Analyze
Analyze runs computational workflows. Tasks are grouped by research intent rather than by software package name.
Overview
Shows task readiness (which steps have completed runs), recommended next actions, and links to stale outputs.
When uploaded AnnData files contain Bioconductor-style layers, relevant steps expose a matrix-source control. Counts-oriented workflows default to detected counts, while expression-oriented workflows default to logcounts or normalized matrices when available. The selected source is recorded in run parameters.
Find Structure
| Step | Purpose |
|---|---|
| QC | Filter cells by counts, genes, and mitochondrial percentage; score doublets per sample (optionally remove them); record assay type (cells, nuclei, fixed cells); score stress and cell-death signatures |
| Ambient RNA Correction | Optional, after QC: estimate and remove ambient RNA (SoupX-style with a raw droplet matrix, or decontX-style); corrected counts feed later steps |
| Normalization | Log-normalize, scale, select highly variable genes |
| Clustering | Neighbors graph, Leiden clustering, UMAP (on the integrated space when Batch Integration has run) |
| Cell Type Annotation | Label clusters from an organism-aware marker panel (blood, adrenal, pituitary, thyroid, testis), your own markers, or a labelled reference dataset in the study (SingleR-style, per cell or per cluster) |
| Trajectory & Cell Cycle | Score cell-cycle programs, compute graph-based pseudotime, review UMAP overlays and assumption warnings, and rank genes associated with pseudotime |
| RNA Velocity | Estimate velocity from spliced/unspliced counts attached on the Data page (stochastic or steady-state model) and draw it on the UMAP; reports the spliced fraction per cell |
| Spatial domains | Cluster spatial spots into tissue domains, rank spatially variable genes, summarize neighborhood enrichment, compare domains by condition, and nominate co-localization niche candidates |
Steps must generally run in order; each saves a checkpoint .h5ad for the next step. See Single-cell analysis for methods and parameters.
Compare Groups
| Step | Purpose |
|---|---|
| Differential expression | Wilcoxon rank-sum on contrasts or cluster markers, or a replicate-aware pseudobulk contrast (choose the method on the tile) |
| Cell Composition | Test whether cell-type proportions differ between conditions across biological replicates (propeller-style: transformed proportions, moderated t / F tests, BH FDR); descriptive when there are no replicates |
| Preparation Method Comparison | Benchmark sample-preparation methods (enzymatic, fixed cells, nuclei): recovery, quality and stress metrics with sample-level tests, composition, marker preservation, and DE between methods |
| Pathway enrichment | ORA, ranked GSEA, ssGSEA/GSVA gene-set scoring against the study organism's GO and Reactome sets, run-scoped custom gene sets, and result gene-set comparison |
For contrast-based DE, pseudobulk DE, and composition, save contrasts on the Data page first (composition can also use a condition column directly). Cluster marker mode compares each cluster to all others.
Bulk RNA-seq
| Step | Purpose |
|---|---|
| Bulk RNA-seq | Sample-level count filtering, DESeq2-style/TMM normalization metadata, VST, differential expression, sample QC, PCA/sample-distance diagnostics, and TSV exports |
| Reference Deconvolution | Estimate cell-type fractions from sample-level or spot-level expression |
Bulk results open in a dedicated view with ranked DE genes, sample QC, PCA/MDS-style sample mapping, sample-distance heatmap data, and mean-variance diagnostics.
Build Predictors
| Step | Purpose |
|---|---|
| Biomarker discovery | Feature selection (mRMR or random forest), optional WGCNA, classifier CV |
| WGCNA Modules | Standalone co-expression modules with soft-threshold diagnostics, hub genes, eigengene/trait review, lightweight hub-edge graph, and TSV exports |
| Coverage DEG | Pairwise class comparisons, cross-contrast DEG recurrence ranking, inline disease evidence, and TSV exports |
Requires a class label column in obs (default condition) with at least two classes.
Tumor / CNV
| Step | Purpose |
|---|---|
| Single-Cell CNV | Expression-derived chromosome-scale CNV signal, malignant-cell evidence, subclones, and state calls |
Integrate / Scale
| Step | Purpose |
|---|---|
| Batch Integration | Harmony (default) or fastMNN batch correction (PCA centring kept as a baseline), integrated clustering, before/after previews, batch mixing (iLISI) and cell-type preservation (cLISI, silhouette) metrics, and confounding warnings |
| Sketch & Scale | Create representative sketches, projection artifacts, downsampled previews, dedicated sketch results, and large-dataset guardrails |
Result pages
Completed steps link to dedicated result views:
- DE Results — volcano plot and sortable gene table
- Pseudobulk Results — per-cell-type/group selector, volcano plot, sample aggregate table, skipped-group warnings, and TSV exports
- Composition Results — per-sample proportion chart, mean proportions per condition, ratio, test statistics and FDR, and TSV exports
- Preparation Method Comparison — recovery, quality-metric table with FDR, per-sample dot strips, composition, marker preservation, and DE between methods with technical genes flagged; the metrics table is also available in the Report Builder
- Integration Results — before/after previews, iLISI / cLISI and silhouette metrics, Harmony iterations or the fastMNN merge table, batch composition, and correction warnings
- Annotation Results — reference and organism, compatibility, per-cluster score heatmap (or label share per cluster for per-cell labelling), confidence/delta review, supporting markers, and each label's marker source
- Trajectory Results — pseudotime/cell-cycle UMAP overlays, assumption warnings, cluster ordering, pseudotime-associated genes, and TSV exports
- Velocity Results — arrows on the UMAP, velocity genes, coherence, gene fits (γ, R²), spliced fraction by group, and TSV export
- Sketch Results — representative-cell preview, projected cluster counts, scale summary, and artifact status
- Bulk Results — sample QC, PCA/sample-distance diagnostics, mean-variance table, and DE gene table
- Biomarker Results — ranked genes and classifier metrics
- Spatial Results — spot/domain viewer, SVG ranked list and overlays, neighborhood enrichment, condition-domain comparison, niche candidates, and TSV exports
- WGCNA Module Explorer — module summaries, hub genes, eigengene/trait tables, selected-module hub-edge graph, and TSV exports
- Coverage DEG Results — recurrent class-comparison genes, pairwise DE details, inline disease evidence, and TSV exports
- Enrichment Results — GO, Reactome, KEGG, GSEA, and score-summary tabs with top-term bars, per-term gene exports, and pairwise set operations across available result gene lists
DE, pseudobulk, composition, preparation comparison, velocity, bulk, enrichment, biomarker, and spatial result views include TSV export buttons for supplement-ready tables.
Errors and reruns
Failed runs show the error under the step's Run button, at the top of Analyze, and in expandable rows on History. Fix parameters or input data, then re-run. Re-running QC or normalization invalidates downstream checkpoints.