the PDMS chip was single use only.

Experimental protocol for Spatial-ATAC-Hi-C

Frozen tissue slides were thawed for 10 min at room temperature. Tissue was fixed with formaldehyde (2%) for 20 min and quenched with 2.5 M glycine (final concentration: 0.2 M) for 5 min. After fixation, the tissue was washed twice with 1 ml of DPBS. The device was assembled by putting a PDMS block with a PDMS reservoir capable of holding ~200 μl liquid on top of the slide and then fixed with an acrylic clamp and four sets of screws and nuts. The tissue was permeabilized with 30 μl Lysis Buffer (1 μl of 1 M Tris, pH 8.0, 0.2 μl of 5 M NaCl, 2 μl 10% NP-40 and proteinase inhibitor) at 4 °C for 20 min. We added 36 μl of 10% SDS solution and incubated the tissue section at 62 °C for 10 min depending on tissue type. Next, 30 μl of 1% Triton X-100 was added to the PDMS block and the clamp was covered with parafilm with incubation at 37 °C for 15 min. A total of 20.5 μl of the enzyme mix (11 μl, 10× cut smart buffer, 1 μl 10% Triton X-100, 8 μl DpnII (NEB, R0543L) and 1.5 μl HinfI (NEB, R0155L)) was added with incubation at 37 °C for 1 h after gently tilting the slide. The tissue section was washed twice with 200 μl of deionized water at room temperature for 2–5 min. We then added 50 μl end repair mix (20 μl 0.4 mM biotin-14-dATP; 1 μl of 10 mM dCTP; 1 μl of 10 mM dGTP, 1 μl of 10 mM dTTP, 5 μl of 5 U μl⁻¹ DNA polymerase I, Large (Klenow) Fragment (NEB, M0210L); 5 μl 10× cut smart buffer and 17 μl H₂O) with incubation at room temperature for 1 h. Then the ligation mix (10 μl 10× T4 DNA ligation buffer; 1 μl 20 mg ml⁻¹ BSA; 1 μl 10% Triton X-100; 10 μl T4 DNA ligase (NEB, M0202L) and 78 μl H₂O) was added and incubated at 16 °C overnight. The tissue section was washed with 200 μl of Pre-tagmentation-wash buffer (1 M Tris pH 7.4; 5 M NaCl, 1 M MgCl₂, 10% Tween-20, 5% BSA, H₂O) at room temperature for 5 min after removing the ligation mix. Then, 100 µl of transposition mix (7.5 μl of home-made transposome, 50 µl 2× tagmentation buffer, 33 µl 1× DPBS, 1 µl 10% Tween-20, 3.5 µl nuclease-free H₂O) was added followed by incubation at 37 °C for 1 h. Next, 100 µl of 40 mM EDTA was added for incubation at room temperature for 5 min to stop transposition. The tissue section was washed twice with 500 µl 1× NEBuffer 3.1 for 5 min after removing all liquid from the reservoir.

For the first barcode (barcode A) in situ ligation, the first PDMS chip was covered in the tissue region of interest. For alignment purposes, a ×10 objective (Nikon ECLIPSE Ts2R) was used to take a brightfield image. The PDMS device and tissue slide were clamped tightly with a home-made acrylic clamp. For each channel, 5 µl of ligation master mixture was prepared with 2 µl of ligation mix (27 µl of T4 DNA ligase buffer, 72.4 µl of RNase-free water, 5.4 µl of 5% Triton X-100 and 11 µl of T4 DNA ligase), 2 µl of 1× NEBuffer 3.1 and 1 µl of each annealed DNA barcode A (A1−A50, 25 µM). A vacuum (Atlas) was used to load the ligation master mixture into 50 channels of the device, followed by incubation at 37 °C for 30 min in a wet box. The PDMS chip and clamp were removed after washing once with water and then air-dried. For the second barcode (barcode B) in situ ligation, the second PDMS chip covered the slide and a second brightfield image was taken with the ×10 objective. An acrylic clamp was applied to clamp the PDMS and tissue slide together. Annealing of barcodes B (B1−B50, 25 µM) and preparation of the ligation master mix were carried out as for barcode A. The device was incubated at 37 °C for 30 min in a wet box. The PDMS chip and clamp were removed, and the slide was washed with water and dried in air. A brightfield image was then taken for further alignment.

For tissue lysis, the region of interest was digested with 100 µl of the reverse crosslinking mixture (2 mg ml⁻¹ proteinase K, TE buffer, pH 8.0 and 1% SDS) at 65 °C for 1 h in a wet box. The lysate was collected in a 1.5-ml tube and incubated at 65 °C overnight. The DNA was purified with the Zymo ChIP-DNA purification kit and then eluted with 50 µl of EB buffer. To prepare for biotin pulldown, 50 μl of 10 mg ml⁻¹ Dynabeads MyOne Streptavidin T1 beads (Life Technologies, 65602) was washed twice with 200 μl of 1× Binding & Wash (B&W) buffer (5 mM Tris-HCl (pH 7.5), 0.5 mM EDTA and 1 M NaCl) on magnet rack. The beads were resuspended in 50 μl of 2× B&W buffer (10 mM Tris-HCl (pH 7.5), 1 mM EDTA and 2 M NaCl). Therefore, 50 µl purified DNA were added to 50 μl of the beads with buffer, mixed well, and then shaken on the rotator (ELMI RM-2L Intelli-mixer; with program UU, 40) for 1 h. The beads were then washed with 200 μl 1× TWB (5 mM Tris-HCl (pH 7.5), 0.5 mM EDTA, 1 M NaCl and 0.05% Tween-20) and the tubes were heated on a Thermomixer at 55 °C for 2 min with mixing. The supernatant was discarded using a magnet, and the wash step was repeated once. The beads were resuspended in 30 μl of EB buffer. The PCR solution (25 µl of 2 × Q5 High-Fidelity 2× Master Mix, 2.5 µl of 25 µM customized indexed i7 primer, 2.5 µl of 25 µM customized P5 PCR primer, 10 µl beads and 10 µl H₂O) was added and mixed well. PCR was performed using the program 72 °C for 5 min, 98 °C for 30 s and cycling at 98 °C for 10 s, 63 °C for 30 s and 72 °C for 1 min, 10–13 times depending on the sample. The final PCR product was purified by 1× Ampure XP beads and eluted in 50 µl of nuclease-free H₂O. Double-size selection step was performed with 0.55× and 0.15× Ampure XP beads. The final library was submitted for Illumina Xplus sequencing with 2 × 150-bp paired-end sequencing.

Histological image and analysis of cell number per pixel

All tissue samples were performed H&E staining following the manufacture’s procedure (Abcam, cat. no. ab245880). High-resolution histological images were obtained with Aperio GT-450 DX Scanner (Leica), and human patient samples were confirmed by a board-certified neuropathologist. These histological images were analyzed using QuPath open-source image-analysis software v.0.6.0. Cells were counted with the cell-detection feature using nuclear segmentation on matched histology images. Thresholds were set by comparison of segmentation of each nucleus at different parameters and then applied to all images of the experiment.

