Identifying differential cell populations in flow cytometry data accounting for marker frequency

We present a statistical test that compares a novel abundance score, SpecEnr, to discover biologically meaningful and interpretable differential cell population that predict a given phenotype or disease from flow cytometry data. Existing methods for differential cell population identification compare a limited set of prespecified cell populations, find differential cell populations as a byproduct of another procedure, or compare overlapping cell populations in a search space of all possible cell populations. Though thorough, analyzing all possible cell populations can be difficult as many cell populations share cells with one another. For example, an increase in one cell population may induce an increase in several other cell populations that share its cells. Our method solves this issue by taking into account these dependencies. By comparing independent score, SpecEnr, we find differential cell populations that are the source of these changes and we show how these results can be easily interpreted via a lattice-based visualization tool.