When to use this
You have gated your samples in FlowJo and exported a table: one row per sample (FCS file), columns such as CD4+/Foxp3+ | Freq. of Parent, Count or Median : FITC-A. You want to compare conditions (for example Control vs EAE), correlate populations with a clinical score, or relate flow results to single-cell RNA from the same animals.
Step by step
- Export the table from FlowJo. In the Table Editor, add the statistics you need (frequencies, counts, MFIs) and export as CSV or TXT. FCS files are not needed.
- Prepare a sample sheet. A small CSV with
sample(the FCS name),conditionand, optionally,animal_id,batch,sexor a clinical score column. - Open FACS–scRNA Integration and load both files. Use FlowJo table export and Sample sheet, then Validate inputs. The tool lists how each column was classified (frequency, count, MFI) and how samples were matched to the sheet.
- Choose the reference condition and run. Run analysis fits each statistic with a model suited to it and adjusts p values (Benjamini–Hochberg) within each statistic type.
- Read the results by tab. Frequencies, MFIs, Ratios, Counts, Correlation and Sample overview (PCA for batch effects and outliers). Every plot exports as SVG/PNG and every table as CSV, with a methods paragraph.
Why the model matters
- Frequencies are proportions. With event counts they are modelled with a quasi-binomial GLM (weighted by the parent count, with overdispersion between animals); without counts, on the empirical-logit scale. Effects are odds ratios plus percentage-point differences.
- MFIs are compared as fold changes of geometric means on the log2 scale, or on the asinh scale if values can be negative. Compare MFIs only within one channel and panel.
- Animals are the replicates. Paired designs (same animal, two tissues) and covariates such as batch are handled explicitly; a covariate confounded with condition is dropped and reported.
- Correlations with clinical scores use tie-aware permutation p values and a condition-adjusted partial correlation, which flags associations driven only by the group difference.
Linking flow to single-cell RNA
Add a pseudobulk matrix, a cell-level table or a DE table from scRNA-seq of the same animals, and map gates to annotated cell types (for example Foxp3+ = Treg). You get flow vs scRNA proportions per animal and protein (MFI) vs mRNA for the same marker.
Open FACS–scRNA Integration How it is validated
Questions
Do I need FCS files?
No. Gating, compensation and population definitions stay in FlowJo; the tool analyses the exported statistics.
Is my data uploaded?
No. Files are read and analysed in your browser.
Which test is used for percentages?
A quasi-binomial GLM when event counts are exported, otherwise an empirical-logit linear model — not a t-test on raw percentages.
How are results checked?
The models are compared with statsmodels, SciPy and NumPy at stated tolerances; see the validation page.
Found something wrong or unclear? Write to contact@solveseverything.com. See also: how results are validated.