Data preprocessing

The raw fastq reads adaptors were first detected and trimmed with TrimGalore (v.0.6.10) (https://github.com/FelixKrueger/TrimGalore). The cleaned fastq files were processed using bbduk function from BBtools (BBMap v.39.01) (https://sourceforge.net/projects/bbmap/), retaining only reads that contained linker sequences (linker1 and linker2). The filtered reads were then converted to Cell Ranger ATAC format (10x Genomics). Specifically, the original paired-end fastq files included Read1 and Read2, with Read2 containing both the genomic DNA sequences and barcode sequences. The original Read2 were split into genomic DNA sequences as the new Read1 and barcode (barcode A and barcode B) sequences as new Read2, and the original Read1 sequences were retained as the new Read3. For spatial-ATAC-seq analysis, the resulting fastq files were aligned to the mouse (mm10) or human (GRCh38) reference genome using Cell Ranger ATAC v.2.0 (https://support.10×genomics.com/single-cell-atac/software/pipelines/latest/what-is-cell-ranger-atac) (10x Genomics). The resulting fragment.tsv file was used for downstream analysis. For spatial Hi-C analysis, the newly generated Read1 and Read3 were aligned to the corresponding reference genome using the runHiC⁹² pipeline (https://github.com/XiaoTaoWang/HiC_pipeline). The resulting pairs files were further split into 2,500 separate pairs files based on the barcode information in the new Read2. These pairs files were used for downstream analysis.

Quality control of Spatial-ATAC-Hi-C data

For Spatial-ATAC-Hi-C data, we used the pairtools stats (v.0.3.0) (https://github.com/open2c/pairtools) to compute summary statistics for all the pairs files in each sample, quantifying cis, trans and long-range (≥10 kb) contacts in single pixels. TSS enrichment of ATAC-seq signal was used to quantify the spatial-ATAC data quality. The ArchR⁷⁴ package was used to calculate the TSS enrichment scores. Specifically, TSS positions were obtained from the precompiled version of hg38 genome for human and mm10 for mouse in ArchR. To approximate a TSS enrichment score for each single pixel, we calculated the average accessibility within the 50-bp window centered on each TSS and normalized this by the average accessibility of the TSS flanking regions (±1,900–2,000 bp).

Image processing

Spatial images were manually processed using the AtlasXbrower (https://github.com/atlasxomics/AtlasXbrowser). The resulting spatial folder for the image was used for the spatial visualization in Seurat⁹³.

Imputation of Spatial-ATAC-Hi-C data

ScHiCluster⁹⁴ (v.1.3.5) (https://github.com/zhoujt1994/scHiCluster) was used to perform contact matrix imputation for individual spatial pixels at three different resolutions (100 kb (for compartment analysis), 25 kb (for visualization) and 10 kb). In brief, contacts from individual pixels were first filtered to exclude contacts overlapping the ENCODE blacklist regions. ScHiCluster then performed matrix imputation using linear convolution and random walk with restart, generating imputed sparse single-pixel contact matrices for each chromosome. To optimize storage efficiency, we output the imputed contact matrices for the whole chromosome only at 100 kb, for the 10.05-Mb window at 25 kb, and for the 5.05-Mb window at 10 kb for each pixel.

Embedding and clustering of Spatial-ATAC-Hi-C data

The GAS and GAD scores were used for spatial data embedding and clustering. To compute the GAS, we first applied the make_gene_matrix function from SnapATAC2⁹⁵ (v.2.6.4), which counts Tn5 insertions within the promoter region of each gene (2 kb upstream of the TSS) and gene body region to generate a raw count matrix. The resulting raw count matrix was used as the gene activity score matrix and imported into Seurat⁹³ (v.5.1.0) for embedding and clustering analysis. Similarly, to compute the GAD scores, we first converted the read pairs into a BED-like tabular file, with each line representing a unique fragment that was one end of the Hi-C pairs. Then, we applied the same analysis pipeline for the Hi-C data and generated the raw count matrix as the GAD score matrix. The GAD score matrix was further used for embedding and clustering analysis.

For clustering of spatial data, we first identified highly variable features from the GAS matrix/GAD score matrix using the FindVariableFeatures function. Linear transformation (scaling) was then applied to these variable features and principal-component analysis (PCA) was performed on the scaled data. We then constructed a k-nearest neighbor (kNN) graph using the FindNeighbors function and applied the Louvain algorithm via the FindClusters function to iteratively group cells together. For visualization, pixels with fewer than 2,000 fragments (empirically threshold) were assigned the majority cluster label or mean GAS/GAD score of their six nearest neighboring pixels based on Euclidean distance.

Integration analysis with single-cell ATAC-seq data in the adult mouse cortex

We used the gene raw count matrix from similar coronal slices for the integration analysis. Specifically, for mouse brain R6 sample, the scATAC-seq data from slice 6 was used; and for mouse brain R8 sample, the scATAC-seq data from slice 8 and 9 was used as a reference. To integrate our spatial data with the single-cell ATAC-seq data from adult mouse brain, we first generated the gene count matrix from the corresponding slices using the same strategy as the spatial data. The integration was then performed using Seurat R package, which basically consisted of three major steps to align the spatial gene count matrices and the single-cell gene count matrices onto the same space: (1) using dimension reduction to derive embeddings of the two datasets; (2) applying canonical component analysis (CCA) to identify transferable anchors; and (3) aligning the low-dimensional representation of the two datasets together with the anchors using MapQuery. It is difficult to scale up to millions of cells using the CCA framework of Seurat due to the memory limitations. To address this, we randomly selected ~10% of cells from the original scATAC-seq datasets as the reference.

A/B compartment analysis of Spatial-ATAC-Hi-C data

Single-pixel A/B compartments were inferred using Higashi at 100-kb resolution⁹⁶, generating a pixel-by-region PC1 value matrix. Cell type-specific A/B compartments were identified using ArchR⁹⁷. In brief, the PC1 matrix was first smoothed using the imputeMatrix function, followed by differential analysis using FindAllMarkers. Regions with P < 0.01 and log₂ fold change > 0.15 were defined as cell type-specific differential A/B compartment regions. To generate pseudobulk A/B compartment profiles for distinct cell types within a sample, we used cooltools⁹⁸ to calculate the PC1 value at 100-kb resolution.

Chromatin loop analysis of Spatial-ATAC-Hi-C data

Cell type-specific chromatin loops were identified using the following strategy: (1) chromatin loops were called from pseudobulk data for each sample using Peakachu at 10-kb resolution⁹⁹. (2) Chromatin interaction strengths at each loop were then quantified for individual spatial pixel. (3) A one-way ANOVA was performed across cell types for each loop, with loops showing P < 0.05 defined as cell type-specific chromatin loops. Cell type- or cluster-based chromatin loops were identified using Peakachu at 25-kb resolution using the corresponding sequencing depth model.

CNV analysis of Spatial-ATAC-Hi-C data

The copy number (CN) ratio was calculated from Hi-C data using the ‘calculate-cnv’ module in NeoLoopFinder⁸². All samples used in this paper were assumed to be diploid and the CN ratio here means the copy number divided by two copies. For each single pixel, the CN ratio at 5-Mb resolution was estimated, and the results were directly used for genome-wide heatmap visualization and clustering. The regional CN ratio visualization was plotted at 100-kb resolution. At sample level, the CN ratio was estimated at 100-kb resolution and a hidden Markov model-based segmentation was further performed using the ‘segment-cnv’ module to determine the boundaries of CN ratio segments.

For spatial visualization of CNVs, we used MAGIC¹⁰⁰ (Markov Affinity-based Graph Imputation of Cells) to further smooth the CN ratio, which could denoise single-pixel CN ratio by sharing information across similar cells and better reveal the spatial patterns. Basically, it (1) computes similarity between cells (typically using PCA output) and then converts to an affinity matrix using adaptive Gaussian Kernel; (2) the affinity matrix is then normalized to a Markov matrix so that each row sums to 1; and (3) the scaled CN ratio is then calculated by X_scaled = P^tX, where P is the Markov matrix, t is the diffusion time which controls the smoothing strength and X represents the original CN ratio matrix (pixels by genomic bins).

