[BioC] Question about F test in limma

Claus-Dieter Mayer claus at bioss.ac.uk
Mon Apr 28 18:44:29 CEST 2008


Dear Lisa,

no this is not a chance finding. Whether you  include all 3 possible 
contrasts or just the two you specify below: if all contrasts are zero 
this is equivalent to the overall null hypothesis that all 3 treatment 
means are the same, i.e. the F-test is testing the same hypothesis in 
both cases and thus gives the same p-value.

Best Wishes

Claus

Lisa Mijung Chung wrote:
> Dear limma expert,
>
> Thank you very much for your prompt reply. After I had sent you
> previous email, I tried with both methods just for my curiosity and
> got the same result. I am wondering if it happens to my data by chance
> or limma calculates moderated F stat by considering all targets
> specified on contrast matrix as different treatment groups, i.e. all
> RNA1, RNA2, and RNA3 appear on (anyways)
>   
>> makeContrasts(RNA2-RNA1, RNA3-RNA2)
>>     
> Thanks, again,
>
> Sincerely,
> Lisa C.
>
>
> On Wed, Apr 23, 2008 at 9:13 PM, Dejian Zhao <zhaodj at ioz.ac.cn> wrote:
>   
>> I think you should make all pairwise comparisons.
>>  Your H0 (no difference) means that all the targets show no
>>  difference. Therefore you should check them pairwise thoroughly.
>>
>>
>>
>>  On Thu, Apr 24, 2008 09:26, Lisa Mijung Chung wrote:
>>  > Dear limma expert,
>>  > I have a question about performing F-test (with several groups) on
>>  > limma.
>>  > If I have 3 targets (using Affy chips) called RNA1, RNA2, RNA3 and
>>  > want to get F statistic to test
>>  > H0: no difference vs. H1: at least one target is different,
>>  > (Similar with example on Section 8.6 of limma user guide at
>>  > pp.42-43)
>>  >
>>  > Should I construct my contrast matrix:
>>  >> makeContrasts(RNA2-RNA1, RNA3-RNA2, RNA3-RNA1) # with all pairwise
>>  >> comparisons
>>  >
>>  > or would it be fine only with:
>>  >> makeContrasts(RNA2-RNA1, RNA3-RNA2)            # comparisons of
>>  >> two adjacent pairs
>>  >
>>  > With best wishes,
>>  > Thank you.
>>  >
>>  >
>>  > --
>>  > Lisa Mijung Chung
>>  > Graduate Student, Biostat Trainee
>>  > Department of Statistics, UW-Madison
>>  > Web: www.stat.wisc.edu/~lchung
>>  >
>>  > _______________________________________________
>>  > 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
>>  >
>>
>>
>>  --
>>  De-Jian Zhao
>>  Institute of Zoology,Chinese Academy of Sciences
>>  +86-10-64807217
>>  zhaodj at ioz.ac.cn
>>
>>     
>
>
>
>   

-- 
***********************************************************************************
 Dr Claus-D. Mayer                    | http://www.bioss.ac.uk
 Biomathematics & Statistics Scotland | email: claus at bioss.ac.uk
 Rowett Research Institute            | Telephone: +44 (0) 1224 716652
 Aberdeen AB21 9SB, Scotland, UK.     | Fax: +44 (0) 1224 715349



More information about the Bioconductor mailing list