[BioC] edgeR: design matrix

Gordon K Smyth smyth at wehi.EDU.AU
Mon May 28 11:02:42 CEST 2012


Dear KJ Lim,

You seem to have posted almost the same question to this list on three 
separate occasions a week apart, and it isn't clear whether you've taken 
much notice of the help you got from Mark Robinson the first time.  So it 
is indeed troubling to us guys.

Rather than trying to design a complete analysis for you, can I refer you 
to advice I gave to another poster recently:

https://www.stat.math.ethz.ch/pipermail/bioconductor/2012-May/045700.html

I suggest that you try using the "longer method", explained in the second 
half of my message to that poster, to setup your design matrix, after 
which you will be to draw any comparisons between the treatments that seem 
sensible to you.  Of course you will need to adapt it somewhat to your 
data.  This is probably better than trying to use model formula in R that 
might not be clear to you.

Best wishes
Gordon

> Date: Sat, 26 May 2012 14:06:29 +0300
> From: KJ Lim <jinkeanlim at gmail.com>
> To: bioconductor at r-project.org
> Cc: bioconductor at stat.math.ethz.ch
> Subject: [BioC] edgeR: design matrix
>
> I'm sorry to trouble you guys. I have a doubt about my design matrix.
>
> I have RNA-seq data for 2 different genotype of trees with
> 0hour(control) and after treatment 3hours,24hours,and 48hours. The
> experiment design like following:
>
>                    Treatment
>  Tree H1  Ctrl  3hrs  24hrs  48hrs
>  Tree H2  Ctrl  3hrs  24hrs  48hrs
>  Tree L1  Ctrl  3hrs  24hrs  48hrs
>  Tree L2  Ctrl  3hrs  24hrs  48hrs
>
> I have assigned 2 factor vectors as:
>  "tree" --> vector for the trees.
>  "trtTime" --> vector for the control and after treatment time.
>
> I would like to study which genes/tags that are differential expressed
> in these H and L trees across the after treatment time points.
>
> Can I assign my design matrix in this way:
>
>  design <- model.matrix(~trtTime+tree) OR design <-model.matrix(~tree+trtTime)
>
> I may wrong in this case as I'm not a statistician nor R programming
> geek. Thus, could someone kindly please light me?
>
> I appreciate very much for your help.
>
> Best regards,
> KJ Lim

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



More information about the Bioconductor mailing list