Misty Mountain Clustering: Application to Fast Unsupervised Flow Cytometry Gating

Flow cytometry experiments record, in individual cells, the fluorescent intensity of different fluorophores that correspond to features such as the levels of specific proteins. An assay typically generates a large number (order 106) of data points in a two or higher dimensional space. The grouping of cells (data points) having similar features, which is referred to as gating, is usually done manually by an expert. We developed software that performs efficient unsupervised gating determining the number of clusters, and the points belonging to each cluster. The program analyses the cross-sections of the histogram created from the data points. The method is particularly efficient in the case of large number of data points such as 104-106. The overall run time for the composite steps of the algorithm increases linearly by the number of data points. In our example 1 million data points, shown in the left part of the figure, were analyzed within 6 seconds on a standard laptop PC. The analysis resulted in 20 clusters, shown in the right side of the figure. The code number of the largest cluster is 1, the second largest is 2, etc.View Large Image | View Hi-Res Image | Download PowerPoint Slide