Help

How to read scAtria

About scAtria

scAtria curates 94 datasets from 82 studies, covering 7,709 samples and 128,508,915 cells across 14 spatial transcriptomic and proteomic technologies. Of these, 80 datasets, 7,487 samples, and 118,589,227 cells have curated tiered cell ontology suitable for cross-study comparison.

Each record links study metadata, sample-level organ identity, disease status, cell-ontology tiers, coordinate-linked in situ views, and available spatial or marker outputs. Start from Home, filter records in Browse, inspect study-level dataset pages, review catalog statistics below, explore Domains, Interaction, and Topology, or retrieve tables from Download.

Current published spatial outputs span 71 datasets, 7,269 sample graphs, 1,273,706 interaction pairs, 216,895 topology nodes, 989,821 topology edges, and 79,830 domain entries, supporting cross-organ and cross-condition comparison of spatial compartments, contact structure, and tissue architecture.

Inspired by Roman atria, architectural spaces where light, air, water, movement, and reception came together, scAtria treats cells like inhabitants of a shared tissue interior: spatial context reveals encounters, neighborhoods, and biological exchange.

A Pompeian Interior by Luigi Bazzani, 1882
A Pompeian Interior, Luigi Bazzani, 1882

Catalog and ontology statistics

94
80 curated datasets
7,709
7,487 curated samples
128,508,915
118,589,227 curated cells
272
curated ontology labels
Spatial transcriptomics 61 datasets · 1,777 samples
Cells61.1%
Datasets64.9%
Spatial proteomics 33 datasets · 5,932 samples
Cells38.9%
Datasets35.1%

Statistics by study

Top studies by cells82 studies
StudyDatasetsSamplesCells
MERFISH 500-plex Vizgen showcase 1 18 10,196,835
NSCLC_38242124 1 2,070 5,984,454
B-NHL_38379051 1 35 5,690,284
10x Genomics Visium HD showcase 1 24 5,379,712
Xenium Human Multi-Tissue and Cancer Panel showcase 1 20 5,095,938
SPATCH_41107232 4 18 4,948,582
10x Genomics Xenium Prime 5K showcase 1 9 4,598,042
DKD_42056516 2 64 4,337,862

Statistics by technique

Platform footprint14 technologies
CODEX 29,954,127
Xenium 28,858,054
CosMx 22,225,127
MERFISH 15,976,076
IMC 12,063,964
Visium HD 7,748,903
CyCIF 3,293,552
CosMx-P 1,638,193

Statistics by anatomy

Organ footprint10 shown
Lymphoid tissue 450 samples
Large intestine 869 samples
Lung 2,814 samples
Breast 1,099 samples
Liver 515 samples
Ovary 223 samples
Skin 224 samples
Kidney 70 samples
Uterus/Cervix 7 samples
Brain 489 samples

