| name | sc-doublet-detection |
| description | Annotate putative doublets in single-cell RNA-seq data using Scrublet, DoubletDetection, DoubletFinder, scDblFinder, or scds. The wrapper preserves the current AnnData matrix semantics, standardizes output columns in `obs`, and exports a reusable figure/table gallery. |
| version | 0.6.0 |
| author | OmicsClaw |
| license | MIT |
| tags | ["singlecell","doublet","scrublet","doubletdetection","doubletfinder","scdblfinder","scds","qc"] |
| metadata | {"omicsclaw":{"domain":"singlecell","allowed_extra_flags":["--method","--expected-doublet-rate","--threshold","--batch-key","--doubletdetection-n-iters","--doubletdetection-standard-scaling","--no-doubletdetection-standard-scaling","--scds-mode","--r-enhanced"],"param_hints":{"scrublet":"[Truncated]","doubletdetection":"[Truncated]","doubletfinder":"[Truncated]","scdblfinder":"[Truncated]","scds":"[Truncated]"},"legacy_aliases":["sc-doublet"],"saves_h5ad":true,"requires_preprocessed":false}} |
Single-Cell Doublet Detection
Why This Exists
- Without it: artificial multiplets can masquerade as transitional or mixed cell states.
- With it: cells receive standardized doublet scores and labels before final clustering, annotation, or DE interpretation.
- Why OmicsClaw: one wrapper harmonizes multiple common backends into the same output columns and gallery layout.
Scope Boundary
Implemented method families:
scrublet
doubletdetection
doubletfinder
scdblfinder
scds (hybrid, cxds, bcds)
This skill annotates doublets in obs. It does not silently remove cells.
Input Expectations
- Preferred state: raw count-like input in
layers["counts"], aligned adata.raw, or count-like adata.X
- Typical stage: after QC review and before final clustering / annotation / DE interpretation
- Important nuance: if the object is already normalized, the wrapper still uses raw counts for calling and preserves the current
adata.X semantics
Public Parameters
Shared controls:
--method
--expected-doublet-rate
Method-specific controls:
scrublet
doubletdetection
--doubletdetection-n-iters
--doubletdetection-standard-scaling
scds
Output Contract
Successful runs write:
processed.h5ad
report.md
result.json
figures/doublet_score_distribution.png
figures/doublet_call_summary.png
figures/embedding_doublet_calls.png when an embedding exists or a preview embedding can be computed