Cytometry advances the study of cellular phenomena at the single-cell level by providing the necessary information to find subpopulation complexity of a heterogeneous sample. The analysis of cytometry datasets is becoming more challenging as cytometry advances increase the data size and dimension. Flow cytometry typically monitors at most 12 genes per cell; however, in recent years, mass cytometry (CyTOF) has taken the routine upper limit to 45 genes per cell. Biologists traditionally analyze the data by manually drawing polygon enclosures around subpopulations on a series of 2D projections of the data; this procedure becomes exponentially harder and time-consuming with increasing dimensions. To aid both experts and non-experts in finding the subpopulation complexity of cytometry samples, we introduce and apply partition-assisted clustering (PAC), which is implemented as an R package called PAC, to enable consistently accurate and efficient analysis of low and high-dimensional single-cell datasets. PAC utilizes the data density implicitly and enables the discovery of cell subpopulations in the dataset without computational bias due to systematic elimination of data points in the analysis.
