Group Test for aCGH
Three different statistical tests are available to determine potential differences in status of copy number alterations between various groups. The testing is recommended to be performed on high-dimensional data nodes containing chromosomal region information.
This analysis plots the copy number aberration frequencies for different groups and indicates significant different regions between these groups.
To begin the analysis, see Running the Analyses, then perform the following steps.
To perform a Group test for aCGH analysis:
Click the Advanced Workflow tab, then open the Analysis menu.
Select Group Test for aCGH.
The Variable Selection section appears.
Define the following variables:
- Region: A high-dimensional data node containing the chromosomal regions.
- Group: Categorical data nodes separating the samples into two or more groups.
- Statistical Test: Select the test to perform:
- Chi-square: Test for the association between alteration pattern and group label. Supports multiple comparisons.
- Wilcoxon: Rank-sum test for two groups.
- Kruskal-Wallis: Generalization for Wilcoxon for more than two groups.
- Alteration type: The type of chromosomal alteration used to test the association (gains, losses, both).
- Permutations: The significance of the p-values is evaluated through permutations, and a false discovery rate is calculated. At least 10,000 permutations are recommended for final calculations. This will require a significant amount of time. (Permutations can be lowered for exploratory purposes in lieu of generating a Frequency Plot for aCGH.)
Click Run Analysis. As this may take a while, consider selecting the option Run Job in Background in the popup window. The analysis can be retrieved at a later time in the Analysis Jobs tab.
Results appear in two sections:
The chromosomal regions present in the high-dimensional data node are shown in a table, appended with p-values and false discovery rates.
Frequency plots of copy number alterations in each defined group are shown. In particular, “Mirror frequency plots” are shown; for example:
- Wiel et al. (2005) “CGHMultiArray: exact p-values for multi-array comparative genomic hybridization data.” Bioinformatics 21: 3193-3194.