Limma Interaction Effect with Covariates
1
0
Entering edit mode
kitzhu • 0
@e03acc02
Last seen 5 days ago
United States

Hi Everyone, Im working on creating a comparison of ( Group1_T - Group2_T ) - (Group1_NT -Group2_NT ) to identify the DMRs that are found between my treatment comparison (Group1_T and Group2_T), and are not found in my control comparison (Group1_NT and Group2_NT). However, Im wondering if its possible to include factorial and continuous covariates in this, as my groups are not perfectly sex or age matched, and I want to include blood cell composition PCs.

Group <- factor(targets$Group)
Treat <-  factor(targets$Treatment)
Sex <- factor(targets$Sex) # covariates
Age <- as.integer(targets$Age)
PC1 <- as.numeric(targets$PC1)
PC2 <- as.numeric(targets$PC2)
PC3 <- as.numeric(targets$PC3)

# My incorrect attempt at a standard interaction effect with a 2x2x2 design, plus the five covariates:
design <- model.matrix(~(Group*Treat)+Sex+Age+PC1+PC2+PC3, data=targets) 
design

which gives me this design:

 (Intercept) PopGroup2 TreatT Sex Age          PC1         PC2          PC3      PopGroup2:TreatT
1         1     1   2  18 -0.806896196  0.10589096 -0.154088271       1
1         0     0   2  24  2.206815782  0.19763909  1.546195905       0
1         0     0   2  19 -2.111827930  0.28779334  0.312779390       0

I also tried to create this matrix an alternative way, where I created a Group_Treatment variable with the four levels: Group1_T, Group2_T, Group1_NT, Group2_NT. However, Im not sure if my covariates are still being factored into my "Interaction" comparison.

Group_Treatment <- factor(targets$Group_Treatment)
design<- model.matrix(~0+Group_Treatment+Sex+Age+PC1+PC2+PC3, data=targets)

contr <- makeContrasts(Interaction=(Group1_T -Group2_T)-(Group1_NT - Group2_NT), 
    levels = design)

myAnnotation <- cpg.annotate("array", object=mVals, what = "M",
    arraytype = "EPICv2",  analysis.type="differential", 
    contrasts = TRUE, cont.matrix = contr, design = design,
     coef="Interaction",  epicv2Filter = "mean")

Is my second attempt actually the correct comparison that Im looking for, with the 5 coefficients factored in? Or am I approaching this incorrectly?

Any help would be appreciated!

limma DMRcate • 129 views
ADD COMMENT
0
Entering edit mode
@james-w-macdonald-5106
Last seen 4 days ago
United States

The second way is how you should do it.

0
Entering edit mode

Thanks so much, James!

ADD REPLY

Login before adding your answer.

Traffic: 770 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6