Quantitative data.

For quantitative data, the underlying model is that the phenotype of a chromosome is drawn from one of two normal distributions with the same standard deviation and differing means, dependent on whether it carries the disease mutation. As those means and standard deviation are usually unknown, the program calculates the posterior likelihood on a grid of values and averages over the grid. When starting the treepeeling step, the user is asked for a minimum and a maximum value for mean and standard deviation and the number of gridpoints that are to be used between those two values. To make sure that the resulting likelihood is independent of the selected grid density, the user may want to experiment with different densities. If the phenotype data is clearly not normal, it may be worth attempting transformations of the phenotype data to make it more normal.

After providing this information, the process will be started by clicking the Start tree peeling button in this dialog box. This will open a window displaying the output from the peeling algorithm for each focal point as they are being generated, indicating the progress of the program. Buttons on the bottom of this window provide the following options to the user:

Save data

Saves the generated likelihoods to disk. If this option is not used, the peeling results will be discarded when TreeLD is closed.

Plot data

Opens a window that plots the peeling likelihood at each focal point.

Compute CI

Calculates a credible interval base on the result of the peeling algorithm. Clicking on this button opens a window that allows you to enter the percentile for the credible region. After this calculation is finished, a box appears, displaying the beginning and the end of the credible interval.

Done

Closes the peeling output window and displays the results as a graph in the Analysis Window.

Cancel

Interrupts the treepeeling analysis or the computation of the credible interval.

When this process is finished, and the window is closed, the estimated posterior distribution is displayed as a graph in the Analysis Window.