[BioC] Design matrix in Combat algorithm

[Johnson, William Evan]

Matei, 

X is a design matrix for your covariates of interest or possible confounding factors. If the covariates are categorical with k levels, then the columns of X will contain k-1 indicator variables for the different factor levels. If this still does not make sense, I would STRONGLY recommend you go talk to a statistician at your university and have her/him help you analyze your data.

Thanks!

Evan



On Jan 22, 2014, at 6:00 AM, <bioconductor-request at r-project.org>
 wrote:

> Date: Wed, 22 Jan 2014 02:30:33 -0800 (PST)
> From: "Neagu Matei Stefan [guest]" <guest at bioconductor.org>
> To: bioconductor at r-project.org, m.neagu at sheffield.ac.uk
> Cc: sva Maintainer <jleek at jhsph.edu>
> Subject: [BioC] Design matrix in Combat algorithm
> Message-ID: <20140122103033.A57C7143D7F at mamba.fhcrc.org>
> Content-Type: text/plain; charset=iso-8859-1
> 
> 
> 
> I am currently trying to use ComBat from sva package in order to eliminate the batch effect from some Ilumina data obtained in an experiment. I have read the article Adjusting batch effects in microarray expression data
> using empirical Bayes method,W. EVAN JOHNSON, CHENG LI???,ARIEL RABINOVIC but I did not understand very well how the elements in matrix X of the model Yijg = ??g + X??g + ??ig + ??ig??ijg are obtained.
> Could you please give me a better insight on this subject?
> 
> Thank you very much,
> Neagu Matei Stefan
> 
> -- output of sessionInfo(): 
> 
> R version 3.0.2 (2013-09-25)
> Platform: x86_64-w64-mingw32/x64 (64-bit)
> 
> locale:
> [1] LC_COLLATE=Romanian_Romania.1250  LC_CTYPE=Romanian_Romania.1250   
> [3] LC_MONETARY=Romanian_Romania.1250 LC_NUMERIC=C                     
> [5] LC_TIME=Romanian_Romania.1250    
> 
> attached base packages:
> [1] stats     graphics  grDevices utils     datasets  methods   base  
> 
> --
> Sent via the guest posting facility at bioconductor.org.