METHODS

How every run works.

The workflow shape, what every run delivers by assay, the exact tools and citations at each step, and the paste-ready Methods paragraph it ends with. Transparency is the trust.

THE SAME WORKFLOW, EVERY TIME

A workflow you can read end to end.

This is the sequencing path; cytometry and bulk share the same skeleton, with assay-specific steps swapped in.

01Ingest + QC Ingest + QC diagnosticsWe profile the dataset (library size, mitochondrial fraction, gene detection) before anything is filtered, so you see the data before deciding.
02Clean + filter Clean + adaptive filterDoublets removed and outliers filtered on cutoffs derived from your data's own distribution, recorded in the run log.
03Normalize Normalize + select featuresNormalization, log1p and highly-variable-gene selection: putting the signal on the scale the downstream statistics assume.
04Integrate (opt) Integrate · optionalscVI batch integration when you provide a batch key; skipped automatically when there's a single batch, and the report says so.
05Cluster Leiden clusteringLeiden community detection groups cells using the neighbourhood graph and the workflow's configured clustering settings.
06Annotate CellTypist annotationA CellTypist reference model auto-selected from your tissue, with a per-cell confidence score and the marker evidence kept alongside.
07Test + pathways Differential testing + pathwaysDifferential expression on the comparison you choose (pseudobulk DESeq2 or Wilcoxon) plus pathway enrichment (GSEA, ORA).
08Report Report + Methods paragraphA self-contained HTML report, downloadable figures, Excel tables, and the paste-ready Methods paragraph.

A cytometry run adds two points where it pauses for your approval: the panel that defines cell identity, and the auto-suggested gating hierarchy. If a step can't run on your data (too few cells for trajectory, no replicates for DE, missing controls for unmixing), the report says which step and why. It never silently skips.

ANALYSES BY ASSAY

What you actually get, by assay.

No lite version, no paywalled analyses. Available analyses depend on the assay, configuration and data. Switch between them:

  • QC + diagnostics, adaptive thresholds, normalization, HVG selection
  • Dimensionality reduction, Leiden clustering, tissue-matched CellTypist annotation with a per-cell confidence score
  • Differential expression on the comparison you choose (pseudobulk DESeq2 with replicates, Wilcoxon otherwise) and pathway enrichment (GSEA, ORA)
  • Trajectory and cell–cell communication when your run enables them and the data supports them
  • scVI integration when a batch key is provided
EVERY RUN ALSO SHIPS

A receipt, not just a picture.

Clustering quality metrics, a self-contained HTML report, downloadable figures, an Excel workbook with a plain-English About sheet, and the paste-ready Methods paragraph.

  • Panel-logic reconstruction from $PnN / $PnS, single-stain and control validation
  • Compensation review (flow) or unmixing review (spectral). Upload multiplexed CyTOF data as separate, already-debarcoded sample FCS files.
  • Uniform-across-samples QC gates, arcsinh transform
  • FlowSOM clustering + UMAP, assisted editable polygon gating, MEM-score labels
  • Replicate-aware differential abundance + differential state across conditions
TWO APPROVAL PAUSES

It stops and waits for you, twice.

Once to approve which markers define cell identity before clustering touches them, once to review and edit the suggested gating hierarchy as real polygons. The hierarchy is written as standards-compliant GatingML 2.0.

  • Library-size + gene-detection QC
  • Sample PCA with a variance-stabilizing transform, or a log-CPM fallback
  • Differential expression (negative-binomial GLM via PyDESeq2), with tests for each requested comparison
  • Gene-set enrichment (GSEA, ORA) when reference collections are available
  • Volcano figures (PNG + interactive HTML); Kaplan–Meier curves when survival metadata is available
EVERY RUN ALSO SHIPS

The numbers behind every plot.

An Excel workbook, per-comparison differential-expression tables, available enrichment tables, a self-contained HTML report, and Methods describing the design, tests and reference collections used.

METHODS FORMAT

See how your analysis is described.

Your completed run provides Methods text for its recorded settings and completed steps. These illustrative previews show the format; use your run's Methods when preparing a manuscript.

ILLUSTRATIVE PREVIEW · NOT A RUN RESULT

Example · scRNA-seq
Methods · illustrative preview