Today Kirsten Diggins presented a webinar for Expert Cytometry entitled, “Identifying and characterizing cell subpopulations from high dimensional flow cytometry data.”
Mass cytometry (CyTOF) and multiplexed fluorescence flow cytometry now allow us to measure upwards of 40 features simultaneously at the single cell level. However, traditional analysis methods like biaxial gating are often insufficient to extract meaningful information from this data. Here, we present a modular workflow for high dimensional flow cytometry data analysis that maximizes the discovery of small or rare cell subsets and provides a population-level view of expression signatures.
1. High quality results come from high quality data. Learn more about the steps involved in data cleanup and pre-processing, including scale transformations, bead normalization, and cleanup gating.
2. Computational tools for clustering, dimensionality reduction, and visualization are used to identify and characterize populations of cells in high dimensional data. Discover how these tools can be combined to automatically identify cell populations and visualize their marker expression profiles.
3. Before applying clustering methods to your data, it can be useful to identify and separate major populations of cells (i.e. CD45 high and low). Learn how to use t-distributed stochastic neighbor embedding (tSNE) for dimensionality reduction and major population gating.
4. Cell populations can be automatically identified using cluster analysis. Learn how to use
spanning-tree progression analysis of density-normalized events (SPADE) to automatically identify populations of cells in your data.
5. Heatmaps of population-level median expression values are useful for interpreting results and generating new hypotheses. Discover how to export cell cluster information from Cytobank, load the data into R, and build heatmaps of expression signatures.
For more information, see Kirsten’s work in Methods and Nature Methods: