Enrichment Tests
Select one or more methods for testing for enrichment. The method is applied separately for each significance variable that you specify.
| • | The Fisher Exact test first creates 0-1 variables from the significance variables based on the {link}{Emphasis}Significance Variable Cutoff for Fisher Exact Tests specified below. (If you have already specified 0-1 variables, then any cutoff between 0 and 1 works.) It then constructs 2 x 2 tables for each category and uses the hypergeometric distribution to compute p-values. Refer to the documentation of PROC MULTTEST in SAS/STAT for details. |
| • | The Cochran-Armitage linear trend test is constructed for each variable by discretizing it into bins, alleviating the need for a single arbitrary significance cutoff. You can adjust the number of bins in the {link}{Emphasis}Number of Bins for Cochran-Armitage Tests specified below. Refer to the documentation of PROC MULTTEST in SAS/STAT for details. |
The PAGE test works by constructing a two-sided z-score for each category as described in Kim and Volsky, BMC Bioinformatics. 2005; 6: 144. The z-score is computed as:
(Category Mean - Overall Mean)*sqrt(Frequency)/(Overall Std).
To Specify Enrichment Tests:
| 8 | Highlight a single test, or hold down while left-clicking on multiple tests, in the Enrichment Tests field. |
