[BioC] design matrix and lmFit problems in limma

pingzhao Hu phu at sickkids.ca
Thu May 13 23:23:56 CEST 2004


Hi,

I have an Affymetrix dataset. There are 24 arrays.
There are three factors: time, sex (M and F) and strain (NOD,R and D4).
Sex has two levels and strain has 3 levels. I am only interested in
interaction between sex and strain. So the formula will like:

~time+sex*strain.

The question I want to  ask is

" Does the effect of strain change between male and female?"

For example I want to select genes:

(1) differentially expressed genes between male and female only at NOD not 
R and D4?
(2) differentially expressed genes between male and female only at R not 
NOD and D4?
(3) differentially expressed genes between male and female only at D4 not 
NOD and R?



I construct design and contrast matrix as follow.


 > design
       maleNOD maleR maleD4 femaleNOD femaleR femaleD4
  [1,]       1     0      0         0       0        0
  [2,]       1     0      0         0       0        0
  [3,]       0     1      0         0       0        0
  [4,]       0     1      0         0       0        0
  [5,]       0     0      1         0       0        0
  [6,]       0     0      1         0       0        0
  [7,]       0     0      0         1       0        0
  [8,]       0     0      0         1       0        0
  [9,]       0     0      0         0       1        0
[10,]       0     0      0         0       1        0
[11,]       0     0      0         0       0        1
[12,]       0     0      0         0       0        1
[13,]       1     0      0         0       0        0
[14,]       1     0      0         0       0        0
[15,]       0     1      0         0       0        0
[16,]       0     1      0         0       0        0
[17,]       0     0      1         0       0        0
[18,]       0     0      1         0       0        0
[19,]       0     0      0         1       0        0
[20,]       0     0      0         1       0        0
[21,]       0     0      0         0       1        0
[22,]       0     0      0         0       1        0
[23,]       0     0      0         0       0        1
[24,]       0     0      0         0       0        1



 > contrast
           maleNOD-femaleNOD maleR-femaleR maleD4-femaleD4
maleNOD                   1             0               0
maleR                     0             1               0
maleD4                    0             0               1
femaleNOD                -1             0               0
femaleR                   0            -1               0
femaleD4                  0             0              -1


Are these correct????


-----------------------------------------------------------

in lmFit there are three methods to fit linear models (gls,ls and robust).
Since affy. data has no duplicate spots for each gene on the same array, Is 
it still
suitable to use gls method???.  The example in user manual at 7.2
used gls (default).

The big question is that the results based on gls and robust methods are
very different:

For example, the number of differentially expressed genes based on these two
mthods are aso follows (p-value<=0.01):


           maleNOD-femaleNOD        maleR-femaleR      maleD4-femaleD4
gls             7                       7                  14
robust         14                      229                 500

How can I decide which method (gls or robust) to use?

Thanks
pingzhao



More information about the Bioconductor mailing list