Separating Outliers from HCS Data

Data

Please see Tutorial 1 for a description of the data set used in this tutorial.

Outliers

HCS data contain a lot of outliers, resulting from primitive image processing algorithms most of the time. Because most of the variables in an HCS data set are directly or indirectly based on nuclei identified by image analysis, it is a good idea first to look at DNA profiles to identify and weed out outliers.

Invoke histogram in the primary console:

to get a histogram menu:
Make the following selections:
After clicking the OK button in the above menu, we'll get the following DNA profile after reshaping the plot window:

Cells in this DNA profile were divided into 100 bins. If you are actually running Panmo to follow this tutorial step by step, the number of bins in your DNA profile may not be 100. You can invoke Set Histogram Number of Bins... from the right click menu over the DNA profile to make it 100.

Identifying Outlier Cells

This tutorial illustrates how to identify the cells behind those outliers. A 147-second screencast of this tutorial and some extra materials is here.

Getting the "Good" Cells

This tutorial illustrates how to separate "good" cells from outlier cells by logical zooming
Start the tutorial or view its screencast (60 seconds).

This tutorial illustrates how to separate "good" cells from outlier cells by painting.
Start the tutorial or view its screencast (56 seconds).