MIFlowCyt Reporting: What a Minimum Information Flow Cytometry Submission Requires

A reviewer sends back your manuscript with one comment on the flow data: list the antibody clones, catalog numbers, and lot where relevant. You go looking, and the panel lives in a lab-notebook photo and a tube label that read “CD4 PE (the good one)”. Now you are reconstructing an experiment from six months ago instead of answering the actual scientific question.

MIFlowCyt — the Minimum Information about a Flow Cytometry Experiment — exists to keep that reconstruction from happening. It is an ISAC reporting standard that defines the minimum a reader needs to interpret your results and reproduce them in another lab. This post walks through what a MIFlowCyt submission actually requires, where the standard is enforced, and how to assemble it without it becoming its own project.

What MIFlowCyt requires

The minimum information about a flow cytometry experiment is organized into four reporting domains. Each one maps to a part of the work you already did — the standard is mostly about writing it down in a structured place, not generating new data.

• Experiment overview. The purpose, the keywords, who ran it, and the experiment-level variables (the comparison you are actually making — treated vs. control, timepoint, donor).
• Samples and reagents. Specimen source and handling, plus every fluorescence reagent with its manufacturer, clone, and catalog number. This is the section reviewers ask about most, because “anti-CD4 PE” is not reproducible — clone RPA-T4 from one vendor behaves differently from a different clone.
• Instrumentation. Cytometer make and model, laser lines, detector configuration, and the detector voltages or gains used for the run. A panel optimized at one set of PMT voltages will not reproduce on another.
• Data analysis. The list-mode file locations, the compensation or unmixing approach, the display transformations, and the full gating strategy — including how each gate was placed.

The first three domains are description; the fourth is where most of the reproducibility risk lives, because gating is the analytical act that turns events into numbers.

The reagent table is where submissions fail

If you do one thing before acquisition, build the reagent table at the bench. For each marker, record the target, the fluorochrome, the clone, the vendor, and the catalog number. Add the titrated concentration if you have it — reviewers in immunology increasingly expect to see that you found the working dilution rather than using the datasheet default. If you are not sure why titration belongs in the methods, the reasoning is the same one behind finding the right antibody concentration by stain index: the concentration changes the signal, so it changes the result.

Report the analysis, not just the result

The data-analysis domain is where MIFlowCyt overlaps with everything that makes flow results hard to compare between groups. Three things need to be explicit:

• Compensation or unmixing. State how the spillover correction was derived — single-color controls on beads or cells, and whether the matrix was reviewed. If you compute the matrix yourself, the spillover matrix calculator will export it in a form you can attach to the supplement.
• Transformations. Name the display transform (biexponential, logicle, arcsinh) and its parameters. A gate placed on a logicle axis is not in the same position on a log axis.
• Gating strategy. Give the full hierarchy, parent to child, with the parameters and the rationale for each gate — ideally with the control (such as an FMO) that set the boundary. This is the part that lets a second analyst reproduce your population, which is the whole point of the standard.

The reason this matters in practice is the same reason gating varies between operators: two people given the same FCS file and no gating record will draw different gates and report different percentages. The gating strategy is the bridge between “here is my number” and “here is how you would get the same number.”

Where MIFlowCyt is actually enforced

This is not an optional best practice you can skip if no one is watching. Three places check for it:

• Public repositories. FlowRepository requires datasets to be annotated to the MIFlowCyt standard before they are released, so depositing data forces you to assemble the report anyway.
• Journals. Several journals, including titles in the Nature and PLOS families, reference MIFlowCyt in their flow cytometry reporting expectations.
• Funder data-sharing policy. Under the NIH Data Management and Sharing Policy, a structured experiment record is the practical way to make flow data shareable and interpretable after the fact.

Assemble it from what the files already carry

The work is smaller than it looks because FCS headers already contain a large share of the required fields. The TEXT segment of an FCS 3.1 file carries the parameter names, detector configuration, and acquisition keywords; the gating hierarchy carries the analysis structure; the compensation matrix is already an object in your analysis. A tool that reads those sources can pre-fill the instrumentation and data-analysis domains and leave you to enter only the fields that live in your head — the specimen handling and the experiment rationale.

The fields that are never in the file are the ones worth writing down at the bench: specimen source, reagent catalog numbers, and the one-sentence purpose of the experiment. Capture those three at acquisition time and the rest is assembly. If you are already exporting numbers for a figure, the same pass is the natural moment to capture the methods — see exporting flow cytometry statistics for a manuscript for how the statistics and the reporting metadata come out of the same step.

A short pre-submission check

• Every fluorochrome has a clone and a catalog number, not just a marker name.
• Detector voltages or gains for the run are recorded.
• The compensation or unmixing method is named, with the control type.
• The display transform and its parameters are stated.
• The gating hierarchy is documented parent-to-child, with the control that set each boundary.

If those five hold, your data-analysis and reagent domains are covered, and the remaining MIFlowCyt fields are description you can write in an afternoon. The standard rewards the lab that captured the details once over the lab that tries to reconstruct them under a revision deadline.