1.33.1
What is SciFlow? Sign In
Read-only view of a live SciFlow workflow — this is the actual pipeline that produced these results, not a mockup. Single-Cell RNA-seq — PBMC Stim vs Control (SciFlow). Open any stage to inspect its parameters, command, versions, inputs and outputs.
Loading workflow...
Name
Modified
Size

No pipeline outputs

Run stages to generate files here.

No reference papers

This demo has no reference papers attached.

Timeline
Set start date
–
Set end date
Owner Hidden for review Tags
No tags
Created Apr 30, 2026 at 01:15
Take quantified single-cell counts through the full analysis arc — normalization, integration, clustering, cell-type labels, marker genes, and a real differential expression test between conditions. Drop in 10x count matrices (the demo combines pbmc3k and pbmc6k for batch integration, then layers on the Kang 2018 IFN-β stimulated vs control PBMCs as the conditional contrast), get back a labeled UMAP, per-cluster markers, and a pseudobulk DE table you can actually publish. Built for immunology and translational labs running stim/ctrl, drug/vehicle, or disease/healthy designs, and for core facilities wanting a defensible default workflow they do not have to reinvent for every project. Counts go in for DE; never TPM, never FPKM. Normalization is SCTransform v2 — variance-stabilized, with mitochondrial percent regressed out. Integration is Harmony on the SCT-corrected embedding, clustering is Leiden on the integrated graph, and cell-type calls come from Azimuth against the PBMC reference so labels are reproducible across runs. Marker detection uses presto for fast Wilcoxon ranks at the cell level, and the cross-condition test is pseudobulk DESeq2 — aggregating to sample-level counts before the model, which is the only DE approach with calibrated false-positive rates on this data type.