Phenotypes

StrataQuest has no engine that makes phenotypes. They live in the diagrams, where you assemble them with the tools a flow cytometrist uses. The User Manual describes Phenotype Diagrams as related to "phenotypes definition and specific cells identifications", and it speaks of the "gating strategy used in standard scattergrams for defining a cell phenotype".

1. Measure. Each cell needs a marker intensity per channel, from Standard MeasurementsLoading....
2. Split each marker. On a scattergramLoading... or histogramLoading..., a cutoffLoading... divides an axis into positive and negative, and two cutoffs make four quadrants. A gateLoading..., a rectangle, ellipse or polygon, selects a region instead. Both are usually set on a marker's mean intensity.
3. Narrow step by step. Set a new diagram's Input Gates to gates, quadrants or phenotypes from other diagrams, and it shows only those events. That is how CD8+ within CD3+ is drawn.
4. Use the result. Phenotypes appear in the Input Gates list of other diagrams, as the Primary and Secondary Phenotype of Phenotype InteractionsLoading..., and as the subset for a Statistics Report value such as Count or Percent. Violin diagrams compare a marker across phenotypes, and 3D diagrams can show several at once.

Two cautions follow from how gates behave. Gates may overlap, so an event can belong to several at once; Bring to front and Apply color to belonging events set only the order in which they are colored. The manual doesn't describe a step that gives every cell exactly one phenotype, so if your populations must not overlap, draw the gates that way. In 8.1 you can also build phenotypes from imported transcriptomics data, with Genes Phenotype in the Diagrams Add list.

Source: StrataQuest 8.1 User Manual §12.6 (pp. 177–205), §12.6.8–12.6.9 (pp. 206–214), §13.1 (pp. 231–234); New Features (pp. 8–9); Engines Reference ch. 34 (pp. 483–484).

A flow cytometry inheritance. Gating comes from flow cytometry, where populations are defined by a sequence of two-parameter plots, each restricted to the cells the previous plot selected. That sequence is the gating strategy, and published studies report it so others can reproduce the populations. Tissue analysis borrows the method and keeps each cell's location.

Regions of marker space. In classification theory, each cell type is a region of the space whose axes are the markers, and a good set of definitions divides that space without gaps or overlaps. Hand-drawn gates promise neither. Cells that fall in no gate, or in two, are worth looking at: they are often the dim, the damaged or the badly segmented.

Too many combinations. With n markers there are 2ⁿ combinations, so ten markers give 1,024. Most are rare or implausible. In multiplex imaging, a cell that seems to carry two lineage markers, such as CD3 and CD20, often turns out to be two touching cells whose signals bled into one outline. Good panels name the handful of combinations that match real populations.

Where the errors live. Every phenotype inherits the boundaries of its gates. Cells near a cutoff, the dim positives, are the ones most often misclassified, and a phenotype that needs four markers to be right collects the boundary errors of all four.

Here is a gating strategy for a six-marker immuno-oncology panel: CD3, CD8, CD20, FOXP3, PD-L1 and CK.

1. Plot CK against CD3 mean intensity and set a cutoff on each axis. The CK+CD3− quadrant holds tumor cells and the CK−CD3+ quadrant T cells.
2. Make a second scattergram with the T-cell quadrant as its Input Gate, plotting CD8 against FOXP3. Its quadrants give CD8+ cytotoxic T cells, CD8−FOXP3+ cells (regulatory T cells, by the usual convention) and CD8−FOXP3− T cells.
3. Make a histogram of PD-L1 with the tumor quadrant as input. A cutoff splits PD-L1+ from PD-L1− tumor cells.
4. Plot CD20 for the CK−CD3− cells and gate the B cells.
5. Check each population with View Backward Data, which outlines its cells in red on the image. Then add Statistics Report columns for the count or percent of each population per region.

The populations at the end of each branch are your phenotypes. A ratio such as cytotoxic to regulatory T cells comes from a Derived Statistic, which divides one statistic by another. In cancer research generally, that ratio is studied as a measure of the immune balance inside a tumor; for your sample, the numbers come from the report.