stereo-seq-analysis-workflow

name: stereo-seq-analysis-workflow description: Use when the user asks for a complete Stereo-seq/STOmics analysis workflow, end-to-end pipeline coverage, which local Stereo-seq skills to run next, raw-to-paper planning, project intake, analysis roadmap, skill routing, milestone checklist, or whether the current skills cover a full Stereo-seq analysis from raw GEM/GEF/SAW/count matrix through QC, image registration, cellbin, annotation, domains, programs, communication, statistics, 3D, visualization, and publication figures.

Stereo-seq Analysis Workflow

This is the routing skill for end-to-end Stereo-seq projects. It does not replace the specialized skills; it decides which ones to use, in what order, and what evidence or metadata must exist before each stage.

Use This For

• Planning a complete Stereo-seq analysis from raw GEM/GEF/SAW/count matrix to figures.
• Auditing whether the current local Stereo-seq skills cover a requested workflow.
• Creating a project manifest, stage checklist, or handoff plan across multiple skills.
• Deciding next steps when inputs are mixed: raw files, h5ad/RDS, images, cellbin masks, sc/snRNA references, replicates, serial sections, or 3D goals.

For paper-story design from a specific biological question, also use stereo-seq-publication-story. For implementation of each analysis stage, load the downstream skill named by this workflow.

Default Requirements

• Start by identifying available inputs, missing inputs, and the intended claim/output.
• Route to existing specialized skills instead of re-implementing their methods.
• Preserve article-derived provenance expectations: each downstream skill should use its own references/source_code.md and bundled scripts before custom code.
• Flag unsupported claims early:
  • no biological replicates for condition inference;
  • no image/mask/landmarks for image registration or cellbin;
  • no sc/snRNA reference for reference-based annotation;
  • no serial sections or z/section metadata for 3D reconstruction;
  • no cleaned count matrix for downstream biological analysis.
• If the user asks whether coverage is complete, report both covered stages and deliberately excluded stages.

Stage Router

1. Project intake and metadata
   • Multi-sample, batches, conditions, donors, replicates, time points, or serial sections: stereo-seq-project-orchestration.
   • Paper-like biological plan or matching previous articles: stereo-seq-publication-story.
2. Raw-to-matrix and QC
   • GEM/GEF/SAW/StereoPy/h5ad/RDS preprocessing, binning, count matrix export, QC maps: stereo-seq-quality-control-preprocessing.
3. Image registration and cellbin
   • ssDNA/DAPI/HE/histology/expression heatmap registration, transform QC: stereo-seq-image-registration.
   • cell/nuclei masks, segmentation, bin-to-cell aggregation, cellbin QC: stereo-seq-cellbin-segmentation.
4. Core biological mapping
   • cell type/state annotation, deconvolution, label transfer: stereo-seq-cell-type-mapping.
   • domain/layer/region discovery and annotation: stereo-seq-spatial-domain-discovery.
   • mapped cell-type spatial interpretation: stereo-seq-spatial-cell-type-interpretation.
5. Programs and mechanisms
   • SVGs, DEG, pathway scores, GO/GSEA, modules, NMF: stereo-seq-spatial-programs.
   • TF regulons, pySCENIC/SCENIC, SpaGRN, AUCell/RSS: stereo-seq-spatial-grn-regulon.
   • CCI/LR/CellChat/NicheNet/STcomm/SPIDER/spatial communication: stereo-seq-spatial-communication or stereo-seq-cell-cell-interaction.
   • pseudotime, regeneration, velocity, stage transitions: stereo-seq-developmental-trajectory.
6. Statistics
   • condition/disease/treatment/time comparisons with biological replicates, pseudobulk, model design, spatial organization metrics: stereo-seq-statistical-design.
7. 3D and visualization
   • multi-slice registration, SABench-informed alignment, atlas/shape matching, 3D coordinate reconstruction: stereo-seq-3d-reconstruction.
   • interactive plotly/htmlwidgets/RColorBrewer 3D visualization or SPACEL/Scube-style snapshots: stereo-seq-3d-visualization.
8. Publication output
   • manuscript-quality maps, dot plots, heatmaps, marker panels, palettes, figure polishing: stereo-seq-publication-plotting.

Reusable Script

• scripts/build_stereo_workflow_manifest.py: create a stage-by-stage manifest from project features such as input type, image availability, scRNA reference, replicates, serial sections, 3D goals, and publication/story goals.

Output Expectations

• A concise stage order with the exact local skill for each stage.
• Inputs required before each stage and blockers if missing.
• Recommended first command/script when the downstream skill has a relevant bundled template.
• Coverage statement: what is covered, what is excluded, and what requires external tools/environments.
• Provenance reminder: downstream analyses should report paper DOI, repository/source file, and dataset-specific edits.