Hi Gordon, Thanks for the clarification. Two more related questions: decideTests(object,method="separate",adjust.method="fdr",p.value=0.05) is the default for command-line limma decideTests? What are the defaut method and adjust.method for affylmGUI? Is there a reference which elaborates other methods such as '"heirarchical"' (adjusts down genes and then across contrasts is given from the help menu)? Thanks again for providing such a great package. Wen -----Original Message----- From: Gordon Smyth [mailto:smyth@wehi.edu.au] Sent: Sunday, September 11, 2005 1:24 AM To: Luo, Wen; Steffen Moeller Cc: Keith Satterley; James Wettenhall Subject: Re: [BioC] Question (vennDiagram vs. topTable) for AffylmGUI 1.3.0 >-----Original Message----- >From: bioconductor-bounces at stat.math.ethz.ch >[mailto:bioconductor-bounces at stat.math.ethz.ch] On Behalf Of Steffen >Moeller >Sent: Friday, June 03, 2005 7:26 AM >To: bioconductor at stat.math.ethz.ch >Subject: [BioC] Question (vennDiagram vs. topTable) + Patch (plotTitle >for vennDiagram) for AffylmGUI 1.3.0 > > >Dear list, dear James and dear Ken, > >the selection for a particular P value in the topTable of affylmGUI and >the selection of genes differentially expressed in the vennDiagram of >affylmGUI show different numbers. This is most likely of no surprise >since the probabilities might have different semantics. > >Does anybody have a hint for me how to get the topTable output in sync >with the vennDiagram? Or should I instead construct the vennCounts from >the results of the topTable function? Wen Luo has written to remind me there hasn't been a response to this question. When using limma (command line) the answer is that you use topTable(..., method="BH") and decideTests(..., adjust="BH") and the results of topTable and vennCounts will agree. This is documented under ?decideTests. afflmGUI is currently is transition from one software engineer to another, and I'm am not sure myself how much of this is implemented in the GUI interface. >Is anybody working on extending affylmGUI for an analysis of associated >GO terms? We provide a web-interface for GO analysis at http://gostat.wehi.edu.au, and Robert Gentleman has provided a nice command line R package GOstats. There are no plans at the moment to incorporate such things into afflmGUI. Gordon >Kind regards and many thanks for affylmGUI >Steffen

At 07:48 AM 23/09/2005, Luo, Wen wrote: >Hi Gordon, > >Thanks for the clarification. Two more related questions: > >decideTests(object,method="separate",adjust.method="fdr",p.value=0.05 ) >is the default for command-line limma decideTests? > >What are the defaut method and adjust.method for affylmGUI? The method used by limmaGUI and afflmGUI is equivalent to the command line decideTests(object, method="nestedF", adjust="none", p.value=0.01) In other words, it is based on the older command classifyTestsF(). >Is there a reference which elaborates other methods such as >'"heirarchical"' (adjusts down genes and then across > contrasts is given from the help menu)? I don't know of any papers at all yet on multiple testing for microarray linear models (i.e., for multple linear models). The methods implemented in limma do not yet have any further documentation. Gordon >Thanks again for providing such a great package. > >Wen