This is in regards to a previous post by Dr. Gordon Smyth from June, 2012. https://stat.ethz.ch/pipermail/bioconductor/2012-June/045950.html In the previous post, Dr. Smyth stated: "voom() isn't currently setup to accept the normalization matrix from cqn, but it will be down the track." Does anyone know whether voom() can now accept CQN normalized output? I was not able to find anything in the limma user documentation on whether voom can now handle CQN normalized data as input. I'm trying to correct for sample-specific variability prior to running voom normalization and limma differential expression. I have been reading publications on sample-specific technical variability (e.g. Risso, et al. BMC Bioinformatics 2011). Is it still recommended to do within- lane quantile normalization prior to differential expression? My data is RNA-Seq from Illumina HiSeq using standard Illumina processing. I definitely have the random hexamer priming issue, but that should (theoretically at least) be consistent across samples for a single gene as Dr. Smyth points out in the previously referenced post. Any insight would be great. Kind regards, Heather [[alternative HTML version deleted]]

Given that voom is not restricted to integer values, is there any reason one couldn't simply apply the normalization manually before calling voom on the data? On Mon Nov 25 16:53:02 2013, Heather Estrella wrote: > This is in regards to a previous post by Dr. Gordon Smyth from June, 2012. > https://stat.ethz.ch/pipermail/bioconductor/2012-June/045950.html > > In the previous post, Dr. Smyth stated: > "voom() isn't currently setup to accept the normalization matrix from cqn, > but it will be down the track." > > Does anyone know whether voom() can now accept CQN normalized output? I was not able to find anything in the limma user documentation on whether voom can now handle CQN normalized data as input. I'm trying to correct for sample-specific variability prior to running voom normalization and limma differential expression. I have been reading publications on sample-specific technical variability (e.g. Risso, et al. BMC Bioinformatics 2011). Is it still recommended to do within-lane quantile normalization prior to differential expression? My data is RNA-Seq from Illumina HiSeq using standard Illumina processing. I definitely have the random hexamer priming issue, but that should (theoretically at least) be consistent across samples for a single gene as Dr. Smyth points out in the previously referenced post. > > Any insight would be great. > > Kind regards, > Heather > > > > [[alternative HTML version deleted]] > > _______________________________________________ > Bioconductor mailing list > Bioconductor at r-project.org > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor

Hi Heather, > Date: Mon, 25 Nov 2013 16:53:02 -0800 > From: Heather Estrella <hestrella at="" regulusrx.com=""> > To: "bioconductor at r-project.org" <bioconductor at="" r-project.org=""> > Subject: [BioC] Voom and CQN input > > This is in regards to a previous post by Dr. Gordon Smyth from June, > 2012. https://stat.ethz.ch/pipermail/bioconductor/2012-June/045950.html > > In the previous post, Dr. Smyth stated: "voom() isn't currently setup to > accept the normalization matrix from cqn, but it will be down the > track." > > Does anyone know whether voom() can now accept CQN normalized output? I > was not able to find anything in the limma user documentation on whether > voom can now handle CQN normalized data as input. No we have not implemented CQN style normalization in conjunction with voom. In the meantime, you could try glmQLFTest() in edgeR which gives comparable results to voom and which does accept cqn normalization matrices. > I'm trying to correct for sample-specific variability prior to running > voom normalization and limma differential expression. I have been > reading publications on sample-specific technical variability (e.g. > Risso, et al. BMC Bioinformatics 2011). Is it still recommended to do > within-lane quantile normalization prior to differential expression? I don't find this necessary in my own RNA-seq analyses, which is why I have not been that motivated to implement it in voom. I usually find TMM scale normalization sufficient. If someone was to present me with a dataset for which it was clearly required, then that would be interesting. Best wishes Gordon > My data is RNA-Seq from Illumina HiSeq using standard Illumina > processing. I definitely have the random hexamer priming issue, but that > should (theoretically at least) be consistent across samples for a > single gene as Dr. Smyth points out in the previously referenced post. > > Any insight would be great. > > Kind regards, > Heather ______________________________________________________________________ The information in this email is confidential and intend...{{dropped:4}}

Maybe that could be a good enough approximation, but in principle the adjustment should be made internally in voom so that voom can figure out the true size of each count, while also preserving the normalization factors, as it currently does with the library sizes. Gordon > Date: Mon, 25 Nov 2013 23:01:39 -0800 > From: Ryan <rct at="" thompsonclan.org=""> > To: Heather Estrella <hestrella at="" regulusrx.com=""> > Cc: "bioconductor at r-project.org" <bioconductor at="" r-project.org=""> > Subject: Re: [BioC] Voom and CQN input > > Given that voom is not restricted to integer values, is there any > reason one couldn't simply apply the normalization manually before > calling voom on the data? > > On Mon Nov 25 16:53:02 2013, Heather Estrella wrote: >> This is in regards to a previous post by Dr. Gordon Smyth from June, 2012. >> https://stat.ethz.ch/pipermail/bioconductor/2012-June/045950.html >> >> In the previous post, Dr. Smyth stated: >> "voom() isn't currently setup to accept the normalization matrix from cqn, >> but it will be down the track." >> >> Does anyone know whether voom() can now accept CQN normalized output? I was not able to find anything in the limma user documentation on whether voom can now handle CQN normalized data as input. I'm trying to correct for sample-specific variability prior to running voom normalization and limma differential expression. I have been reading publications on sample-specific technical variability (e.g. Risso, et al. BMC Bioinformatics 2011). Is it still recommended to do within-lane quantile normalization prior to differential expression? My data is RNA-Seq from Illumina HiSeq using standard Illumina processing. I definitely have the random hexamer priming issue, but that should (theoretically at least) be consistent across samples for a single gene as Dr. Smyth points out in the previously referenced post. >> >> Any insight would be great. >> >> Kind regards, >> Heather ______________________________________________________________________ The information in this email is confidential and intend...{{dropped:4}}