func-refin − refines the results of the category test
refin-YEAR-MM-DD.sh [pvalue] [pvalue-after-refinement] [cutoff]
The refinement algorithm removes significant groups on a higher level if they are only significant because they include genes from significant subgroups. The result of the refinement are the most specific significant groups (as lower groups give a more specific description of the data).
The refin-YEAR-MM-DD.sh shellscript is created by one of the four func(1) tests. It will use the same output-directory, termdb-tables and root-nodes as the invoked test.
[pvalue]
only the nodes, that are significant at this alpha are used for the refinement.
[pvalue-after-refinement]
after removal of the data of significant subgroups a significant group is tested with this alpha.
[cutoff]
only groups with at least cutoff genes are tested. This cutoff is only applied before genes are removed.
The refinement will place an resultfile for each root node into the output directory. The first columns describe the Gene Ontology groups. The remaining 5 columns show
|
1. |
+ if the group is significant after refinement, - otherwise | ||
|
2. |
left pvalue before refinement | ||
|
3. |
right pvalue before refinement | ||
|
4. |
left pvalue after refinement | ||
|
5. |
right pvalue after refinement |
A value of -1 after refinement means, that this node was’nt significant before refinement. If the p value is -1 before the refinement, then the node does’nt contain any data.
The resultfiles will be named refinement-GO-ID-pvalue-pvalue-after-refinement.txt.
func(1)
http://func.eva.mpg.de/
/usr/local/doc/func-0.4.7/
Kay Pruefer <pruefer@eva.mpg.de>, Bjoern Muetzel <muetzel@eva.mpg.de>