Study workflow

The product follows a fixed left-to-right flow. You can revisit earlier sections at any time, but downstream results may become stale if inputs change.

Data → Explore → Analyze → Interpret → Figures → Runs

Data — prepare inputs

Upload files, validate ingestion, map metadata, define experimental design, and save contrasts. Nothing in Analyze runs meaningfully until at least one dataset is ready.

Readiness signal: green dot on Datasets when a processed .h5ad exists.

Explore — first look

Inspect QC metrics, sample relationships, embeddings, and spatial layouts before committing to formal analysis. Explore is read-only with respect to the pipeline; it displays outputs from completed runs.

Analyze — compute

Run structured workflows: QC, normalization, clustering, differential expression, enrichment, spatial domains, biomarker discovery. Parameters are saved with every job. Failed runs show actionable errors on the Analyze and Runs pages.

Interpret — biological meaning

Connect numeric outputs to biology: cluster annotations, pathway enrichment, importable gene sets, and (planned) AI-assisted summaries. Interpret aggregates results from multiple completed runs.

Figures — publish

Drag plots from completed runs onto a canvas, arrange panels, and export a letter-size PDF. Layout is saved per study in your browser until server-side figure persistence ships.

Runs — provenance

Review active and completed jobs, expand rows for parameters and errors, manage snapshots, and address stale outputs. Deleting runs is blocked when a snapshot references them or while a job is still running.

Typical single-cell path

Step	Section	Action
1	Data	Upload .h5ad or convert 10x/CSV
2	Data	Save metadata and contrasts
3	Analyze	QC → Normalization → Clustering
4	Explore	Review UMAP and QC charts
5	Analyze	DE → Enrichment
6	Interpret	Review enrichment and gene sets
7	Figures	Build multi-panel figure
8	Runs	Create snapshot

See section-specific guides: Data, Explore, Analyze, Interpret, Figures and Runs.