Dear all, I am grateful to you for the prompt reply to my query. 1) So I think adj.p.value with "BH" adjust.method (which also takes " FDR" issues into consideration) will be my choice for extracting significant DEGs. But I am looking into the propnulltrue function also to get more ideas. 2) I do however need some ideas on your suggestion regarding "If you want to give even more priority to larger fold changes, then we recommend that you use treat(). This is better than just cutting on estimated logFC value." Can you please enlighten me on the issue? 3) I have one more question regarding the utility of genefilter() in pre-filtering normalised data based on mean expression intensity prior to LIMMA. I all advisable examples filtering has been done for minimum expression intensity as 3.5 and follow-up statistics is multtest,SAM or t-test with multiple adjustment. Do you recommend to do it before LIMMA? Thanks in advance for your suggestions. Deb. On Wed, Dec 4, 2013 at 1:48 AM, Gordon K Smyth <smyth@wehi.edu.au> wrote: > Dear Deb, > > Well, this is not really a simple question at all. The way that you > prioritise your discoveries is not just a matter of statistics, but also > depends on the context and aims of your study. That is why limma offers > different options. > > If you want to know what the limma developers do, have a look at the case > studies in the limma User's Guide. We do not actually recommend either of > the options that you mention. > > The most common analysis would be to simply choose genes by FDR. But > please don't ask me what cutoff you should use for FDR. It is quite common > to use 0.05 or 0.1, but there is no correct value and this is for you to > decide on the basis of the science of your own study. > > If you want to give even more priority to larger fold changes, then we > recommend that you use treat(). This is better than just cutting on > estimated logFC value. > > There is no B.value cutoff. The B-statistic is for ranking, not for > absolute cutoff. I do not know where you might have got the cutoff value > "4" from. I have not seen anyone suggest this. To use the B-statistic for > absolute cutoff would require estimating the overall proportion of DE > genes, which limma can do by propTrueNull but doesn't do by default. > > Best wishes > Gordon > > Date: Mon, 2 Dec 2013 10:15:53 -0800 (PST) >> From: "deb [guest]" <guest@bioconductor.org> >> To: bioconductor@r-project.org, devbt15@gmail.com >> Subject: [BioC] (adj.p.value & log2FC) or (B.value & log2FC) >> >> >> Hi Sir, >> I have a simple question regarding cut-off parameter to be used for >> filtering out DEGs from the topT object obtained using LIMMA. >> Which statistics is preferred more and why? >> 1) filter of adj.p.value and log2FC. >> 2) filter of B.value and log2FC. >> I mean both give similar ranking of genes but usually I have seen people >> using an adj.p.value cut-off of 0.01. What is the minimum cut-off value for >> B.value?(Is it 4?) >> It is a technical question so I do not have any output to be put in the >> next sessionInfo() field. >> Thank you for your input Sir. >> Regards. >> Deb. >> >> >> >> -- output of sessionInfo(): >> >> topT<- topTable(fit3, coef=1, adjust="BH", sort.by="B",number=nrow(data)) >> > > ______________________________________________________________________ > The information in this email is confidential and inte...{{dropped:10}}