[BioC] EdgeR for proteomics data

[Pekka Kohonen]

Hello,

I have used RNA-seq statistics to analyze proteomics data myself. I
tried using edgeR but think that the spectral count numbers in
proteomics data are too small that you get artefacts, especially of
lowly expressed proteins. I then turned to limma/voom to estimated
mean-variance relationship for the data and then did analysis with
limma/eBayes. I was pleased with the results and the speciicity of the
method, limma/voom is generally a bit more conservative than edgeR
(but not much less sensitive). I experimented a bit with the loess
span for the voom (increasing it to 0.75 I think), since there was a
dip in the distribution otherwise (probably due to low amount of data)
that could lead to some spurious hits.

I think that RNA-seq analysis packages are great for label-free
proteomics, but one neds to be a bit careful with them. Otherwise the
procedure is the same (starting with a matrix of counts).

Best, Pekka

2014/1/14 Ryan <rct at thompsonclan.org>:
> Hi,
>
> As mentoined in the help text for calcNormFactors, the TMM normalization
> method is described in the paper "A scaling normalization method for
> differential expression analysis of RNA-seq data" by Robinson & Oshlack. The
> best way to familiarize yourself with this method would be to read the
> paper: http://genomebiology.com/2010/11/3/r25
>
> For what it's worth, one of my colleagues used edgeR on some proteomic data
> and decided that the default normalization strategy was not suitable for his
> data. I don't remember exactly what he ended up using instead.
>
> -Ryan Thompson
>
>
> On Mon Jan 13 17:32:36 2014, Phinney, Brett wrote:
>>
>> Hi everyone, I have been experimenting with using EdgeR with proteomics
>> data (spectral counts for now). I was a little confused how the TMM
>> normalization works on proteomics data. I  basically just read in my
>> spectral counting data  as a data matrix
>>
>> And then
>>
>> normFactors <- calcNormFactors(counts)
>>
>> but I'm not sure exactly how it is calculating the normalization factors?
>>
>> Any help would be greatly appreciated
>>
>> Cheers
>>
>> Brett
>>
>>
>> ---
>> Brett S. Phinney, Ph D.
>> UC Davis Genome Center
>> www.proteomics.ucdavis.edu<http://www.proteomics.ucdavis.edu>
>> cell = 530-771-7055
>>
>>
>>         [[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
>
>
> _______________________________________________
> 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