![]() |
Cluster Explorer |
Copyright © 2019, 2020, 2021 Joseph Trotter
ClusterExplorer is a FlowJo version 10.7.x plugin. The tool creates a cluster Expression Profile graph, Heatmap, and multiple tSNE/UMAP or measurement biaxial data plots. The plots are dynamic, may be copied to the clipboard and pasted into the FlowJo Layout (or other app), and allow the user to select in one view and display the results in the other plots, especially useful when identifying clusters to select for sorting in a near real-time workflow where HyperFinder is used to find gates that will efficiently sort the desired population defined by the clustering algorithm and/or data projection.
The Title item on the 2D Plots is an alternative active Menu to bring up a FlowJo Workspace Node tree usable for selecting population nodes defined in the FlowJo Workspace.
The Heatmap data may be copied as either an image, or as comma delimited text to the Clipboard. If using Excel, paste the content using the Text Import Wizard and select comma as the delimiter.
If multiple cluster sets are loaded into the plugin, Cluster Explorer solves the population assignment problem using the Hungarian Algorithm (as in Samusik et. al.1) such that the common populations will be rendered with the same color in the 2D plots, and the cluster IDs the same as in the FlowJo Workspace.
ClusterExplorer can be used to do additional cluster cleanup by removing outliers on currently selected Clusters (top menu bar), default is by a robust Mahalanobis approach (see link above to Minimum Covariance Determinant method). Events are removed by selecting a probability level. Once cleaned, modified selected clusters may then be exported into the FlowJo Workspace by using the Export Clusters command (see menu bar).
After downloading the plugin, place the JAR file into your local FlowJo plugins folder and restart FlowJo.
May 20, 2021
joe@cytometrygroup.org
La Jolla, California
1Samusik N, Good Z, Spitzer MH, Davis KL, Nolan GP. Automated mapping of phenotype space with single‐cell data. Nat Methods 2016; 1– 4.