Gene Set Test

The Gene Set Test tool tests whether GO terms are over-represented in a set of differentially expressed genes (input as a statistical comparison track) using a hypergeometric test. The tool will require a GO annotation file that must be previously saved in the Navigation Area of the workbench.

GO annotation files are available from several sources (Blast2Go, GO ontology database). Before import, check that the table does have a GO column, and if not, edit the table to change the relevant column header to GO before import in the workbench using the Standard Import function. For GO annotation files in GAF format, use the option "Force import as type: Gene Ontology Annotation file" from the drop down menu at the bottom of the Standard Import dialog.

RefSeq files are available via the Reference Data Manager, and are saved in the "CLC_Reference" folder in the Navigation Area if you have already downloaded a Reference Data Set.

It is also possible to format a text file of custom annotations into a format the Gene Set Test tool can use (see Generic annotation file for expression data format and Generic ontology annotation files).

This custom annotation file can be imported using the Standard Import functionality.

To start the tool:

        Toolbox | RNA-Seq and Small RNA Analysis (Image rna_seq_group_closed_16_n_p)| Differential Expression (Image rna_expression_folder_closed_16_n_p) | Gene Set Test (Image identify_differentially_expressed_genes_16_n_p)

Select a statistical comparison track (Image stats_track_16_n_p) and click Next (see figure 31.86). To run several statistical comparisons at once, use the batch function.

Image genesettest_output
Figure 31.89: Select one statistical comparison.

In the "Annotation testing parameters" dialog, you need to specify a GO annotation file and have several annotation testing options(see figure 31.87).

Image genesettest_output1
Figure 31.90: Select annotation tetsing parameters.

Click Next to access the "Filtering parameters" dialog (see figure 31.88).

Image genesettest_output2
Figure 31.91: Specify filtering parameters.

Instead of annotating all genes present in the statistical comparison track, it is possible to focus on the subset of genes that are differentially expressed. The filtering parameters allow you to define this subset:

Click Finish to open or save the file in a specified location of the Navigation Area.

During analysis, a black banner in the left hand side of the workbench warns if duplicate features were found while processing the file. If you get this warning message, consider unchecking the "Ignore gene name capitalization" option.

The output is a table called "GO enrichment analysis" (see figure 31.89). The table is sorted in order of ascending p-values but it can easily be sorted differently, as well as filtered to highlight only the GO terms that are over-represented.

Image genesettest_output3
Figure 31.92: The GO enrichment analysis table generated by the Gene Set Test tool.

The table lists for each GO term the number and names of Detected Genes, i.e., the total number of genes in the annotation for a given GO term which is being considered for the analysis, and of DE (Differentially Expressed) Genes. Genes that are not detected (i.e., genes that have Max group mean = 0, meaning they are not expressed in any sample) are not included in the analysis. By excluding undetected genes, we make the background of the test specific to the experiment (for example, if someone is comparing liver cells under two conditions, then the most appropriate background is the set of genes expressed in the liver).

The table also provides FDR and Bonferroni-corrected p-values. When testing for the significance of a particular GO term, we take into account that GO has a hierarchical structure. For example, when testing for the term "GO:0006259 DNA metabolic process", we include all genes that are annotated with more specific GO terms that are types of DNA metabolic process such as "GO:0016444 somatic cell DNA recombination". Also note that the p-values provided in the table are meant as a guide, as GO annotations are not strictly independent of each other (for example, "reproduction" is a broad category that encompass a nested set of terms from other categories such as "pheromone biosynthetic process").



Subsections