[BioC] LIMMA design question

Prasad Siddavatam siddavatam at gmail.com
Fri Apr 8 19:59:00 CEST 2011


Hello Group,

I am analyzing individual patient arrays and need some help on the following 
questions.  

1. If I have a dataset with four types of patients (disease1, disease2,
disease3, healthy) and want to get the differential genes between disease 
and healthy. I can get the contrasts between healthy and disease 
comparisons using contrasts in LIMMA. But here each disease has 2-3 
arrays from different patient blood samples. These are not true
replicates, per se but thats all we have. My question is how do we 
treat them? Can I consider them as replicates? (I understand LIMMA will 
combine the replicate data.)
here is the design........
1. disease1 patient1
2. disease1 patient2
3. disease1 patient3
4. disease2 patient4
5. disease2 patient5
6. disease3 patient6
7. disease3 patient7
8. disease3 patient8
9. healthy patient1
10. healthy patient2
11. healthy patient4

2. In the above scenario, If I want differentially expressed genes in 
disease1 and disease2 combined w.r.t to healthy, is the following is 
correct way in LIMMA?
makeContrats(dis1&dis2 <- "(disease1 + disease2) /2 - healthy", 
                   levels = design);

If I want differentially expressed genes in all three diseases combined
w.r.t to healthy 
makeContrats(dis <- "(disease1 + disease2 + disease3) /3 - healthy", 
              levels = design);

please let me know is it the right approach.

Greatly greatly appreciate your help,

Sincerely

Prasad



More information about the Bioconductor mailing list