Vaccine and Infectious Disease Division

Optimizing flow cytometry data transformation

High-throughput screening of cell populations by flow cytometry has dramatically increased the productivity of cellular phenotyping assays, and is often used by VIDD scientists to study different populations of immune cells, but requires automated processing methods to maximize its potential.  Many of these computational processing steps have been rigorously scrutinized, but the nuances of the data transformation step, which affect population identification, remain unclear.  Many transformations offer user-customizable parameters which are meant to fine-tune their output, but the effects are not intuitive for most users.

VIDD post-doctoral research fellow Greg Finak, VIDD associate member Raphael Gottardo and colleagues analyzed several commonly used transformation methods, studying their effects on both hypothetical and real-world samples to determine the merits of each algorithm and their optimal parameters.  They determined that for most data the currently preferred transformations are best, but that the optimal transformation and parameters for that transformation often depend on the specific data.  They have developed an algorithm for the BioConductor flow cytometry analysis software package that will help users optimize transformation parameters for their own experiments.

Finak G, Perez JM, Weng A, Gottardo R.  Optimizing transformations for automated, high throughput analysis of flow cytometry data.  BMC Bioinformatics. 2010 Nov 4;11:546.

Fred Hutchinson Cancer Research Center is a world leader in research to prevent, detect and treat cancer and other life-threatening diseases.