Uncertainty Quantification in Flow Cytometry Using a Cell Sorter

• Published in: Cytometry. Part A Vol. 107; no. 4; pp. 248 - 262
• Main Authors: Krishnaswamy‐Usha, Amudhan, Cooksey, Gregory A., Patrone, Paul N.
• Format: Journal Article
• Language: English
• Published: Hoboken, USA John Wiley & Sons, Inc 01.04.2025; Wiley Subscription Services, Inc
• Subjects: Affine transformations; beads; Cell Separation - instrumentation; Cell Separation - methods; FACS; Flow cytometry; Flow Cytometry - instrumentation; Flow Cytometry - methods; Flow rates; gating; Humans; sorting; Uncertainty; uncertainty quantification; Variation; gating; beads; uncertainty quantification; FACS; sorting
• DOI: 10.1002/cyto.a.24925
• ISSN: 1552-4922; 1552-4930

ABSTRACT In cytometry, it is difficult to disentangle the contributions of population variance and instrument noise toward total measured variation. Fundamentally, this is due to the fact that one cannot measure the same particle multiple times. We propose a simple experiment that uses a cell sorter to distinguish instrument‐specific variation. For a population of beads whose intensities are distributed around a single peak, the sorter is used to collect beads whose measured intensities lie below some threshold. This subset of particles is then remeasured. If the variation in the measured values is only due to the sample, the second set of measurements should also lie entirely below our threshold. Any “spillover” is therefore due to instrument‐specific effects—we demonstrate how the distribution of the post‐sort measurements is sufficient to extract an estimate of the cumulative variability induced by the instrument. A distinguishing feature of our work is that we do not make any assumptions about the sources of said noise. We then show how “local affine transformations” let us transfer these estimates to cytometers not equipped with a sorter. We use our analysis to estimate noise for a set of three instruments and two bead types, across a range of sample flow rates. Lastly, we discuss the implications of instrument noise on optimal classification, as well as other applications.