Introduction

The application of statistics to experimental data is usually a prerequisite for publication and obviously applies to the discipline of flow cytometry. However, in this field, the application of statistics to individual data files, which can often be informative and becomes essential when examining small populations, has often been overlooked. The probable reason for this oversight is the inherent ‘reliability’ of individual datasets when considering significant subpopulations and acquiring information on at least 5000 events, which is often the case in flow cytometry.

The statistics pertaining to these ‘reliable’ situations will be outlined and developed into the realm of rare event analysis. During this process, the additional requirements placed on datasets to produce statistics of a predetermined quality will become apparent.

Flow cytometers, by definition, operate on a suspension of cells or other particles that are hydrodynamically focused in a stream of fluid. Assuming that the sample has been suitably agitated, the particles will arrive at a point for analysis in a random order. The time interval between these events is, therefore, also random and this has important consequences if the cytometer is being used for sorting. Analysis is performed either online, in realtime, or offline from a dataset stored electronically. Data from flow cytometers collected in list mode to the Flow Cytometry Standard (FCS) will contain a sequential list of all the parameters collected for each of these randomly ordered events and realtime can, on some instruments, be one of these measurements. Whether the analysis is online or offline, decisions are usually being made as to whether events fall within certain regions that can be defined by the user or by some particular software.