Ethical approval

The glioma specimen collection was approved by the Institutional Review Board at Northwestern University (STU00095863). Before sample collection, informed consent was obtained from each patient.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Online content

Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41592-026-03217-4.

Supplementary information

Supplementary Information^{(24.2MB, pdf)}

Supplementary Figs. 1–4 and Supplementary Tables 1–3 and 7.

Reporting Summary^{(1.7MB, pdf)}

Supplementary Table 4^{(1,022.3KB, xlsx)}

Summary of metrics for Hi-C of each pixel in spatial Hi-C for all the samples.

Supplementary Table 5^{(537KB, xlsx)}

Summary of metrics for ATAC-seq of each pixel in spatial Hi-C for all the samples.

Supplementary Table 6^{(173KB, xlsx)}

The location of differential ATAC-seq peaks between subtypes/clusters in mouse brain and tumor samples.

Supplementary Table 8^{(14.4MB, xlsx)}

TAD boundaries, chromatin loops and PC1 value for each subtype/cluster in each sample.

Supplementary Table 9^{(143.6KB, xlsx)}

Differential chromatin loops between subtypes/clusters in mouse brain and tumor samples.

Supplementary Table 10^{(13KB, xlsx)}

SV events for each sample.

Source data

Source Data Fig. 1^{(784.7KB, xlsx)}

Violin plot source data.

Source Data Fig. 2^{(9KB, xlsx)}

GO enrichment source data.

Source Data Fig. 3^{(13.5KB, xlsx)}

GO enrichment and transcription factor enrichment source data.

Source Data Fig. 5^{(223.3KB, xlsx)}

Violin plot and GO enrichment source data.

Source Data Extended Data Fig. 2^{(12.3MB, xlsx)}

Scatter-plot source data.

Source Data Extended Data Fig. 3^{(666.9KB, xlsx)}

Bar plots and Venn diagram source data.

Source Data Extended Data Fig. 4^{(1.6MB, xlsx)}

Box plot and violin plots source data.

Source Data Extended Data Fig. 6^{(10KB, xlsx)}

GO enrichment source data.

Source Data Extended Data Fig. 7^{(8.9KB, xlsx)}

Bar plot source data.

Acknowledgements

We thank all members of the Yue Laboratory for discussions throughout this project. This work made use of the NUFAB facility (RRID:SCR_017779) of Northwestern University’s NUANCE Center, which has received support from the SHyNE Resource (NSF ECCS-2025633), the IIN and Northwestern’s MRSEC program (NSF DMR-2308691). The Northwestern Nervous System Tumor Bank is supported by the P50CA221747 SPORE for Translational Approaches to Brain Cancer. We are grateful to R. Gao and M. Wang in Northwestern University for granting us access to the Nikon ECLIPSE Ts2R in Gao Laboratory and for helpful discussions.

Extended data

Author contributions

F.Y. and R.F. conceptualized the study. P.W. was responsible for the methodology. J.W. led the computational analysis. P.W., Y.C., J.H.-Y.W. and H.N. performed the experimental investigation. J.W., Q.W., S.Y. and Y.F. performed the data analysis. M.W.Y. and K.M. were responsible for clinical sample processing. B.T. and K.R.N. assisted in establishing the microfluidic platform. Y.L. and D.Z. provided technical support for data analysis. Y.D., A.S., A.B.H., A.M.S., C.M.H. and R.F. aided in the interpretation of results. P.W., J.W., Q.W. and F.Y wrote the paper with input from all authors. All authors revised, edited and approved the paper.

Peer review

Peer review information

Nature Methods thanks Yodai Takei and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editors: Lei Tang and Nina Vogt, in collaboration with the Nature Methods team.

Funding

F.Y. is supported by National Institutes of Health grants R35GM124820, 5R01LM015300 and R01HG011207. J.H.-Y.W. is supported by T32 CA009560. A.S. is supported by T32 CA070085. M.W.Y is supported by a Neurosurgery Research and Education Foundation Fellowship, the United States Department of Defense (CA230856), American Brain Tumor Association Basic Research Fellowship (fully supported by Tap Cancer Out) and the Society of Neurological Surgeons.

Data availability

Raw sequencing data, spatial barcode locations and processed files are available from the Gene Expression Omnibus under accession number GSE307620. The Brain Initiative Cell Census Network (BICCN) whole mouse brain snATAC-seq datasets were downloaded from the Gene Expression Omnibus under accession number GSE246791. The snm3C-seq data for adult mouse brain were downloaded from the Neuroscience Multi-omic Data Archive (assets.nemoarchive.org/dat-sig83t9). Source data are provided with this paper.

Code availability

The code used in this study is available at GitHub under an MIT License (https://github.com/wangjuan001/Spatial-ATAC-Hi-C).

Competing interests

A provisional patent application related to this work has been filed by Northwestern University and Yale University. F.Y. is a scientific co-founder of Sariant Therapeutics. R.F. is a scientific founder and adviser for IsoPlexis, Singleron Biotechnologies and AtlasXomics. The interests of R.F. were reviewed and managed by the Yale University Provost’s Office in accordance with the university’s conflict of interest policies. The other authors declare no competing interests.

Footnotes

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

These authors contributed equally: Ping Wang, Juan Wang, Qixuan Wang, Mark W. Youngblood.

Contributor Information

Rong Fan, Email: rong.fan@yale.edu.

Feng Yue, Email: yue@northwestern.edu.

Extended data

is available for this paper at https://doi.org/10.1038/s41592-026-03217-4.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41592-026-03217-4.

