[BioC] Limma design and contrast matrix question.

Thornton, Matthew Matthew.Thornton at med.usc.edu
Fri Jan 24 18:37:40 CET 2014

Hi Dario!

Thank you for your reply. I am glad that I have a proper design matrix. I will try the 'Group1 -Group2 -Group3 + Control' contrasts.matrix. I am not very experienced with general linear models. So let me guess why that would be an appropriate contrast matrix, is it because the null hypothesis would be the mean of Group1 = mean Group2 + mean Group3 - mean Control?  So if I am interested in comparing Group2 to Control, I should have included 'Group2 + Control' in the contrasts.matrix

I really appreciate your help.

Thank you again!



Matthew E. Thornton

Research Lab Specialist
Saban Research Institute

USC/Children’s Hospital Los Angeles
513X,  Mail Stop 35
4661 W. Sunset Blvd.
Los Angeles, CA 90027-6020

matthew.thornton at med.usc.edu

From: Dario Strbenac [dstr7320 at uni.sydney.edu.au]
Sent: Thursday, January 23, 2014 5:00 PM
To: Thornton, Matthew; bioconductor at r-project.org
Subject: RE: Limma design and contrast matrix question.


Your design matrix is correct in the group means style. It is not the only correct way to do it, though.

How you tested for the interaction effect of the two treatments is wrong. The contrast should be Group1 - Group2 - Group3 + Control. I also don't think your Group1 - Control, Group3 - Group1 and Group2 - Group1 contrasts are meaningful.

Dario Strbenac
PhD Student
University of Sydney
Camperdown NSW 2050

More information about the Bioconductor mailing list