Guides / Flow cytometry basics
What is compensation, and why does it matter?
When you label cells with several fluorescent dyes, each dye also leaks a little light into detectors meant for the other dyes. This is called spillover. Compensation is the arithmetic that removes it.
The problem
Consider two dyes. A cell stained with only the first dye still produces a small signal in the second dye's detector. Without correction, that cell appears "double positive" and the gate counts the wrong events.
How it is corrected
Each dye gets measured on its own, in a single-stain control, to learn how much it leaks into every other detector. Those numbers form a spillover matrix. Compensation applies the inverse of that matrix to every event, so the leaked signal is subtracted.
What you need
- A single-stain control for every dye in the panel, at least as bright as your real samples, and treated the same way. Within each control, the positive and negative events need the same autofluorescence, and the control should use the same dye lot as the samples.
- A negative (unstained) control, to set the baseline.
- The spillover matrix. Many acquisition programs store one inside the FCS file, and OpenFlo applies it when it loads the file.
In OpenFlo
Tools > Compensation & spectral > Compensation... shows and edits the matrix, imports one from a workspace or a CSV file, and can estimate one from single-stain controls. Compensation QC... shows the spillover as a heatmap and lists the strongest pairs, so you can spot a suspicious entry.
Common mistakes
- Compensating on a different day, voltage or instrument setup from the experiment.
- Using a control that is dimmer than the sample, or a different dye lot.
- Forgetting that compensation changes what log-scale plots look like, which is why fluorescence is shown on a logicle scale, with a small linear region around zero so that compensated negatives do not pile up against the axis.
Compensation is a correction and cannot repair poor controls: a matrix computed from poor controls will be inaccurate.