Getting started

Sample datasets.

Public datasets each of the 33 templates was built around, so you can reproduce the documented result before trying your own data. 8 files cover 22 templates; the other 11 expect your own data.

Sample data is not bundled in the installer

Templates open with the loader path empty. Two small files can be downloaded directly below; the others come from their public sources. Save the file anywhere, then click the template’s first node and choose it — the paths shown are only the names the templates were developed against.

Download

Two small files you can download straight from this page — the desktop app’s Welcome screen links here. Load either with the Load H5AD node: add it to the canvas, pick the file in the Properties panel, then run.

PBMC 3k — raw counts

pbmc3k_raw.h5ad · 2,700 cells × 32,738 genes · integer UMI counts, stored sparse

Loader: Load H5AD. X holds raw counts, so start with Quality Control — do not skip ahead to Normalize.
Fits: Every template built around it (scRNA QC Overview, scRNA Standard Clustering, scRNA Full Pipeline with Visualization, scRNA Marker Discovery, scRNA DEG + Gene List Export, scRNA Doublet Screening, scRNA Harmony Integration, Trajectory with PAGA+DPT, Pathway Enrichment Analysis, GO & KEGG Pathway Visualization, Cell-Cell Interaction (CellPhoneDB)) and the PBMC tutorial: QC → Normalize → Log1p → HVG → PCA → Neighbors → Leiden → UMAP → DEG Analysis. The CellRank templates accept it too.
Source: 10x Genomics, “3k PBMCs from a healthy donor” (Cell Ranger 1.1.0 filtered matrix) — the dataset behind scanpy.datasets.pbmc3k(). When you publish, cite 10x Genomics for the data and Wolf, Angerer & Theis, Genome Biology 2018 for Scanpy.

PBMC 3k — raw counts (.h5ad, 7.7 MB)

Xenium demo

xenium_demo.h5ad · 500 synthetic cells × 300 genes · x_centroid / y_centroid / cell_type in obs

Loader: Load H5AD → Spatial Density QC works directly: the node reads coordinates from the x_centroid / y_centroid obs columns. Run Add Coordinates on those columns before Spatial Neighbors or any spatial plot, which need obsm["spatial"].
Fits: The Spatial Density QC step of the Xenium tutorial — a quick way to try the QC workspace, the grid filter and the spatial scatter without a real bundle.
Source: A small derived fixture made for Oreumics: synthetic cells, not a tissue section, and not for biological conclusions. For real data use Load Xenium on a 10x Xenium output bundle.

Xenium demo (.h5ad, 0.4 MB)

Other datasets

The remaining files are the public datasets each template was built around, grouped by modality. Download them from the linked source (10x Genomics, GEO, Bioconductor, the scVelo / squidpy projects) and open them with the loader named on each card.

Single-cell RNA-seq

PBMC 3k

sampledata/scrna/pbmc3k_raw.h5ad

Download Source

Peripheral blood mononuclear cells from a healthy donor on 10x Chromium — the standard scanpy tutorial dataset, with integer counts (Cell Ranger filtered output converted to H5AD). Contains the major PBMC populations, so QC → clustering → marker discovery gives textbook results.

Format: H5AD
Loader: Load H5AD
Source: 10x Genomics · scanpy.datasets.pbmc3k()

Templates built around it (11)

scRNA QC Overview scRNA Standard Clustering scRNA Full Pipeline with Visualization scRNA Marker Discovery scRNA DEG + Gene List Export scRNA Doublet Screening scRNA Harmony Integration Trajectory with PAGA+DPT Pathway Enrichment Analysis GO & KEGG Pathway Visualization Cell-Cell Interaction (CellPhoneDB)

Dentate gyrus (scVelo)

sampledata/velocity/dentategyrus_velocity.h5ad

Source

Mouse hippocampal dentate gyrus neurogenesis (Hochgerner et al., Nature Neuroscience 2018), packaged by scVelo with spliced, unspliced and ambiguous layers and an existing UMAP, so RNA Velocity runs directly.

Format: H5AD with spliced / unspliced layers
Loader: Load H5AD
Source: scVelo datasets · scvelo.datasets.dentategyrus()

Templates built around it (2)

RNA Velocity Stream (H5AD)RNA Velocity Latent Time

Velocyto .loom demo

sampledata/velocity/velocyto_demo.loom

Source

A small demo in the exact format the velocyto command-line tool writes — spliced / unspliced / ambiguous layers in one .loom. Use it to check the Load Velocyto Loom node before pointing it at your own velocyto output.

Format: velocyto .loom
Loader: Load Velocyto Loom
Source: velocyto.org (format) · demo file

Templates built around it (1)

RNA Velocity from Velocyto Loom

Spatial transcriptomics

Visium mouse brain

sampledata/spatial/visium_mouse_brain.h5ad

Source

A coronal section of adult mouse brain profiled with 10x Visium, pre-processed with spot coordinates in obsm["spatial"] and the histology image attached — the reference dataset of the squidpy tutorials, with clear anatomical regions for neighborhood and autocorrelation analyses.