References

1.Palla, G., Fischer, D. S., Regev, A. & Theis, F. J. Spatial components of molecular tissue biology. Nat. Biotechnol.40, 308–318 (2022). [DOI] [PubMed] [Google Scholar]
2.Seferbekova, Z., Lomakin, A., Yates, L. R. & Gerstung, M. Spatial biology of cancer evolution. Nat. Rev. Genet.24, 295–313 (2023). [DOI] [PubMed] [Google Scholar]
3.Stahl, P. L. et al. Visualization and analysis of gene expression in tissue sections by spatial transcriptomics. Science353, 78–82 (2016). [DOI] [PubMed] [Google Scholar]
4.Rodriques, S. G. et al. Slide-seq: a scalable technology for measuring genome-wide expression at high spatial resolution. Science363, 1463–1467 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Stickels, R. R. et al. Highly sensitive spatial transcriptomics at near-cellular resolution with Slide-seqV2. Nat. Biotechnol.39, 313–319 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Liu, Y. et al. High-spatial-resolution multi-omics sequencing via deterministic barcoding in tissue. Cell183, 1665–1681.e18 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Cho, C. S. et al. Microscopic examination of spatial transcriptome using Seq-Scope. Cell184, 3559–3572.e22 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Fu, X. et al. Polony gels enable amplifiable DNA stamping and spatial transcriptomics of chronic pain. Cell185, 4621–4633.e17 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Chen, A. et al. Spatiotemporal transcriptomic atlas of mouse organogenesis using DNA nanoball-patterned arrays. Cell185, 1777–1792.e21 (2022). [DOI] [PubMed] [Google Scholar]
10.Deng, Y. et al. Spatial profiling of chromatin accessibility in mouse and human tissues. Nature609, 375–383 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Deng, Y. et al. Spatial-CUT&Tag: spatially resolved chromatin modification profiling at the cellular level. Science375, 681–686 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Zhao, T. et al. Spatial genomics enables multi-modal study of clonal heterogeneity in tissues. Nature601, 85–91 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Liu, Y. et al. High-plex protein and whole transcriptome co-mapping at cellular resolution with spatial CITE-seq. Nat. Biotechnol.41, 1405–1409 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Hu, B. et al. High-resolution spatially resolved proteomics of complex tissues based on microfluidics and transfer learning. Cell188, 734–748.e22 (2025). [DOI] [PubMed] [Google Scholar]
15.Guo, P. et al. Multiplexed spatial mapping of chromatin features, transcriptome and proteins in tissues. Nat. Methods22, 520–529 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Zhang, D. et al. Spatial dynamics of brain development and neuroinflammation. Nature647, 213–227 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Du, Z. et al. Allelic reprogramming of 3D chromatin architecture during early mammalian development. Nature547, 232–235 (2017). [DOI] [PubMed] [Google Scholar]
18.Dixon, J. R. et al. Chromatin architecture reorganization during stem cell differentiation. Nature518, 331–336 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Lin, D. et al. Decoding the spatial chromatin organization and dynamic epigenetic landscapes of macrophage cells during differentiation and immune activation. Nat. Commun.13, 5857 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Liu, Z., Lee, D. S., Liang, Y., Zheng, Y. & Dixon, J. R. Foxp3 orchestrates reorganization of chromatin architecture to establish regulatory T cell identity. Nat. Commun.14, 6943 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Adams, N. M. et al. Cohesin-mediated chromatin organization controls the differentiation and function of dendritic cells. Sci. Immunol.11, eadx4622 (2026). [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Xu, J. et al. Subtype-specific 3D genome alteration in acute myeloid leukaemia. Nature611, 387–398 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Flavahan, W. A. et al. Insulator dysfunction and oncogene activation in IDH mutant gliomas. Nature529, 110–114 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Ooi, W. F. et al. Integrated paired-end enhancer profiling and whole-genome sequencing reveals recurrent CCNE1 and IGF2 enhancer hijacking in primary gastric adenocarcinoma. Gut69, 1039–1052 (2020). [DOI] [PubMed] [Google Scholar]
25.Feng, Y. et al. Rewiring of 3D chromatin topology orchestrates transcriptional reprogramming and the development of human dilated cardiomyopathy. Circulation145, 1663–1683 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Lieberman-Aiden, E. et al. Comprehensive mapping of long-range interactions reveals folding principles of the human genome. Science326, 289–293 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Mumbach, M. R. et al. HiChIP: efficient and sensitive analysis of protein-directed genome architecture. Nat. Methods13, 919–922 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Fullwood, M. J. et al. An oestrogen-receptor-α-bound human chromatin interactome. Nature462, 58–64 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Krietenstein, N. et al. Ultrastructural details of mammalian chromosome architecture. Mol. Cell78, 554–565.e7 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Fang, R. et al. Mapping of long-range chromatin interactions by proximity ligation-assisted ChIP-seq. Cell Res.26, 1345–1348 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Lai, B. et al. Trac-looping measures genome structure and chromatin accessibility. Nat. Methods15, 741–747 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Wei, X. et al. HiCAR is a robust and sensitive method to analyze open-chromatin-associated genome organization. Mol. Cell82, 1225–1238.e6 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Dixon, J. R. et al. Topological domains in mammalian genomes identified by analysis of chromatin interactions. Nature485, 376–380 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Nora, E. P. et al. Spatial partitioning of the regulatory landscape of the X-inactivation centre. Nature485, 381–385 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Rao, S. S. et al. A 3D map of the human genome at kilobase resolution reveals principles of chromatin looping. Cell159, 1665–1680 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
36.Liu, Z. et al. Linking genome structures to functions by simultaneous single-cell Hi-C and RNA-seq. Science380, 1070–1076 (2023). [DOI] [PubMed] [Google Scholar]
37.Qu, J. et al. Simultaneous profiling of chromatin architecture and transcription in single cells. Nat. Struct. Mol. Biol.30, 1393–1402 (2023). [DOI] [PubMed] [Google Scholar]
38.Zhou, T. et al. GAGE-seq concurrently profiles multiscale 3D genome organization and gene expression in single cells. Nat. Genet.56, 1701–1711 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Wu, H. et al. Simultaneous single-cell three-dimensional genome and gene expression profiling uncovers dynamic enhancer connectivity underlying olfactory receptor choice. Nat. Methods21, 974–982 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
40.Chang, L. et al. Droplet Hi-C enables scalable, single-cell profiling of chromatin architecture in heterogeneous tissues. Nat. Biotechnol.43, 1694–1707 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
41.Lee, D. S. et al. Simultaneous profiling of 3D genome structure and DNA methylation in single human cells. Nat. Methods16, 999–1006 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Wu, H., Zhang, J., Tan, L. & Xie, X. S. Single-cell Micro-C profiles 3D genome structures at high resolution and characterizes multi-enhancer hubs. Nat. Genet.57, 1777–1786 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Arrastia, M. V. et al. Single-cell measurement of higher-order 3D genome organization with scSPRITE. Nat. Biotechnol.40, 64–73 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Yuang, M. et al. Embryo spatial 3D genomics. Preprint at bioRxiv 10.1101/2024.05.07.592900 (2024). [DOI]
45.Su, J. H., Zheng, P., Kinrot, S. S., Bintu, B. & Zhuang, X. Genome-scale imaging of the 3D organization and transcriptional activity of chromatin. Cell182, 1641–1659.e6 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
46.Liu, S. et al. Cell type-specific 3D-genome organization and transcription regulation in the brain. Sci. Adv.11, eadv2067 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
47.Takei, Y. et al. Integrated spatial genomics reveals global architecture of single nuclei. Nature590, 344–350 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Takei, Y. et al. Single-cell nuclear architecture across cell types in the mouse brain. Science374, 586–594 (2021). [DOI] [PubMed] [Google Scholar]
49.Takei, Y. et al. Spatial multi-omics reveals cell-type-specific nuclear compartments. Nature641, 1037–1047 (2025). [DOI] [PubMed] [Google Scholar]
50.Liu, M. et al. Tracing the evolution of single-cell 3D genomes in Kras-driven cancers. Nat. Genet.57, 3075–3087 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
51.Bintu, B. et al. Super-resolution chromatin tracing reveals domains and cooperative interactions in single cells. Science362, eaau1783 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
