In conventional flow cytometry sorting, hardware and instrument restrictions permit only one or two-dimensional regions to be used for gating. In addition, these gates can only be applied to measurements made during data collection. Once the data is collected, sophisticated machine learning algorithms present in software packages like FlowJo™ may be used to identify cell populations using dimensionality reduction and clustering techniques. These methods create derived parameters, such as cluster ID or locations on a t-SNE map. These parameters are a vital component of modern multi-dimensional data analysis and are not measurements present within the cell sorter hardware.
The authors of this poster have developed a workflow featuring HyperFinder™, a plug-in to FlowJo™, which allows populations identified by any appropriate means, including machine learning techniques and derived parameters, to become training sets that HyperFinder™ can use to create a fully optimized gating strategy. The workflow additionally utilizes other new FlowJo™ features that can export the HyperFinder™ generated gating strategy into a new BD FACSDiva™experiment where the optimised solution becomes a new worksheet along with any modified compensation or scaling performed in FlowJo™.