Entering edit mode
3.0 years ago
Palgrave
▴
110
I am doing limma on a matrix with 150 samples, three groups and 50 samples in each group. If I do limma on group1 vs groups2 by inputting the whole matrix I get different results than If I input a matrix that includes only the measurements for the two groups I am interested in?
Why does the limma results differ if I remove a set of measurements that is not used in the contrasts? Is it due to normalisation?
It makes sense to me, but, please share the exact code that you've used.