[BioC] assist in constructing a design matrix & contrast matrix for Illumina microarray

[Wei Shi]

Hi,

    Just adding to Steve's comments that the limma userguide in the 
latest release ( Oct. 28) has a case study for processing Illumina 
BeadChip data now.

Cheers,
Wei

Steve Lianoglou wrote:
> Howdy,
>
> On Nov 11, 2009, at 10:43 AM, James W. MacDonald wrote:
>
>> Hi Maria,
>>
>> Maria Dolores Serafica wrote:
>>>   Dear Bioconductor Staff,
>>>   I sent this email on Oct 27 & would like to know if it was 
>>> received and if
>>>   someone can help.
>>>   I am analysing Illumina Beadchips (HumanWG6) for 42 SAMPLES.
>>>   Experimental design: (factorial) time course, samples grown on 4
>>>   media; have zero control sampled at onset only[ one way multigroup
>>>   analysis]; 2-3 reps per treatment.
>>>   I used the lumi script by Pan Du et al (2009) for preprocessing
>>>   and normalisation . This part  was okay. The samples were subsetted
>>>   and are R objects.
>>>   SAMPLE 1 REFERENCE SAMPLE  DAY 0 (triplicate)
>>>   SAMPLE 2 MEDIA A DAY 7 (triplicate)
>>>   SAMPLE 3 MEDIA A DAY 14 (triplicate)
>>>   SAMPLE 4 MEDIA A Day 30 (triplicate)
>>>   SAMPLE 5 MEDIA A DAY60  (singleton)
>>>   SAMPLE 6 MEDIA A DAY90 (replicates)
>>>   SAMPLE 7 MEDIA B DAY 7 (triplicate)
>>>   SAMPLE 8 MEDIA B DAY 14 (triplicate)
>>>   SAMPLE 9 MEDIA B DAY 30 (triplicate)
>>>   SAMPLE 10 MEDIA B DAY60 (singleton
>>>   SAMPLE 11 MEDIA B DAY 90 (TRIPLICATE)
>>>   SAMPLE 12 MEDIA C DAY 7 (REPLICATES)
>>>   SAMPLE 13 MEDIA C DAY 14 (REPLICATES)
>>>   SAMPLE 14 MEDIA C DAY 30 (REPLICATES)
>>>   SAMPLE 15 MEDIA D DAY 7 (REPLICATES)
>>>   SAMPLE 16 MEDIA D DAY 14 (REPLCATES)
>>>   SAMPLE 17 MEDIA D DAY 30 (REPLICATES)
>>>   SAMPLE 18 MEDIA D (DAY 90 (REPLICATES
>>>   This is a differentiation experiment.
>>>    I tried the lumi script using the modified limma (PanDu et al 
>>> 2009) based
>>>   on the Barnes subset data but the script does not apply to my 
>>> experimental
>>>   set up.  Can someone assist me in making a design matrix and a 
>>> contrast
>>>   matrix.
>>>   I am new to R and would like to use limma for identifying 
>>> differentially
>>>   expressed genes.  I have read the limma users guide(Gordon Smyth), 
>>> LAB4
>>>   limma 2005, James MacDonald's BioC 2009.
>>>   The contrasts (comparisons) I am looking for:
>>>   a. differential gene expression of each sample WITH the reference
>>>   b. early differentiation genes as affected by media, time and both
>>>   c. late differentiation genes as affected by media or both.
>>
>> It's good that you have done some homework and are trying to learn 
>> how to do the analysis, but in the end you are asking somebody to do 
>> your work for you (for free), and it's not likely that you will get 
>> many people who want to do that.
>
> Or, at the very least, you can show us how you're trying to do what 
> you're doing, and someone will likely comment on what's going awry.
>
> To be honest, I have no idea what reference Pan Du (2009) is, and 
> don't really have any desire to add that to my reading list, atm.
>
> So, why don't you start with an easy task. Specifically, by reading 
> the limma user guide, it *should* be pretty straight forward to 
> achieve (a) in your goals: finding differential expression of a 
> perterbation-sample vs. its reference is in there.
>
> To be even more specific:
>
> You should be able to find differential expression between your SAMPLE 
> 1 (reference triplicate) vs. SAMPLE 2 (media day 7 triplicate). I'm 
> guessing your code will look very similar to section 8.2.
>
> I recall at first I wasn't very comfortable using the 
> limma::modelMatrix function, and I I would create my own design matrix 
> by hand. Once you have that (with appropriate column names), building 
> the contrast matrix should be easy rather straightforward.
>
> So .. have fun with that. I'd imagine that might take a while (the 
> limma guide is *big*), so I guess we'll get your follow up questions 
> sometime tomorrow :-)
>
> And, like I said, the more specific you are about what's troubling you 
> for this specific task (getting differential expression between 
> sample2 and sample1, for example), the greater the likelihood you'll 
> get an informative answer.
>
> Good luck,
>
> -steve
>
> -- 
> Steve Lianoglou
> Graduate Student: Computational Systems Biology
>   |  Memorial Sloan-Kettering Cancer Center
>   |  Weill Medical College of Cornell University
> Contact Info: http://cbio.mskcc.org/~lianos/contact
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives: 
> http://news.gmane.org/gmane.science.biology.informatics.conductor