[BioC] choosing normalization method for RNA-seq analysis

[Biase, Fernando]

Hi Dr. Huber,

I believe you are correct in terms of the weakness of the experimental design. Unfortunately, I could not change it. I have 9 biological replicates sequenced for one group and 5 for another. Is there a down side of using DEseq to compare conditions with unbalanced sample number?

One can argue that choosing either one of the normalization factors does not matter because I have about 85% of concordance of the results, but others can easily argue that it matters because of the difference on the results. And to add to the discrepancy on the results, when I analyze the data using one or the other normalization factor, there is a shift on the number of up regulated genes.

For example:
	Genes Up in A 		genes Up in B 
RLM        	2953			3322
TMM		3114			2632	

I think it may easily affect the biological interpretation. 

I appreciated your reply,

Thanks,
Fernando



-----Original Message-----
From: bioconductor-bounces at r-project.org [mailto:bioconductor-bounces at r-project.org] On Behalf Of Wolfgang Huber
Sent: Saturday, March 19, 2011 5:21 AM
To: bioconductor at r-project.org
Subject: Re: [BioC] choosing normalization method for RNA-seq analysis


Fernando,

does it matter? I.e., for your data, do you get substantially different results from the different methods? If so, that might indicate a weakness of the data quality or of the experimental design that you would need to investigate in a manner specific to the dataset.

Unsurprisingly, my personal favorite is "RLE".

	Wolfgang

Biase, Fernando scripsit 17/03/11 22:53:
> Hi
>
> Working with edgeR package to analyze RNA-seq data, we have three 
> options of normalization within calcNormFactors() function. Is there a 
> way to test which one is more suitable for the data? If there is no 
> test, is there a criteria to choose one over the other?
>
> Thanks in advance, Fernando
>
> _______________________________________________ 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


-- 


Wolfgang Huber
EMBL
http://www.embl.de/research/units/genome_biology/huber

_______________________________________________
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