Cell Atlas

Every cell is one point. Drag to pan, scroll to zoom, shift-drag to select a region; click any group in the legend to isolate it. Use Colour cells by to switch between cell type, cluster, library, quality-control metric and the expression of a single gene.

UMAP — this re-analysis

Drawn from the same exported coordinates and labels the interactive atlas above uses, so the figure and the atlas cannot disagree. 36,550 cells.

Coloured by cell type (8)
UMAP of the re-analysed NVECT_gastruloid, coloured by annotated cell type, with a legend naming each type

The legend in the panel names every type and its cell count.

Coloured by Leiden cluster (17)
UMAP of the re-analysed NVECT_gastruloid, coloured by Leiden cluster, each labelled with its number

Each cluster's number is placed at its own centroid, so the grouping can be read without relying on colour. These are the clusters the cell-type annotation was built on.

Cell types in this dataset

Click a cell type to isolate it in the atlas above. The bar beside each percentage is drawn to scale: a cell type at 2.6 % fills 2.6 % of the track, so the rows can be ranked by eye without reading the numbers.

Cell type Cells Percent Markers
Ectoderm 25,986 marker genes
Secretory Progenitor 2,713 marker genes
Endoderm 2,055 marker genes
Unannotated 1,542 marker genes
Mesoderm 1,489 marker genes
Cnidocytes 1,471 marker genes
Neuronal 730 marker genes
Gland Cells 564 marker genes
8 cell types 36,550

Quality control for this dataset

These are the figures actually applied to the cells shown above, not the values requested in a configuration file. Every dataset is filtered independently; see Single-cell Data for the full table across all datasets.

Source BioProject PRJNA1327231
SRA study SRP619054
GEO series GSE307733
Library type 10x Genomics Chromium (high confidence)
Libraries integrated 7
Cells before filtering 36,568
Cells after cell filtering 36,568
Doublets removed 18 (0.05%)
Cells after filtering 36,550
Retained 100.0%
Filtering strategy Per-library MAD outlier detection (adaptive)
Thresholds applied adaptive: per-library MAD (5.0 MADs) on UMI, genes and MT%, floor 200 genes, floor 500 UMI
Mitochondrial genes found 0 — too few for a reliable MT% filter, so none was applied
Doublet method Scrublet (per library)
Integration method Harmony on sample (7 libraries)
Cell-type annotation Inherited from the source publication
Cell-type labels from GSE307733_g.cell_to_cts.csv.gz (100% of cells matched)
Pipeline version cnidosite-sc-1.0.0
Processed 2026-09-20T03:34:23+00:00
Clusters / cell types 17 / 8
TOP