Format: H5AD with obsm["spatial"] and the H&E image
Loader: Load H5AD
Source: 10x Genomics Visium · squidpy datasets

Templates built around it (3)

Spatial Neighborhood Analysis Spatial Autocorrelation Scan Spatial Full Analysis

Bulk RNA-seq

Airway (featureCounts)

sampledata/bulk/airway_featurecounts.txt

Source

The Bioconductor airway experiment (Himes et al., PLOS ONE 2014; GEO GSE52778): human airway smooth muscle cells, dexamethasone-treated versus untreated, in featureCounts output format with its header and annotation columns.

Format: featureCounts .txt
Loader: Load featureCounts
Source: NCBI GEO GSE52778 · Bioconductor airway

Templates built around it (2)

Bulk RNA-seq Differential Expression Bulk RNA-seq with GSEA

Airway sample sheet

sampledata/bulk/airway_metadata.csv

Source

The matching sample sheet with a condition column (control / treated) that Merge Cell Meta attaches, and Differential Expression tests on.

Format: CSV
Loader: Merge Cell Meta
Source: Derived from GSE52778 metadata

Templates built around it (2)

Bulk RNA-seq Differential Expression Bulk RNA-seq with GSEA

STAR ReadsPerGene

sampledata/bulk/star_ReadsPerGene.out.tab

Source

A synthetic single-sample STAR GeneCounts table (N_unmapped / N_multimapping / N_noFeature / N_ambiguous summary rows plus per-gene unstranded / forward / reverse counts) — a format demo for the loader, not a biological dataset.

Format: STAR --quantMode GeneCounts .tab
Loader: Load featureCounts (auto-detects STAR)
Source: Synthetic (format demo)

Templates built around it (1)

STAR ReadsPerGene — Single-sample Load Demo

ATAC-seq (chromatin accessibility)

Cell Ranger ATAC peak matrix

sampledata/atac/filtered_peak_bc_matrix

Source

A small filtered_peak_bc_matrix directory in the layout Cell Ranger ATAC writes. Any 10x scATAC-seq dataset — for example the 500-cell PBMC run — provides the same three files.

Format: MEX directory (matrix.mtx.gz + peaks.bed + barcodes.tsv.gz)
Loader: Load ATAC MEX
Source: 10x Genomics scATAC-seq datasets

Templates built around it (2)

ATAC Differential Accessibility ATAC Motif Discovery

Templates that expect your own data (11)

These ship with the graph fully wired and the loader waiting for your directory — the data they need is too large, too licence-bound, or too specific to a lab to point at a public file.

RNA Velocity from STARsolo: Velocity — A STARsolo run with --soloFeatures Gene Velocyto: the filtered/ directory holding spliced, unspliced and ambiguous .mtx.gz with barcodes and features.
MERFISH / MERSCOPE Spatial Analysis: Spatial — A MERSCOPE / MERFISH export — cell_by_gene.csv plus cell_metadata.csv with x / y centroids — loaded with Load CSV and Add Coordinates.
Methylation — Full Pipeline (IDAT to DMP / DMR): Methylation — A folder of Illumina IDAT pairs (Red + Grn per sample) from a GEO series or your core facility, or a vendor beta-value matrix (CSV / TSV), plus a sample sheet keyed by sample ID for Merge Cell Meta.
Methylation — Full Pipeline (Beta Matrix to DMP / DMR, EPIC v2): Methylation — A folder of Illumina IDAT pairs (Red + Grn per sample) from a GEO series or your core facility, or a vendor beta-value matrix (CSV / TSV), plus a sample sheet keyed by sample ID for Merge Cell Meta.
Methylation — Visual EDA (no phenotype required): Methylation — A folder of Illumina IDAT pairs (Red + Grn per sample) from a GEO series or your core facility, or a vendor beta-value matrix (CSV / TSV), plus a sample sheet keyed by sample ID for Merge Cell Meta.
Methylation — Pathway Discovery (Volcano + GOmeth): Methylation — A folder of Illumina IDAT pairs (Red + Grn per sample) from a GEO series or your core facility, or a vendor beta-value matrix (CSV / TSV), plus a sample sheet keyed by sample ID for Merge Cell Meta.
LLM Auto Annotation: Annotation — Any scRNA-seq H5AD for the query; Reference Label Transfer additionally wants a reference atlas H5AD with cell-type labels in obs.
Reference Label Transfer: Annotation — Any scRNA-seq H5AD for the query; Reference Label Transfer additionally wants a reference atlas H5AD with cell-type labels in obs.
CellRank — CytoTRACE Fate Analysis: Single-cell — Any scRNA-seq H5AD with raw counts — the PBMC 3k file above works for the CellRank templates too.
CellRank — Pseudotime Fate Analysis: Single-cell — Any scRNA-seq H5AD with raw counts — the PBMC 3k file above works for the CellRank templates too.
CellRank — Full Pipeline (CytoTRACE + All Viz): Single-cell — Any scRNA-seq H5AD with raw counts — the PBMC 3k file above works for the CellRank templates too.

All datasets are public

They come from 10x Genomics, NCBI GEO, Bioconductor and the scVelo / squidpy projects under their respective open-access terms. Cite the original publication when you use one in your work; the Methods paragraph the app generates cites the software, not the data.