Optimization of flow cytometry panel and gating strategy design with Discriminative Information Measure Evaluation (DIME)

Cliburn Chan, Lin Lin, Jacob Frelinger, Patricia D'Souza, Valerie Hebert , Dominic Gagnon, Claire Landry, Jennifer Enzor, Janet Ottinger, Maria Jaimes and Mike West

May 2010

The design of a panel to identify target cell subsets in flow cytometry can be difficult when specific markers unique to each cell subset do not exist, and a combination of markers must be used to identify target cells of interest and exclude irrelevant events. In the absence of any standard for designing gating strategies, different laboratories may end up using very different panels and gating strategies to isolate the same target cell subset. In the context of highly standardized multi-center studies, local variations in marker selection and gating strategies may result in increased variability and reduced reproducibility of the assay. Thus, the ability to objectively measure the contribution of a marker or group of markers towards target cell identification independent of any gating strategy could be very helpful for both panel design and gating strategy design. In this paper, we propose a Discriminative Information Measure Evaluation (DIME) based on statistical mixture modeling; DIME provides numerical measures of the "usefulness" of markers for identifying a target cell subset. We show how DIME provides an objective basis for inclusion or exclusion of specific markers in a panel, and how ranked sets of such markers can be used to optimize gating strategies. An illustrative example of the application of DIME to streamline the gating strategy for a highly standardized CFSE assay is described.

Keywords: CFSE standardization, Discriminative Information Measure Evaluation (DIME), Gating strategy optimization, Mixture model, Panel design.

Supplementary material: Data and Matlab code (with some supporting utility functions) for the DIME computations are available here . The example is detailed in the short Matlab code ExampleReadme.m

The mixture model fitting used in the paper is based on standard Bayesian Dirichlet process mixtures of Gaussians, and was performed using the Bayesian EM (BEM) algorithm in the software available from our group at the GPU StatSci page under the software link.

Research reported here was partially supported by grants from the U.S. National Science Foundation (DMS-0342172) and National Institutes of Health (U54-CA-112952, P50-GM081883 and RC1 AI086032). Any opinions, findings and conclusions or recommendations expressed in this work are those of the authors and do not necessarily reflect the views of the NIH and/or NSF.