Kevin Coombes, PhD ProfessorMedical College of Georgia, Department of Population Health ScienceAugusta University
Mass cytometry is a relatively new technology that combines antibody-based protein recognition with mass spectrometry. It is capable of simultaneously measuring the levels of 40-50 proteins in hundreds-of-thousands of single cells from every sample. Most existing analyses of single-cell data (whether proteomics or RNA) have concentrated on first separating/clustering cell types and then assessing changes in the expression of one protein at a time. Using a published data set containing a total of 58 samples either from normal B cells or from patients with different subtypes of acute myeloid leukemia (AML), we have explored how relationships between collections of proteins change depending on the context. This analysis includes changes both between different kinds of samples and between different cell types within a sample. In this talk, I will first describe a new statistic we have developed to compare changes in correlation between pairs of proteins. Most of the talk, however, will describe applications of topological data analysis to discover higher-dimensional non-linear relationships between sets of proteins.