[BioC] (Limma) different toptable results for the same dataset using 2 different designs

Celine Carret ckc at sanger.ac.uk
Tue Dec 6 12:37:47 CET 2005


Dear gordon, thank you very much for your answer.
However, when I do that, I end up with the table below:

>>Here is what I've done:
>> > design2 <- cbind(A=1, BvsA=c(0,0,1,1))
>> > fit2 <- lmFit(eset, design2)
>> > fit2eB <- eBayes(fit2)
>> > toptable(fit2eB, n=10)
>>    
>>
>
>topTable(fit2eB, coef=2)
>
>Gordon
>
>             M          t   P.Value          B
>311  -3.997113 -13.961019 0.5794648 -0.5112598
>1327 -1.461801 -11.334987 0.5794648 -0.6889117
>113  -1.690073 -10.880814 0.5794648 -0.7308602
>4408 -3.066882 -10.019535 0.5794648 -0.8232774
>1825 -3.576034  -9.781223 0.5794648 -0.8523026
>1224 -1.099785  -9.445264 0.5794648 -0.8961448
>289  -2.800736  -9.306995 0.5794648 -0.9152559
>288  -1.689312  -8.759499 0.5794648 -0.9977157
>2892 -2.513426  -8.675603 0.5794648 -1.0113879
>3311  3.005392   8.377018 0.5794648 -1.0625077
>  
>
It does solve one problem, which is obatining 2 similar toptables 
whatever the design lay-out, but it doesn't explain why I have those 
awfully wrong values...
Would have an idea by any chance?

All the best,
Celine



More information about the Bioconductor mailing list