52.Liu, M. et al. Multiplexed imaging of nucleome architectures in single cells of mammalian tissue. Nat. Commun.11, 2907 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
53.Mateo, L. J. et al. Visualizing DNA folding and RNA in embryos at single-cell resolution. Nature568, 49–54 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
54.Cardozo Gizzi, A. M. et al. Microscopy-based chromosome conformation capture enables simultaneous visualization of genome organization and transcription in intact organisms. Mol. Cell74, 212–222.e5 (2019). [DOI] [PubMed] [Google Scholar]
55.Payne, A. C. et al. In situ genome sequencing resolves DNA sequence and structure in intact biological samples. Science371, eaay3446 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
56.Labade, A. S. et al. Expansion in situ genome sequencing links nuclear abnormalities to aberrant chromatin regulation. Science389, eadt2781 (2025). [DOI] [PubMed] [Google Scholar]
57.Zhang, D. et al. Spatial epigenome–transcriptome co-profiling of mammalian tissues. Nature616, 113–122 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
58.Tan, L. et al. Changes in genome architecture and transcriptional dynamics progress independently of sensory experience during post-natal brain development. Cell184, 741–758.e17 (2021). [DOI] [PubMed] [Google Scholar]
59.Gorkin, D. U. et al. An atlas of dynamic chromatin landscapes in mouse fetal development. Nature583, 744–751 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
60.Nagano, T. et al. Single-cell Hi-C reveals cell-to-cell variability in chromosome structure. Nature502, 59–64 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
61.Ramani, V. et al. Massively multiplex single-cell Hi-C. Nat. Methods14, 263–266 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
62.Shen, S., Zheng, Y. & Keles, S. scGAD: single-cell gene associating domain scores for exploratory analysis of scHi-C data. Bioinformatics38, 3642–3644 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
63.Pliner, H. A. et al. Cicero predicts cis-regulatory DNA interactions from single-cell chromatin accessibility data. Mol. Cell71, 858–871.e8 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
64.Zhang, M. et al. Molecularly defined and spatially resolved cell atlas of the whole mouse brain. Nature624, 343–354 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
65.Zu, S. et al. Single-cell analysis of chromatin accessibility in the adult mouse brain. Nature624, 378–389 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
66.Yao, Z. et al. A high-resolution transcriptomic and spatial atlas of cell types in the whole mouse brain. Nature624, 317–332 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
67.Stanley, G., Gokce, O., Malenka, R. C., Sudhof, T. C. & Quake, S. R. Continuous and discrete neuron types of the adult murine striatum. Neuron105, 688–699.e8 (2020). [DOI] [PubMed] [Google Scholar]
68.Zhou, Q. & Anderson, D. J. The bHLH transcription factors OLIG2 and OLIG1 couple neuronal and glial subtype specification. Cell109, 61–73 (2002). [DOI] [PubMed] [Google Scholar]
69.Mallika, C., Guo, Q. & Li, J. Y. Gbx2 is essential for maintaining thalamic neuron identity and repressing habenular characters in the developing thalamus. Dev. Biol.407, 26–39 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
70.Liu, H. et al. Single-cell DNA methylome and 3D multi-omic atlas of the adult mouse brain. Nature624, 366–377 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
71.Ding, B. et al. Reciprocal autoregulation by NFI occupancy and ETV1 promotes the developmental expression of dendrite-synapse genes in cerebellar granule neurons. Mol. Biol. Cell27, 1488–1499 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
72.Miyata, T., Maeda, T. & Lee, J. E. NeuroD is required for differentiation of the granule cells in the cerebellum and hippocampus. Genes Dev.13, 1647–1652 (1999). [DOI] [PMC free article] [PubMed] [Google Scholar]
73.Li, Y. E. et al. A comparative atlas of single-cell chromatin accessibility in the human brain. Science382, eadf7044 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
74.Dixon, J. R. et al. Integrative detection and analysis of structural variation in cancer genomes. Nat. Genet.50, 1388–1398 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
75.Wang, S. et al. HiNT: a computational method for detecting copy number variations and translocations from Hi-C data. Genome Biol.21, 73 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
76.Wang, X., Luan, Y. & Yue, F. EagleC: a deep-learning framework for detecting a full range of structural variations from bulk and single-cell contact maps. Sci. Adv.8, eabn9215 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
77.Mathur, R. et al. Glioblastoma evolution and heterogeneity from a 3D whole-tumor perspective. Cell187, 446–463.e16 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
78.Wang, Q. et al. Spatial 3D genome organization reveals intratumor heterogeneity in primary glioblastoma samples. Sci. Adv.11, eadn2830 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
79.Liu, T. et al. Enhancer coamplification and hijacking promote oncogene expression in liposarcoma. Cancer Res.83, 1517–1530 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
80.Okonechnikov, K. et al. 3D genome mapping identifies subgroup-specific chromosome conformations and tumor-dependency genes in ependymoma. Nat. Commun.14, 2300 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
81.Zhao, S. G. et al. Integrated analyses highlight interactions between the three-dimensional genome and DNA, RNA and epigenomic alterations in metastatic prostate cancer. Nat. Genet.56, 1689–1700 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
82.Wang, X. et al. Genome-wide detection of enhancer-hijacking events from chromatin interaction data in rearranged genomes. Nat. Methods18, 661–668 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
83.Rausch, T. et al. DELLY: structural variant discovery by integrated paired-end and split-read analysis. Bioinformatics28, i333–i339 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
84.Layer, R. M., Chiang, C., Quinlan, A. R. & Hall, I. M. LUMPY: a probabilistic framework for structural variant discovery. Genome Biol. 15, R84 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
85.Mamlouk, S. et al. DNA copy number changes define spatial patterns of heterogeneity in colorectal cancer. Nat. Commun.8, 14093 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
86.Wu, S. et al. Circular ecDNA promotes accessible chromatin and high oncogene expression. Nature575, 699–703 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
87.Noorani, I. et al. Extrachromosomal DNA-driven oncogene spatial heterogeneity and evolution in glioblastoma. Cancer Discov.15, 2078–2095 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
88.Zhao, Y. et al. Stereo-seq V2: spatial mapping of total RNA on FFPE sections with high resolution. Cell188, 6554–6571.e21 (2025). [DOI] [PubMed] [Google Scholar]
89.Bai, Z. et al. Spatially exploring RNA biology in archival formalin-fixed paraffin-embedded tissues. Cell187, 6760–6779.e24 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
90.Guo, P. et al. Spatial profiling of chromatin accessibility in formalin-fixed paraffin-embedded tissues. Nat. Commun.16, 5945 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
91.Su, G. et al. Spatial multi-omics sequencing for fixed tissue via DBiT-seq. STAR Protoc.2, 100532 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
92.Wang, X. runHiC: a user-friendly Hi-C data processing software based on hiclib. Zenodo 10.5281/zenodo.55324 (2016). [DOI]
93.Hao, Y. et al. Dictionary learning for integrative, multimodal and scalable single-cell analysis. Nat. Biotechnol.42, 293–304 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
94.Zhou, J. et al. Robust single-cell Hi-C clustering by convolution- and random-walk-based imputation. Proc. Natl Acad. Sci. USA116, 14011–14018 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
95.Zhang, K., Zemke, N. R., Armand, E. J. & Ren, B. A fast, scalable and versatile tool for analysis of single-cell omics data. Nat. Methods21, 217–227 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
96.Zhang, R., Zhou, T. & Ma, J. Multiscale and integrative single-cell Hi-C analysis with Higashi. Nat. Biotechnol.40, 254–261 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
97.Granja, J. M. et al. ArchR is a scalable software package for integrative single-cell chromatin accessibility analysis. Nat. Genet.53, 403–411 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
98.Open2C, et al. Cooltools: enabling high-resolution Hi-C analysis in Python. PLoS Comput. Biol.20, e1012067 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
99.Salameh, T. J. et al. A supervised learning framework for chromatin loop detection in genome-wide contact maps. Nat. Commun.11, 3428 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
100.van Dijk, D. et al. Recovering gene interactions from single-cell data using data diffusion. Cell174, 716–729.e27 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
101.Llorens-Bobadilla, E. et al. Solid-phase capture and profiling of open chromatin by spatial ATAC. Nat. Biotechnol.41, 1085–1088 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Information^{(24.2MB, pdf)}

