Enumeration of Merged Cell Subpopulations
Francis L. Battye
The Walter & Eliza Hall Institute
Several solutions have been proposed for the problem of enumerating weakly fluorescent cells that are not fully resolved from the unstained population in a flow cytometric histogram. Most of these solutions require a negative control histogram that also characterizes the negative cells within the stained sample, a requirement often unfulfillable because of non-specific staining. To circumvent that requirement, a background subtraction method was described in which the negative distribution was shifted according to a model of non-specific staining. This technique had the disadvantage of requiring some subjective choice of computational parameters and also often resulted in incomplete subtraction of the putative ÒnegativeÓ cells.
In an alternative approach, a method of fitting histogram data using synthetic curves has been investigated. These curves contain parameters for amplitude, position, width skew and skirt. The flexibility of peak shape available when using these parameters unconstrained, while allowing excellent fitting to experimental histograms, can give rise to an unlikely variety of shapes for the subpopulation distributions. Consequently this implementation allows parameters to be individually locked or those of separate curves to be linked to mimic sub-population distributions all having similar shapes.
As an adjunct to these dissections of single parameter histograms, the other measured flow cytometric parameters should be considered. More accurate information often may be extracted by considering a two-parameter display exploiting differences of size, nuclear shape or autofluorescence between the positive and negative cells. The partitioning then may be aided by cluster analysis, multi-dimensional curve fitting or by simple construction of 2-dimensional regions.