This method can help answer key questions in the biomedical field, such as understanding the phenotype of biologically relevant sub-populations. The main advantage of this technique is that it allows an individual with no programming experience to analyze their cytometry data with the latest in high-dimensional techniques. To begin an analysis pipeline, first select the type of cytometry, and the number of events to sample from the file.
Next, click Gate Population, select the cell populations of interest, and enter the percentage of events for downstream analysis. Then select the number of channels to be used for the analysis in the list box. For T-distributed Stochastic Neighbor Embedding, or t-SNE analysis, click t-SNE to begin computing the reduced dimensionality dataset.
When the set has been computed, click Save TSNE Image, and in the Marker-Specific t-SNE popup menu, select a specific marker of interest. A figure will appear showing a heat map representation of the t-SNE plot that can be saved for figure generation. To begin clustering analysis, select an option in the Clustering Method list box, and click Cluster.
To sort the clusters by a marker of interest, select the appropriate corresponding option from the Sort popup menu, and click Ascending, Descending, to update the list of clusters in the Clusters list box. To set a minimal threshold value for a given cluster across a certain channel, select an option from Threshold popup menu, and set an appropriate threshold. Once threshold has been set, click Add Above Threshold, or Add Below Threshold, to specify the direction of the threshold, and enter a numerical cut-off in the Cluster Frequency Threshold box, in the Cluster Filter panel, to set a minimal threshold for the frequency of a cluster.
To select clusters for further analysis individualization, select the clusters of interest, in the Clusters list box. And use the Select button, to move the options to the Cluster Analyze list box. To create heat maps of the clusters, select the clusters of interest in the Cluster Analyze list box, and click the HeatMap of Clusters button.
To create a high-dimensional box plot, or a high-dimensional flow plot, select the clusters of interest in the Cluster Analyze list box, and click either High-Dimensional Box Plot, or High-Dimensional Flow Plot, to visually assess the distribution of given channels of various clusters across all dimensions. To show clusters in traditional 2D flow plots, select the appropriate transformation and channel in the Conventional Flow Plot panel, and click Conventional Flow Plot. Here, a representative t-SNE analysis of heat maps for various markers within a myeloid panel analysis pipeline, are shown.
Using the fast greedy implementation within ExCYT to cluster the data with 100, 000 of the nearest neighbors, 19 sub-populations of cells were revealed. Comparison of the original heat maps to the clusters created by ExCYT allowed similar clusters of the myeloid cells to be identified between the two data groups. Analysis of the lymphoid panel with a more conventional and faster hierarchical clustering approach, yielded similar marker distributions via t-SNE heat maps.
Further, clustering of the data via hierarchical clustering demonstrated similar clusters of lymphoid cells. Notably, a unique regulatory T cell population was also identified via a high-dimensional flow plot. To quickly and quantitatively assess co-associations among markers, first a hard K-means clustering algorithm was used to lay down 5000 clusters on the two dimensional t-SNE data.
The median expression of all the markers from all the clusters was then used, to create a heat map from these clusters, allowing co-associations to be easily identified, such as the co-association of Tim-3, PD-1, CD38 and 4-1BB. While attempting this procedure, it's important to remember to explore the different parameters, such as different clustering methods, to fully explore the data you are studying.