Supplementary Figs. 1–4 and Supplementary Tables 1–3 and 7.

Reporting Summary^{(1.7MB, pdf)}

Supplementary Table 4^{(1,022.3KB, xlsx)}

Summary of metrics for Hi-C of each pixel in spatial Hi-C for all the samples.

Supplementary Table 5^{(537KB, xlsx)}

Summary of metrics for ATAC-seq of each pixel in spatial Hi-C for all the samples.

Supplementary Table 6^{(173KB, xlsx)}

The location of differential ATAC-seq peaks between subtypes/clusters in mouse brain and tumor samples.

Supplementary Table 8^{(14.4MB, xlsx)}

TAD boundaries, chromatin loops and PC1 value for each subtype/cluster in each sample.

Supplementary Table 9^{(143.6KB, xlsx)}

Differential chromatin loops between subtypes/clusters in mouse brain and tumor samples.

Supplementary Table 10^{(13KB, xlsx)}

SV events for each sample.

Source Data Fig. 1^{(784.7KB, xlsx)}

Violin plot source data.

Source Data Fig. 2^{(9KB, xlsx)}

GO enrichment source data.

Source Data Fig. 3^{(13.5KB, xlsx)}

GO enrichment and transcription factor enrichment source data.

Source Data Fig. 5^{(223.3KB, xlsx)}

Violin plot and GO enrichment source data.

Source Data Extended Data Fig. 2^{(12.3MB, xlsx)}

Scatter-plot source data.

Source Data Extended Data Fig. 3^{(666.9KB, xlsx)}

Bar plots and Venn diagram source data.

Source Data Extended Data Fig. 4^{(1.6MB, xlsx)}

Box plot and violin plots source data.

Source Data Extended Data Fig. 6^{(10KB, xlsx)}

GO enrichment source data.

Source Data Extended Data Fig. 7^{(8.9KB, xlsx)}

Bar plot source data.

Data Availability Statement

The code used in this study is available at GitHub under an MIT License (https://github.com/wangjuan001/Spatial-ATAC-Hi-C).

[CR1] 1.Palla, G., Fischer, D. S., Regev, A. & Theis, F. J. Spatial components of molecular tissue biology. Nat. Biotechnol.40, 308–318 (2022). [DOI] [PubMed] [Google Scholar]

[CR2] 2.Seferbekova, Z., Lomakin, A., Yates, L. R. & Gerstung, M. Spatial biology of cancer evolution. Nat. Rev. Genet.24, 295–313 (2023). [DOI] [PubMed] [Google Scholar]

[CR3] 3.Stahl, P. L. et al. Visualization and analysis of gene expression in tissue sections by spatial transcriptomics. Science353, 78–82 (2016). [DOI] [PubMed] [Google Scholar]

[CR4] 4.Rodriques, S. G. et al. Slide-seq: a scalable technology for measuring genome-wide expression at high spatial resolution. Science363, 1463–1467 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR5] 5.Stickels, R. R. et al. Highly sensitive spatial transcriptomics at near-cellular resolution with Slide-seqV2. Nat. Biotechnol.39, 313–319 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR6] 6.Liu, Y. et al. High-spatial-resolution multi-omics sequencing via deterministic barcoding in tissue. Cell183, 1665–1681.e18 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR7] 7.Cho, C. S. et al. Microscopic examination of spatial transcriptome using Seq-Scope. Cell184, 3559–3572.e22 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR8] 8.Fu, X. et al. Polony gels enable amplifiable DNA stamping and spatial transcriptomics of chronic pain. Cell185, 4621–4633.e17 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR9] 9.Chen, A. et al. Spatiotemporal transcriptomic atlas of mouse organogenesis using DNA nanoball-patterned arrays. Cell185, 1777–1792.e21 (2022). [DOI] [PubMed] [Google Scholar]

[CR10] 10.Deng, Y. et al. Spatial profiling of chromatin accessibility in mouse and human tissues. Nature609, 375–383 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR11] 11.Deng, Y. et al. Spatial-CUT&Tag: spatially resolved chromatin modification profiling at the cellular level. Science375, 681–686 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR12] 12.Zhao, T. et al. Spatial genomics enables multi-modal study of clonal heterogeneity in tissues. Nature601, 85–91 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR13] 13.Liu, Y. et al. High-plex protein and whole transcriptome co-mapping at cellular resolution with spatial CITE-seq. Nat. Biotechnol.41, 1405–1409 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR14] 14.Hu, B. et al. High-resolution spatially resolved proteomics of complex tissues based on microfluidics and transfer learning. Cell188, 734–748.e22 (2025). [DOI] [PubMed] [Google Scholar]

[CR15] 15.Guo, P. et al. Multiplexed spatial mapping of chromatin features, transcriptome and proteins in tissues. Nat. Methods22, 520–529 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR16] 16.Zhang, D. et al. Spatial dynamics of brain development and neuroinflammation. Nature647, 213–227 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR17] 17.Du, Z. et al. Allelic reprogramming of 3D chromatin architecture during early mammalian development. Nature547, 232–235 (2017). [DOI] [PubMed] [Google Scholar]

[CR18] 18.Dixon, J. R. et al. Chromatin architecture reorganization during stem cell differentiation. Nature518, 331–336 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR19] 19.Lin, D. et al. Decoding the spatial chromatin organization and dynamic epigenetic landscapes of macrophage cells during differentiation and immune activation. Nat. Commun.13, 5857 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR20] 20.Liu, Z., Lee, D. S., Liang, Y., Zheng, Y. & Dixon, J. R. Foxp3 orchestrates reorganization of chromatin architecture to establish regulatory T cell identity. Nat. Commun.14, 6943 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR21] 21.Adams, N. M. et al. Cohesin-mediated chromatin organization controls the differentiation and function of dendritic cells. Sci. Immunol.11, eadx4622 (2026). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR22] 22.Xu, J. et al. Subtype-specific 3D genome alteration in acute myeloid leukaemia. Nature611, 387–398 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR23] 23.Flavahan, W. A. et al. Insulator dysfunction and oncogene activation in IDH mutant gliomas. Nature529, 110–114 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR24] 24.Ooi, W. F. et al. Integrated paired-end enhancer profiling and whole-genome sequencing reveals recurrent CCNE1 and IGF2 enhancer hijacking in primary gastric adenocarcinoma. Gut69, 1039–1052 (2020). [DOI] [PubMed] [Google Scholar]

[CR25] 25.Feng, Y. et al. Rewiring of 3D chromatin topology orchestrates transcriptional reprogramming and the development of human dilated cardiomyopathy. Circulation145, 1663–1683 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR26] 26.Lieberman-Aiden, E. et al. Comprehensive mapping of long-range interactions reveals folding principles of the human genome. Science326, 289–293 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR27] 27.Mumbach, M. R. et al. HiChIP: efficient and sensitive analysis of protein-directed genome architecture. Nat. Methods13, 919–922 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR28] 28.Fullwood, M. J. et al. An oestrogen-receptor-α-bound human chromatin interactome. Nature462, 58–64 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR29] 29.Krietenstein, N. et al. Ultrastructural details of mammalian chromosome architecture. Mol. Cell78, 554–565.e7 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR30] 30.Fang, R. et al. Mapping of long-range chromatin interactions by proximity ligation-assisted ChIP-seq. Cell Res.26, 1345–1348 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR31] 31.Lai, B. et al. Trac-looping measures genome structure and chromatin accessibility. Nat. Methods15, 741–747 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR32] 32.Wei, X. et al. HiCAR is a robust and sensitive method to analyze open-chromatin-associated genome organization. Mol. Cell82, 1225–1238.e6 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR33] 33.Dixon, J. R. et al. Topological domains in mammalian genomes identified by analysis of chromatin interactions. Nature485, 376–380 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR34] 34.Nora, E. P. et al. Spatial partitioning of the regulatory landscape of the X-inactivation centre. Nature485, 381–385 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR35] 35.Rao, S. S. et al. A 3D map of the human genome at kilobase resolution reveals principles of chromatin looping. Cell159, 1665–1680 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR36] 36.Liu, Z. et al. Linking genome structures to functions by simultaneous single-cell Hi-C and RNA-seq. Science380, 1070–1076 (2023). [DOI] [PubMed] [Google Scholar]

