Mathematical evaluation of two parameter flow cytometric histograms

Abstract

A general procedure was developed to evaluate two‐parameter flow cytometric data. These data usually show overlapping distributions representing different subpopulations. To calculate the fractions of cells in the various subpopulations, the maximum likelihood method is used, assuming an unknown number of Gaussian distributions with constant coefficient of variation in the histogram. A test on simulated data shows that Gaussians can be separated clearly by the procedure if the distance between their mean values is larger than twice the maximum standard deviation. The analysis of different measurements of double‐stained (DNA/protein) cultured cclls out of the same pool yield the same fractions in the phases of the cell cycle with acceptable error rates. To simulate samples with different fractions in the phases of the cell cycle, defined mixtures of proliferating and resting cells of the same type were prepared and measured for DNA and cell size. The calculated fractions are consistent with the expected variation of G1, S and G2 + M fractions depending on the mixture ratio.