Revealing the Immune Population Hierarchy Using Novel Interactive Visualization Tools, a Comparison of Methods: Sunburst, viSNE, and Spade on Mass Cytometry Data

Abstract

The development of new technologies for high-parameter data has resulted in a critical bottleneck: identification of immune subsets is restricted to expert-based analysis, focusing on post-acquisition characterization of cell populations. Identification of cell subsets in flow cytometry has primarily focused on manual analysis, despite the fact that computational tools have proven useful for high-parameter and cross-sample comparisons. Sharing well-annotated data improves transparency and facilitates vital reproduction of results by external groups. Adoption of these new tools for immune subset discovery requires thorough collaborative investigation and validation of identified cell populations. To this end, in this study we compare the ease of discovery of immune subsets by comparing analysis through the use of three visualization tools: the sunburst hierarchy, the SPADE tree, and dimensionality reduction using viSNE. The sunburst hierarchy is a visual and interactive representation of traditional manual gating, whereas the SPADE tree is a semi-automated clustering and visualization tool for identification of cell subsets. viSNE allows interaction with high parameter data in the context of two-dimensional space where gating can be accomplished. In this study, we demonstrate the ability to automatically elucidate many immune subsets using Cytobank via an iterative analytic approach, combining computational tools (viSNE and SPADE) to recapitulate manually derived cell subsets. DisclosuresChen:Cytobank, Inc: Employment, Equity Ownership. Kotecha:Cytobank, Inc: Employment, Equity Ownership.