[CR37] 37.Qu, J. et al. Simultaneous profiling of chromatin architecture and transcription in single cells. Nat. Struct. Mol. Biol.30, 1393–1402 (2023). [DOI] [PubMed] [Google Scholar]

[CR38] 38.Zhou, T. et al. GAGE-seq concurrently profiles multiscale 3D genome organization and gene expression in single cells. Nat. Genet.56, 1701–1711 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR39] 39.Wu, H. et al. Simultaneous single-cell three-dimensional genome and gene expression profiling uncovers dynamic enhancer connectivity underlying olfactory receptor choice. Nat. Methods21, 974–982 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR40] 40.Chang, L. et al. Droplet Hi-C enables scalable, single-cell profiling of chromatin architecture in heterogeneous tissues. Nat. Biotechnol.43, 1694–1707 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR41] 41.Lee, D. S. et al. Simultaneous profiling of 3D genome structure and DNA methylation in single human cells. Nat. Methods16, 999–1006 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR42] 42.Wu, H., Zhang, J., Tan, L. & Xie, X. S. Single-cell Micro-C profiles 3D genome structures at high resolution and characterizes multi-enhancer hubs. Nat. Genet.57, 1777–1786 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR43] 43.Arrastia, M. V. et al. Single-cell measurement of higher-order 3D genome organization with scSPRITE. Nat. Biotechnol.40, 64–73 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR44] 44.Yuang, M. et al. Embryo spatial 3D genomics. Preprint at bioRxiv 10.1101/2024.05.07.592900 (2024). [DOI]

[CR45] 45.Su, J. H., Zheng, P., Kinrot, S. S., Bintu, B. & Zhuang, X. Genome-scale imaging of the 3D organization and transcriptional activity of chromatin. Cell182, 1641–1659.e6 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR46] 46.Liu, S. et al. Cell type-specific 3D-genome organization and transcription regulation in the brain. Sci. Adv.11, eadv2067 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR47] 47.Takei, Y. et al. Integrated spatial genomics reveals global architecture of single nuclei. Nature590, 344–350 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR48] 48.Takei, Y. et al. Single-cell nuclear architecture across cell types in the mouse brain. Science374, 586–594 (2021). [DOI] [PubMed] [Google Scholar]

[CR49] 49.Takei, Y. et al. Spatial multi-omics reveals cell-type-specific nuclear compartments. Nature641, 1037–1047 (2025). [DOI] [PubMed] [Google Scholar]

[CR50] 50.Liu, M. et al. Tracing the evolution of single-cell 3D genomes in Kras-driven cancers. Nat. Genet.57, 3075–3087 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR51] 51.Bintu, B. et al. Super-resolution chromatin tracing reveals domains and cooperative interactions in single cells. Science362, eaau1783 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR52] 52.Liu, M. et al. Multiplexed imaging of nucleome architectures in single cells of mammalian tissue. Nat. Commun.11, 2907 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR53] 53.Mateo, L. J. et al. Visualizing DNA folding and RNA in embryos at single-cell resolution. Nature568, 49–54 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR54] 54.Cardozo Gizzi, A. M. et al. Microscopy-based chromosome conformation capture enables simultaneous visualization of genome organization and transcription in intact organisms. Mol. Cell74, 212–222.e5 (2019). [DOI] [PubMed] [Google Scholar]

[CR55] 55.Payne, A. C. et al. In situ genome sequencing resolves DNA sequence and structure in intact biological samples. Science371, eaay3446 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR56] 56.Labade, A. S. et al. Expansion in situ genome sequencing links nuclear abnormalities to aberrant chromatin regulation. Science389, eadt2781 (2025). [DOI] [PubMed] [Google Scholar]

[CR57] 57.Zhang, D. et al. Spatial epigenome–transcriptome co-profiling of mammalian tissues. Nature616, 113–122 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR58] 58.Tan, L. et al. Changes in genome architecture and transcriptional dynamics progress independently of sensory experience during post-natal brain development. Cell184, 741–758.e17 (2021). [DOI] [PubMed] [Google Scholar]

[CR59] 59.Gorkin, D. U. et al. An atlas of dynamic chromatin landscapes in mouse fetal development. Nature583, 744–751 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR60] 60.Nagano, T. et al. Single-cell Hi-C reveals cell-to-cell variability in chromosome structure. Nature502, 59–64 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR61] 61.Ramani, V. et al. Massively multiplex single-cell Hi-C. Nat. Methods14, 263–266 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR62] 62.Shen, S., Zheng, Y. & Keles, S. scGAD: single-cell gene associating domain scores for exploratory analysis of scHi-C data. Bioinformatics38, 3642–3644 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR63] 63.Pliner, H. A. et al. Cicero predicts cis-regulatory DNA interactions from single-cell chromatin accessibility data. Mol. Cell71, 858–871.e8 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR64] 64.Zhang, M. et al. Molecularly defined and spatially resolved cell atlas of the whole mouse brain. Nature624, 343–354 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR65] 65.Zu, S. et al. Single-cell analysis of chromatin accessibility in the adult mouse brain. Nature624, 378–389 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR66] 66.Yao, Z. et al. A high-resolution transcriptomic and spatial atlas of cell types in the whole mouse brain. Nature624, 317–332 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR67] 67.Stanley, G., Gokce, O., Malenka, R. C., Sudhof, T. C. & Quake, S. R. Continuous and discrete neuron types of the adult murine striatum. Neuron105, 688–699.e8 (2020). [DOI] [PubMed] [Google Scholar]

[CR68] 68.Zhou, Q. & Anderson, D. J. The bHLH transcription factors OLIG2 and OLIG1 couple neuronal and glial subtype specification. Cell109, 61–73 (2002). [DOI] [PubMed] [Google Scholar]