Statistics by cell ontology

Tier summaries272 labels
Tier 1 labelCellsDatasetsStudies
Immune 40,692,308 77 66
Tumor/Malignant 30,818,752 51 47
Stromal 24,107,754 78 67
Epithelial 15,091,372 41 36
Other 3,479,430 42 39
Neural/Glial 2,225,441 21 18
Hematopoietic 1,302,223 9 8
Tier 2 labelCellsDatasetsStudies
Tumor/Malignant 30,818,752 51 47
Lymphoid 26,640,949 73 63
Epithelial 15,091,372 41 36
Stromal 14,600,779 69 62
Myeloid 12,667,969 73 63
Vascular stromal 9,377,092 73 63
Other 3,479,430 42 39
Neural/Glial 2,218,239 21 18
Immune 1,383,390 15 14
Erythroid 1,078,806 7 7
Hematopoietic progenitor 199,693 3 3
Immune-associated stromal 96,170 5 5
Skeletal stromal 28,867 3 3
Megakaryocytic 17,430 5 4
Tier 3 labelCellsDatasetsStudies
Tumor/Malignant cell 30,818,752 51 47
T cell 14,157,555 71 62
Fibroblast 10,719,947 56 49
B cell 9,159,236 64 56
Macrophage 6,156,901 64 56
Endothelial cell 5,344,234 73 63
Mural cell 4,032,858 39 36
Epithelial cell, unspecified 3,935,409 23 20
Stromal cell, unspecified 3,696,984 24 24
Other 3,415,470 41 38
Tubular epithelial cell 3,196,347 5 4
Enterocyte 2,160,453 9 8
Plasma cell 1,956,982 43 41
Myeloid cell, unspecified 1,736,266 14 14
Tier 4 labelCellsDatasetsStudies
Tumor/Malignant cell 30,818,752 51 47
B cell 9,024,583 64 56
Fibroblast 9,004,274 56 49
Macrophage 6,149,500 64 56
CD4 Tconv 5,595,250 42 34
Endothelial cell 4,050,020 72 62
Pericyte/Smooth muscle 3,964,135 37 34
Epithelial cell, unspecified 3,777,697 17 14
Stromal cell, unspecified 3,661,788 23 23
Other 3,415,470 41 38
CD8 T 3,083,277 45 37
T cell, unspecified 2,970,584 29 24
Enterocyte 2,072,333 6 5
Plasma cell 1,956,982 43 41

Spatial analyses

Result pages show precomputed modules; Help keeps the interpretation rules.

TiersSeparated by chart; never mixed in one analytical plot
GraphPer-sample Delaunay with distance filtering
NANot available in the current release; not a biological zero

Domains

Label-only, data-driven neighborhood compartments from coordinates plus Tier 2 or Tier 3 labels.

Domain IDSample-local; D1 is not cross-study equivalent
FamilyCross-sample grouping by composition profile
TierTier 2 is stable; Tier 3 is more specific
Method and parameters
InputsCoordinates and tiered labels only
OutputsDomain summary, composition, interfaces, mixing metrics, PNG previews
CaveatData-driven label neighborhoods, not histology-confirmed regions

Interaction Propensity

Directed cell-type contacts on a sample-level spatial graph. Matrix X is from_label; Y is to_label.

DirectionA to B and B to A remain separate
ColorRed enriched; blue depleted
EngineimcRtools on distance-limited Delaunay graph
Method and parameters
InputsCoordinates, graph, Tier 3 or Tier 4 labels
Permutations100 for routine atlas updates
OutputsNative engine fields plus display-ready CSVs

Network Topology

Sample-level graph architecture after cells are grouped by curated ontology labels.

ModularityCommunity separation
AssortativitySimilar labels preferentially connect
BetweennessBridge-like role between graph regions or lineages
Method and parameters
InputsCoordinates, graph, Tier 1 through Tier 4 labels
SummariesNode centrality, degree, betweenness, lineage and bucket edges
CaveatCompare within the same graph rule and label tier first

Browse and search

Browse is both catalog and search-results workspace.

SearchDataset key, title, PMID, DOI, organ, disease, technique
Dataset viewOne row per deployed dataset
Sample viewOne row per sample, best for organ and module readiness

Dataset pages

Dataset pages are terse records for one deployed dataset.

DefaultOverview, reference, organ inclusion, sample manifest
OntologyComposition and raw-to-curated flow when available
ModulesPer-sample links to spatial and DEG results

Curated tiers

Raw labels are preserved; curated tiers add a cross-study comparison layer.

Tier 1Super-broad groups
Tier 2Broad biology
Tier 3Default cross-study comparison layer
Tier 4Source-supported detail
A valid label is not automatically a strong pooled comparison label; reuse across datasets matters.
); if (!tierButtons.length || !tierTables.length) { return; } tierButtons.forEach((button) => { button.addEventListener('click', () => { const tier = button.getAttribute('data-help-tier') || ''; tierButtons.forEach((candidate) => { candidate.classList.toggle('is-active', candidate === button); }); tierTables.forEach((table) => { table.classList.toggle('is-active', table.getAttribute('data-help-tier-table') === tier); }); }); }); })();