After normalizing the data, we can perform differential expression analysis.

• Select the Normalized counts data node
• Select Statistical analysis from the task menu
• Select Detect differential expression (GSA) from the Statistical analysis section of the task menu (Figure)

The Include attributes page shows all available attributes for analysis (Figure). Here, we only have one attribute, Treatment.

• Select Next to continue with Treatment as the selected attribute

The Comparison selector page will open (Figure). 

It is easiest to think about comparisons as the questions we are asking. In this case, we want to know what are the differentially expressed genes between untreated and treated cells. We can ask this for each dose individually and for both collectively. 

The upper box will be the numerator and the lower box will be the denominator in the comparison calculation so we will select the 0μM control in the lower box. 

• Select 5μM in the upper box
• Select 0μM in the lower box 
• Select Add comparison to add 5μM vs. 0μM to the comparison table (Figure)
• Repeat to create comparisons for 10μM vs. 0μM and 5μM:10μM vs. 0μM (Figure)

By default, the GSA will use a lognormal with shrinkage model for its analysis of variance of each gene. This is appropriate for most data sets and will tend to give accurate and reproducible results. We will use the default settings under Advanced Options in this tutorial. To learn more about the GSA, please see the Differential Gene Expression - GSA user guide. 

• Select Finish to perform GSA as configured

A Gene analysis task node and a Feature list data node will be added to the pipeline (Figure).