How would you compare longitudinal cytometry data collected on different instruments?

Why

Two cytometers with different lasers, filters and detector voltages give different numbers for the same cell. Over months, lasers age and reagents fade. The Cossarizza guidelines show that samples from one subject correlated best when run on one day and one machine, and less well across days or cytometers.

Perfetto and colleagues describe a program of optimization, calibration and standardization with beads. It is repeated in full when an instrument is installed or its optical path changes, with daily checks between. The point is to adjust voltages so that a reference bead lands on the same target intensity in every detector of every instrument.

Reagent lots, staining protocol, sample handling and analysis templates matter as much as hardware. Frequencies of well-separated populations are more robust than intensities.

How to work it through

1. Define which outputs must be comparable, such as population frequencies or marker intensities.
2. Set target bead values for every detector and adjust each instrument to hit them, then verify with a shared control sample.
3. Freeze the panel, reagent lots where possible, staining protocol and analysis template.
4. Run a cryopreserved reference sample with every batch and chart its key readouts.
5. Inspect the data for batch effects and apply normalization only when the reference sample shows it is needed.
6. Report which comparisons are within-instrument and which are cross-instrument.

Where the answer stops holding

Beads differ from cells in spectrum and autofluorescence, so matched bead values do not guarantee matched cell values in every channel. Tandem dye lots and sample age can drift in ways no instrument setting corrects. Computational batch correction can remove real biology if batches are confounded with groups, so the design should mix groups within each batch.