[BioC] Contrast, help

[Sean Davis]

On 1/29/04 2:16 PM, "Marcelo Luiz de Laia" <mlaia at fcav.unesp.br> wrote:

> My apologize for sending off topic to the list.
> 
> I, with the help of the user's guide and of the html_help, I set up the
> following contrast matrix:
> 
>> design
> time1 time2 time3
> 1     1     0     0
> 2     1     0     0
> 3     1     0     0
> 4     0     1     0
> 5     0     1     0
> 6     0     1     0
> 7     0     0     1
> 8     0     0     1
> 9     0     0     1
>> contrast.matrix <- makeContrasts(time2-time1,
>> time3-time2,time3-time1,levels=design)
>> contrast.matrix
>     time2 - time1 time3 - time2 time3 - time1
> time1            -1             0            -1
> time2             1            -1             0
> time3             0             1             1
> 
> Considering that my experiment relates levels of hidric stress, in other
> words, in the time zero there is no lack of water and, with the increase of
> the time, the shortage of water went increasing. Therefore, in the time x,
> larger than 3 (x>3),  the plants died for lack of water.
> 
> I can generate 3 tables of genes, one for the contrast 1, one for the 2 and
> one for the contrast 3.
> 
> My doubt is the following: can I say that the genes of the table 1 are the
> genes differentialy expressed in the time 1? Can I say that the genes of the
> table 3 are the genes differentialy expressed in the time 2? same for the time
> 3?

Actually, as I understand what you are doing, the gene lists associated with
the contrasts would represent the genes differentially expressed between
time 1 and time 2, time 2 and time 3, and time 1 and time 3.

Are these two-color arrays?  If so, then the gene lists you find by fitting
your data with the design matrix will give you the genes differentially
expressed between your "treated" group and your control group.  In other
words, the first column would represent genes differentially expressed at
time 1 (compared to the control), the second at time 2, and the third at
time 3.  

If they are not two-color arrays, then you probably want to decide on a
reference (perhaps the "no-lack-of-water" state at time 1 and your contrasts
would be between time 2 and time 1 and time 3 and time 1.

All of this depends, of course, on the biology and the hypotheses being
tested.

Sean

-- 
Sean Davis, M.D., Ph.D.

Clinical Fellow
National Institutes of Health
National Cancer Institute
National Human Genome Research Institute

Clinical Fellow, Johns Hopkins
Department of Pediatric Oncology
--