Weasel can perform cluster analysis of multi-parameter data "on the fly". A "seed cell" is selected from any 3D display and Weasel searches the data for all similar cells belonging to the cluster on the basis of all parameters reflecting cellular properties. Following that, a collection of conventional 2D regions, using all parameters of the data, are constructed and used to form a conventional gate.
The seed cell may be selected from a 3D dotplot but usually a 3D biplot gives better spread of the data points, enabling selection of a suitable individual cell. An example of this, where a seed cell that is not contained in a major population, is selected is as follows:
|1. In this biplotted data, cluster colour coding* shows five major populations and several sparse minor populations. We may select an example cell outside the major populations. Hold CTRL and left-click on that cell.||2. Following the left-click, the selected seed cell is highlighted (see the large green dot). Now, hold CTRL, right-click and select "Find Seed Cluster".||3. The members of the cluster that includes the seed cell are then all highlighted. Finally, hold CTRL, right-click and select "Create Live Gate for Cluster". A set of conventional 2D regions are created that enclose the cluster and these are combined to define a "Live Gate" (see below).|
|4. The live gate utilizes all parameters displayed in the biplot. If the seed cell is selected from a 3D dotplot, the live gate parameters are user-selectable. These should include all parameters reflecting cellular properties (e.g. FSC, SSC, all fluorescences). If FSC-Area and FSC-Width are available, both should be selected but there is little advantage in selecting all three of FSC-Area, FSC-Width and FSC-Height.|
Note that any cell that is suspected to be of interest may be selected. This cell may also be selected from a conventional 3D dotplot display which has the advantage that the intensities of at least 3 of the cell's parameters are shown more clearly, but has the disadvantage that multiple clusters are more likely to be interspersed at any location on the display. In any case, if a single cell is visible, it can be selected.
*Data need not be clustered first using the ACluE program. That was done only for illustration in the above example. It is not clear from the static images that the biplot display is 3-dimensional but it has been dragged by mouse to vary rotation and tilt to best expose the minor population that is identified and selected for the live gate.