Post Sat Nov 09, 2013 12:08 am

Principle Components Analysis (PCA)

Principle Components Analysis, or PCA, is a well-known computational technique, that was popularized for CyTOF analysis by Newell et al. [1]. Basically, PCA attempts to separate a group of events (cells, patients, etc.) according to their measured attributes. For our purposes, the attributes are each cell's expression level of markers measured in CyTOF. One can imagine the cells plotted in an N-dimensional graph, where N is the number of CyTOF markers. The algorithm then choses a set of vectors (principle components) through these N dimensions, that maximize the spatial separation of the events. Each principle component is thus "weighted" to varying degrees by different dimensions (or different markers). This allows one to see clusters of cells that are phenotypically distinct from other cells. The main caveat is that it is not obvious from a PCA graph what the actual markers influencing the separation are; for that, one must scrutinize the weighting of each of the principle components. It is, however, an effective visualization for asking how different one set of cells is from another.

Reference

1. Newell, E. W., Sigal, N., Bendall, S. C., Nolan, G. P., & Davis, M. M. (2012). Cytometry by Time-of-Flight Shows Combinatorial Cytokine Expression and Virus-Specific Cell Niches within a Continuum of CD8+ T Cell Phenotypes. Immunity, 36(1), 142–152. doi:10.1016/j.immuni.2012.01.002