[Bioc-devel] GOHyperGParams Minimum Size

[Dario Strbenac]

Hello,

When doing a gene ontology analysis, it is common to remove gene sets that are small to avoid a problem of their statistics spuriously being significant. For example, I have a list of 681 significantly differentially expressed genes, but some expected overlap values are much smaller than 1.

> head(GOsummary)
       GO ID                                   Term Overlap Expected Overlap
1 GO:0051591                       response to cAMP       4       0.06257822
2 GO:0046683           response to organophosphorus       4       0.08635795
3 GO:0014074 response to purine-containing compound       4       0.10137672
4 GO:1901700 response to oxygen-containing compound       7       0.99374218
5 GO:0035914   skeletal muscle cell differentiation       3       0.05256571
6 GO:0009725                    response to hormone       6       0.64205257

This functionality is missing. Also, in the documentation of the constructor of GOHyperGParams, there is no description of what ... is for. Although it can accept variables with the same names as the slots, this is not explained. It would be good if there was a real constructor, instead of the user calling new explicitly.

--------------------------------------
Dario Strbenac
PhD Student
University of Sydney
Camperdown NSW 2050
Australia