Hi there, I am wondering: i) When do you perform count outlier detection step? Do you do it after filtering/differential expression analysis OR before both steps? ii) It doesn't seem like DESeq2 take into account normalizing samples against endogenous or consistently expressed genes for differential expression analysis. Do you recommend that? Thank you in advance!! Yoong -- output of sessionInfo(): NA -- Sent via the guest posting facility at bioconductor.org.

hi Yoong, On Wed, Nov 20, 2013 at 5:56 PM, FeiYian Yoong [guest] < guest@bioconductor.org> wrote: > > Hi there, > > I am wondering: > > i) When do you perform count outlier detection step? Do you do it after > filtering/differential expression analysis OR before both steps? > Count outlier detection is performed before the p-value adjustment, which is the point where it would make a difference before or after. So it does not count in multiple testing to the number of tests performed. ​​ > > ii) It doesn't seem like DESeq2 take into account normalizing samples > against endogenous or consistently expressed genes for differential > expression analysis. Do you recommend that? > ​DESeq2 uses the median ratio method described in Anders & Huber 2010 over all genes. It does not take into account any prior information about the genes. However, if you know a certain subset of genes are not differentially expressed, you could use the following code to produce size factors over these genes, indexed by 'nonDEgenes': sf <- estimateSizeFactorsForMatrix​( counts(dds)[ nonDEgenes , ] ) ​sizeFactors(dds) <- sf​ ​hope this helps, Mike​ > > > Thank you in advance!! > > Yoong > > -- output of sessionInfo(): > > NA > > -- > Sent via the guest posting facility at bioconductor.org. > [[alternative HTML version deleted]]