[BioC] duplicateCorrelation and design matrix

Carolyn Fitzsimmons Carolyn.Fitzsimmons at imbim.uu.se
Tue Jul 5 11:27:30 CEST 2005


Hello again Gordon,

> > But if I need to group the individuals differently to test for
> differential
> > expression between different groupings of individuals (i.e. between
> > WLm/WLf/RJm/RJf and WL/RJ), the use of 2 different design matrixies in the
> > dupCorrelation function is warrented, yes?
> 
> No.  Unless you have a good reason to do otherwise, set the full design
> matrix and use
> contrasts.fit() to group the individuals for differential expression tests.
> 
> Gordon
>

Then I would have to set a contrasts matrix for a comparison between WJ and RJ
like this:
    WL.RJ
RJf  -0.5
RJm  -0.5
WLf   0.5
WLm   0.5

Instead of this:
     WL.RJ
RJf   -1
RJm   -1
WLf    1
WLm    1

Because it I get inflated m-values with the second matrix. Is this what you
would do?

Regards,  Carolyn

-- 
Carolyn Fitzsimmons
Dept. Medical Biochemistry and Microbiology
Uppsala University
Box 597/BMC
SE-751 24
SWEDEN

E-mail: Carolyn.Fitzsimmons at imbim.uu.se
Tel: +46 (0)18 471 4593
Mobile: +46 (0)73 704 1248



More information about the Bioconductor mailing list