|
FlowJo draws only one type of contour
plot, using equal probability contouring. This algorithm generates,
in general, graphs which are most accurately interpreted by our brains,
in terms of relatively frequencies of subpopulations. All contouring
algorithms have advantages and disadvantages, but probability contouring
is probably the most informative. Combined with the display of outliers,
this generates an extremely informative graphic that does not over-emphasize
narrow peaks nor hide low-frequency information.
There are several options available in drawing the
plots (available from the menu). These
include the option of drawing "2%", "5%", "10%",
or "Logarithmic" contour plots, "smoothed" or
not, and "outliers" or not. Contrary to the belief of
many people, smoothing does not alter the data nor does it distort
it. Thus, it is by default left on. Note that turning smoothing
off can cause the program to take much longer to draw the plots.
"Show Outliers" controls whether events falling outside
of the lowest contour are drawn with dots--thus, combining the best
advantages of dot plots (revealing low-frequency data) with the
best of contour plots (allow high-frequency data to be accurately
displayed). 2%, 5% and 10% plots have 50, 20, and 10 contour lines
(respectively) that are distributed so that the same number of cells
fall between each pair of contour lines. Logarithmic plots distribute
the contour levels logarithmically in height; each contour line
encloses twice as many events as the previous.
The first set of graphs below compares the same
data using different contouring thresholds. The second set of graphs
compares 5% probability plots with different options.
You may view a comparison
of all the bivariate displays for this
same data.
|
|