[BioC] edgeR: design matrix for different condition

[KJ Lim]

Dear Prof Gordon,

Good day. Thanks for your prompt replied.

Please allow me to rephrase my previous question.

This model,  ~tree+treat,  is assumed effect of the time of treatment
will be the same irrespective of genotype. Thus, test for the coef=3
will gives me the differential expression at 96H irrespective of
genotype.

I would like to learn the differential expression of treeHS vs treeLS
at specific time points i.e. 24H or if possible across the whole time
points. Should I fit my model as

  model A: ~tree*treat    OR   model B: ~tree+tree:treat ?

The design matrix columns of model B give:
  "(Intercept)"  "treeLS"  "treeHS:treat03H"  "treeLS:treat03H"
"treeHS:treat24H"  "treeLS:treat24H"  "treeHS:treat96H"
"treeLS:treat96H"

Am I doing right if I fit the model B and test for the coef=3:4 to
learn the differential expression of treeHS vs treeLS at specific time
points i.e. 03H? Does this test could tells me which set of genes
up/down-regulated in treeLS or treeHS?

I'm not good in statistic and I'm learning,  kindly please correct me
if I'm wrong.

Thank you very much for your time and suggestion.

Best regards,
KJ Lim




On 26 July 2012 03:39, Gordon K Smyth <smyth at wehi.edu.au> wrote:
>
> Dear KJ Lim,
>
> I don't quite understand your question, because you seem to be asking for something that isn't a test for differential expression, which is what edgeR does.  You question "I would like to learn the genes are express in treeHS but not treeLS and vice versa" seems to be about absolute expression levels.  edgeR doesn't test for genes that are not expressed in particular conditions.
>
> I'll refer you to the limma section on interaction models in case that helps, see Section 8.5 of:
>
> http://bioconductor.org/packages/2.11/bioc/vignettes/limma/inst/doc/usersguide.pdf
>
> Setting up a design matrix is the same for edgeR as it is for limma.
>
> Best wishes
> Gordon
>
>> Date: Tue, 24 Jul 2012 17:12:00 +0300
>> From: KJ Lim <jinkeanlim at gmail.com>
>> To: Bioconductor mailing list <bioconductor at r-project.org>
>> Subject: [BioC] edgeR: design matrix for different condition
>>
>> Dear the edgeR community,
>>
>> Good day.
>>
>> I'm analyzing my RNA-Seq experiment with edgeR. My study has 2
>> different genotypes (treeHS, treeLS) and time of treatment (0H, 3H,
>> 24H, 96H).
>>
>> I first assumed that the treatment effect will be the same for each
>> genotype and I have the design matrix as:
>>
>>  design <- model.matrix(~tree+treat)
>>
>>> design
>>
>>   (Intercept)  treat03H  treat24H  treat96H  treeLS
>> 1             1        0        0        0      0
>> 2             1        0        0        0      0
>> 3             1        1        0        0      0
>> 4             1        1        0        0      0
>> 5             1        0        1        0      0
>> 6             1        0        1        0      0
>> 7             1        0        0        1      0
>> 8             1        0        0        1      0
>> 9             1        0        0        0      1
>> 10           1        0        0        0      1
>> 11           1        1        0        0      1
>> 12           1        1        0        0      1
>> 13           1        0        1        0      1
>> 14           1        0        1        0      1
>> 15           1        0        0        1      1
>> 16           1        0        0        1      1
>>
>> I used coef=4 to test for the differential expressions between treeHS
>> and treeLS within the time of treatment, coef=3 to learn the
>> differential expressions in 2 genotypes at time of treatment 96H.
>>
>> ** I would like to learn the genes are express in treeHS but not
>> treeLS and vice versa at certain time of treatment let's say 24H or
>> across the whole time of treatment, should I have the design matrix as
>> below or more steps I need to do?
>>
>>  design <- model.matrix(~tree*treat)
>>
>> Kindly please light me on this. I appreciate very much for your help and time.
>>
>> Have a nice day.
>>
>> Best regards,
>> KJ Lim
>>
>
> ______________________________________________________________________
> The information in this email is confidential and inte...{{dropped:6}}