[BioC] edgeR

[Gordon K Smyth]

Dear Jim,

No, edgeR does not generate simulated data sets for you.

Let me gently point out that this is no unique "null" for RNA-Seq or SNP 
data.  Considerable thought needs to go into what distributional 
assumptions are appropriate for genomic data.  You might start by reading 
some published papers that compare differential expression methods for 
RNA-Seq, to see what simulations they chose to do.  Some authors make 
available R code to reproduce their simulations.

I have no plans to add a simulation function to the public edgeR package. 
There are plenty of functions in R that readily simulate data, from any 
distribution that you like.  Anyone undertaking a simulation study is 
undertaking methodological research in statistical bioinformatics. 
Anyone qualified to undertake methodological research will have particular 
requirements for their simulation that are hard to predict in advance, and 
should be easily able to write simulation code for themselves, or so it 
seems to me.

Best wishes
Gordon


> Date: Thu, 1 Sep 2011 09:25:23 -0400
> From: Jim Silverton <jim.silverton at gmail.com>
> To: bioconductor at r-project.org
> Subject: Re: [BioC] edgeR
>
> Using edgeR can differentiate difference between Treatment and Control using
> a negative binomial assumption of the data. However, can edgeR be used to
> simulate data from the null?
> -- 
> Thanks,
> Jim.


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