Automatic Clustering of Flow Cytometry Data with Density-Based Merging

Abstract

The ability of flow cytometry to allow fast single cell interrogation of a large number of cells has made this technology ubiquitous and indispensable in the clinical and laboratory setting. A current limit to the potential of this technology is the lack of automated tools for analyzing the resulting data. We describe methodology and software to automatically identify cell populations in flow cytometry data. Our approach advances the paradigm of manually gating sequential two-dimensional projections of the data to a procedure that automatically produces gates based on statistical theory. Our approach is nonparametric and can reproduce nonconvex subpopulations that are known to occur in flow cytometry samples, but which cannot be produced with current parametric model-based approaches. We illustrate the methodology with a sample of mouse spleen and peritoneal cavity cells.

Validation of cell‐based fluorescence assays: Practice guidelines from the ICSH and ICCS – part III – analytical issues

“…Automatic gating software is designed to eliminate subjective decisions (58)(59)(60)(61)(62)(63), and is an available option in commercial products. If a gating approach is appropriate for an assay, it is possible to design the test initially with a manual gating strategy, and then to evaluate automatic gating solutions prior to putting the test into routine use.…”

Automated identification of stratifying signatures in cellular subpopulations

“…For instance, a variety of automated gating approaches were developed to ease the bottleneck of manual gating. Methods use various strategies for identifying cell subsets, including nonparametric clustering (9-11), model-based approaches (12)(13)(14), density-based methods (15)(16)(17), and combinations thereof (18). However, estimating the true number of clusters in a dataset remains a challenge for these methods (SI Appendix, Fig.…”

Estimation of level set trees using adaptive partitions

“…Walther et al (2009) constructs a regular grid with associated weights that are derived by binning the data. The density of a grid point is estimated using a kernel density estimator.…”

“…The use of a regular grid limits the use of the method to lower dimensional cases. Walther et al (2009) suggest to use the method sequentially for two-dimensional projections, and to use two-dimensional scatter plots for visualization. The curvHDR method of Naumann et al (2010) mimics manual gating by applying kernel density estimation in a two-stage procedure that first finds regions of high negative curvature using the Hessian of the density estimate; see also Duong et al (2008).…”