[BioC] edgeR: Value of variation of biological variation (BCV)

Gordon K Smyth smyth at wehi.EDU.AU
Tue Aug 14 01:50:51 CEST 2012

Dear Lucia,

I don't have yet any experience with working with in-house generated 
RNA-seq data from brain tissue, so I can't say, but I think I would be 
looking for what sources of variation can be controlled in a case such as 
yours.  When we have had larger BCV from mouse experiments, we've been 
able to track the cause down to sex differences or batch effects or RNA 
contamination.  I have analysed mouse brain tissue using microarrays, and 
found that it was important to control for exactly what aspect of the 
brain was being sampled.  See Achtman et al (Science Translational 
Medicine, 23 May 2012) for such an analysis.

I have noticed that many of the data sets published as examples in 
statistical papers on differential expression for RNA-seq show very much 
larger BCV than I would be happy to have from our in-house controlled 
mouse experiments.  Whether that reflects innate variability or something 
else I don't know.

Best wishes

Professor Gordon K Smyth,
Bioinformatics Division,
Walter and Eliza Hall Institute of Medical Research,
1G Royal Parade, Parkville, Vic 3052, Australia.

On Mon, 13 Aug 2012, Lucia Peixoto wrote:

> Hi,

> I had a related question regarding BCV, it has been my recent experience
> that working in vivo with heterogeneous tissues such as brain yields a much
> higher BCV than 10%, usually close to 30%
> even though I am working with genetically identical organisms (mice, all
> males as well).
> Is that your observation as well, or does this mean that variation is being
> introduced at some of the sample processing steps?
> I do not have technical replicates, although I have n=5 biological
> replicates for each of my conditions
> Interestingly, filtering out low count reads does not seem to lower the
> average BCV much
> thanks very much for your answer
> Lucia
> On Fri, Aug 10, 2012 at 9:40 PM, Gordon K Smyth <smyth at wehi.edu.au> wrote:
>> See fourth paragraph of the Discussion of the edgeR glm paper:
>> http://nar.oxfordjournals.org/**content/40/10/4288.long<http://nar.oxfordjournals.org/content/40/10/4288.long>
>> Gordon
>>  Date: Fri, 10 Aug 2012 11:45:31 +0300
>>> From: KJ Lim <jinkeanlim at gmail.com>
>>> To: Bioconductor mailing list <bioconductor at r-project.org>
>>> Subject: [BioC] edgeR: Value of variation of biological variation (BCV)
>>> Dear edgeR users and R community,
>>> I'm analysing a set of RNA-Seq data which has 2 different genotypes
>>> (treeHS, treeLS) and time of treatment (0H, 3H, 24H, 96H) with edgeR.
>>> After carried out estimating the common dispersion, the value of BCV
>>> for my RNA-Seq data was 0.428.
>>> > hl <- estimateGLMCommonDisp(hl, hl.design, verbose=TRUE)
>>>  Disp = 0.18319 , BCV = 0.428
>>> May I ask, it is common to see the value of variation of biological
>>> variation (BCV) as such? What this value can tells about the RNA-Seq
>>> data? I have these thoughts wondering me for a while. Forgive me if I
>>> have asked a stupid question.
>>> Thank you very much for your time.
>>> Have a nice weekend.
>>> Best regards,
>>> KJ Lim
>> ______________________________**______________________________**__________
>> The information in this email is confidential and intend...{{dropped:4}}
>> ______________________________**_________________
>> Bioconductor mailing list
>> Bioconductor at r-project.org
>> https://stat.ethz.ch/mailman/**listinfo/bioconductor<https://stat.ethz.ch/mailman/listinfo/bioconductor>
>> Search the archives: http://news.gmane.org/gmane.**
>> science.biology.informatics.**conductor<http://news.gmane.org/gmane.science.biology.informatics.conductor>

The information in this email is confidential and intend...{{dropped:4}}

More information about the Bioconductor mailing list