[CR69] 69.Mallika, C., Guo, Q. & Li, J. Y. Gbx2 is essential for maintaining thalamic neuron identity and repressing habenular characters in the developing thalamus. Dev. Biol.407, 26–39 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR70] 70.Liu, H. et al. Single-cell DNA methylome and 3D multi-omic atlas of the adult mouse brain. Nature624, 366–377 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR71] 71.Ding, B. et al. Reciprocal autoregulation by NFI occupancy and ETV1 promotes the developmental expression of dendrite-synapse genes in cerebellar granule neurons. Mol. Biol. Cell27, 1488–1499 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR72] 72.Miyata, T., Maeda, T. & Lee, J. E. NeuroD is required for differentiation of the granule cells in the cerebellum and hippocampus. Genes Dev.13, 1647–1652 (1999). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR73] 73.Li, Y. E. et al. A comparative atlas of single-cell chromatin accessibility in the human brain. Science382, eadf7044 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR74] 74.Dixon, J. R. et al. Integrative detection and analysis of structural variation in cancer genomes. Nat. Genet.50, 1388–1398 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR75] 75.Wang, S. et al. HiNT: a computational method for detecting copy number variations and translocations from Hi-C data. Genome Biol.21, 73 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR76] 76.Wang, X., Luan, Y. & Yue, F. EagleC: a deep-learning framework for detecting a full range of structural variations from bulk and single-cell contact maps. Sci. Adv.8, eabn9215 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR77] 77.Mathur, R. et al. Glioblastoma evolution and heterogeneity from a 3D whole-tumor perspective. Cell187, 446–463.e16 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR78] 78.Wang, Q. et al. Spatial 3D genome organization reveals intratumor heterogeneity in primary glioblastoma samples. Sci. Adv.11, eadn2830 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR79] 79.Liu, T. et al. Enhancer coamplification and hijacking promote oncogene expression in liposarcoma. Cancer Res.83, 1517–1530 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR80] 80.Okonechnikov, K. et al. 3D genome mapping identifies subgroup-specific chromosome conformations and tumor-dependency genes in ependymoma. Nat. Commun.14, 2300 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR81] 81.Zhao, S. G. et al. Integrated analyses highlight interactions between the three-dimensional genome and DNA, RNA and epigenomic alterations in metastatic prostate cancer. Nat. Genet.56, 1689–1700 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR82] 82.Wang, X. et al. Genome-wide detection of enhancer-hijacking events from chromatin interaction data in rearranged genomes. Nat. Methods18, 661–668 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR83] 83.Rausch, T. et al. DELLY: structural variant discovery by integrated paired-end and split-read analysis. Bioinformatics28, i333–i339 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR84] 84.Layer, R. M., Chiang, C., Quinlan, A. R. & Hall, I. M. LUMPY: a probabilistic framework for structural variant discovery. Genome Biol. 15, R84 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR85] 85.Mamlouk, S. et al. DNA copy number changes define spatial patterns of heterogeneity in colorectal cancer. Nat. Commun.8, 14093 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR86] 86.Wu, S. et al. Circular ecDNA promotes accessible chromatin and high oncogene expression. Nature575, 699–703 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR87] 87.Noorani, I. et al. Extrachromosomal DNA-driven oncogene spatial heterogeneity and evolution in glioblastoma. Cancer Discov.15, 2078–2095 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR88] 88.Zhao, Y. et al. Stereo-seq V2: spatial mapping of total RNA on FFPE sections with high resolution. Cell188, 6554–6571.e21 (2025). [DOI] [PubMed] [Google Scholar]

[CR89] 89.Bai, Z. et al. Spatially exploring RNA biology in archival formalin-fixed paraffin-embedded tissues. Cell187, 6760–6779.e24 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR90] 90.Guo, P. et al. Spatial profiling of chromatin accessibility in formalin-fixed paraffin-embedded tissues. Nat. Commun.16, 5945 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR91] 91.Su, G. et al. Spatial multi-omics sequencing for fixed tissue via DBiT-seq. STAR Protoc.2, 100532 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR92] 92.Wang, X. runHiC: a user-friendly Hi-C data processing software based on hiclib. Zenodo 10.5281/zenodo.55324 (2016). [DOI]

[CR93] 93.Hao, Y. et al. Dictionary learning for integrative, multimodal and scalable single-cell analysis. Nat. Biotechnol.42, 293–304 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR94] 94.Zhou, J. et al. Robust single-cell Hi-C clustering by convolution- and random-walk-based imputation. Proc. Natl Acad. Sci. USA116, 14011–14018 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR95] 95.Zhang, K., Zemke, N. R., Armand, E. J. & Ren, B. A fast, scalable and versatile tool for analysis of single-cell omics data. Nat. Methods21, 217–227 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR96] 96.Zhang, R., Zhou, T. & Ma, J. Multiscale and integrative single-cell Hi-C analysis with Higashi. Nat. Biotechnol.40, 254–261 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR97] 97.Granja, J. M. et al. ArchR is a scalable software package for integrative single-cell chromatin accessibility analysis. Nat. Genet.53, 403–411 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR98] 98.Open2C, et al. Cooltools: enabling high-resolution Hi-C analysis in Python. PLoS Comput. Biol.20, e1012067 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR99] 99.Salameh, T. J. et al. A supervised learning framework for chromatin loop detection in genome-wide contact maps. Nat. Commun.11, 3428 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR100] 100.van Dijk, D. et al. Recovering gene interactions from single-cell data using data diffusion. Cell174, 716–729.e27 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR101] 101.Llorens-Bobadilla, E. et al. Solid-phase capture and profiling of open chromatin by spatial ATAC. Nat. Biotechnol.41, 1085–1088 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Spatial chromatin architecture and accessibility co-profiling of mammalian tissues

Ping Wang

Juan Wang

Qixuan Wang

Mark W Youngblood

Yang Cheng

Bo Tao

Josiah Hiu-Yuen Wong

Yu Luan

Shaojun Yu

Alok Swaroop

Khizar R Nandoliya

Hinda Najem

Yihao Fu

Kathleen McCortney

Di Zhang

Yarui Diao

Amy B Heimberger

Adam M Sonabend

Craig M Horbinski

Rong Fan

Feng Yue

Abstract

Main

Results

Development of Spatial-ATAC-Hi-C technology

Fig. 1. Design and evaluation of Spatial-ATAC-Hi-C data.

Quality assessment and benchmarking of Spatial-ATAC-Hi-C

Extended Data Fig. 1. Quality-control assessment of Hi-C data modality from Spatial-ATAC-Hi-C libraries.

Extended Data Fig. 2. Quality-control assessment of ATAC-seq data modality from Spatial-ATAC-Hi-C libraries.

Extended Data Fig. 3. Quality-control assessment of the Spatial-ATAC-Hi-C libraries.

Extended Data Fig. 4. Quality-control assessment of the Spatial-ATAC-Hi-C libraries at single pixel level.

Spatial-ATAC-Hi-C captures complex cell identities in adult mouse brain

Fig. 2. Spatial-ATAC-Hi-C reveals spatially resolved cell types of mouse brain.

Extended Data Fig. 5. Spatial co-profiling of chromatin structure and accessibility in mouse brain R8.

Spatial-ATAC-Hi-C reveals spatially resolved, cell-type-specific 3D chromatin architecture

Extended Data Fig. 6. Spatial-ATAC-Hi-C reveals spatially resolved, cell-type-specific chromatin loops in mouse brain samples.

Fig. 3. Spatial-ATAC-Hi-C reveals spatial heterogeneity of 3D genome architectures in mouse brain.

Spatial co-profiling of chromatin structure and accessibility in human brain

Spatial-ATAC-Hi-C reveals spatial distribution of CNVs

Fig. 4. Spatial-ATAC-Hi-C can faithfully capture spatial copy number differences.

Extended Data Fig. 7. Structural variations (SVs) identified by Spatial-ATAC-Hi-C were supported by WGS data.

Spatial 3D genome structure and chromatin accessibility co-profiling of glioblastoma

Fig. 5. Spatial 3D genome and epigenome co-profiling of GBM sample.

Discussion

Methods

Preparation of tissue slides

Preparation of Tn5 transposome

DNA barcode sequencing, annealing and DNA oligonucleotides

Photomask fabrication

Polydimethylsiloxane mold fabrication

PDMS microfluidic chip fabrication

Experimental protocol for Spatial-ATAC-Hi-C

Histological image and analysis of cell number per pixel

Data preprocessing

Quality control of Spatial-ATAC-Hi-C data

Image processing

Imputation of Spatial-ATAC-Hi-C data

Embedding and clustering of Spatial-ATAC-Hi-C data

Integration analysis with single-cell ATAC-seq data in the adult mouse cortex

A/B compartment analysis of Spatial-ATAC-Hi-C data

Chromatin loop analysis of Spatial-ATAC-Hi-C data

CNV analysis of Spatial-ATAC-Hi-C data

Ethical approval

Reporting summary

Online content

Supplementary information

Source data

Acknowledgements

Extended data

Author contributions

Peer review

Peer review information

Funding

Data availability

Code availability

Competing interests

Footnotes

Contributor Information