Carnegie Mellon University

Causal Discovery from Mass Cytometry Data: Ioannis Tsamardinos

Abstract: The emergence of Mass Cytometry technology presents unprecedented opportunities for causal discovery algorithms. This technology measures protein concentrations in single cells; in contrast, all other mass-throughput technologies in biology measure averages over millions of cells creating statistical problems to standard causal discovery algorithms. In addition, mass cytometers can measure about 10K cells per second, creating datasets with millions of samples (vs. at most a couple of hundred samples for mass-throughput omics data). In this talk, we present our preliminary results causally analyzing public mass-cytometry datasets. In addition, we present new advances in Integrative Causal Analysis and a proof-of-concept application to the co-analysis of multiple, heterogeneous cytometry datasets obtained under different experimental conditions and measuring overlapping variable sets.