Statistical Methods for Flow Cytometry Cross Sample Comparison

• Activity Number: 253
• Type: Contributed
• Date/Time: Monday, July 30, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Mengya Liu*+
• Companies: Southern Methodist University
• Keywords: Multivariate distributions ; Flow Cytometry ; cross sample coparison

Abstract

Flow cytometry (FCM) is a powerful technique to measure the multiple characteristics of individual particles, usually cells. In the FCM data analysis, a gate is a numerical or graphical boundary that can be used to identify the cell subpopulations which share a set of common properties. After gating, another significant challenge in FCM data analysis pipeline is to match corresponding subpopulations across multiple samples for comparisons across phenotypes or time points. From statistical point of view, the n-parameter clinical FCM data can be treated as n-dimensional data sampled from different multivariate distributions corresponding to different cell subpopulations. Therefore, the problem is simplified to multi-dimensional distribution similarity measure. I propose three potential methods for cross-sample comparison, namely Friedman-Rafsky Test, Maximum Mean Discrepancy method and Density Estimation and K-S Distance method. Those methods are suitable for measuring the similarity between multi-dimensional distributions, but have not been applied to FCM analysis before.