Confused about the design matrix in edgeR: How to tell R to consider one level as a reference and the other is a treatment to be compared to that ref.
1
0
Entering edit mode
Mohamed ▴ 30
@aa1ae679
Last seen 15 months ago
United Kingdom

This question relates to my previous post (Advice on designing an EdgeR object in complex experimental setup ?)

I have an experimental design like in this image it aims to compare treated Vs control, adjusting for animals (sex + individual) l

I first created a data frame for the two groups (Animals) and (Status):

Animal <- factor(paste(Sample_info$Individual, Sample_info$Breed, sep = '.'))
Status<-factor(Sample_info$Status, levels = c('Treated', 'Control'))

Then I made the 'design matrix'

design <- model.matrix(~Animal+Status)

which produced a design matrix that looks like this design matrix.

My question is why the design matrix named its last column ' StatusControl' and not 'statustreated' ? this caused it to label the control samples as 1 and treated as 0, which I assume that this told the program to consider the treated groups as a 'reference' and control group as 'comparison' group so that all my genes (which in the count was higher in treated samples than in control) appeared as downregulated ? It should be the reverse..... So I want to tell R to consider the control as a 'reference group' and I think label it as 0 (not 1). How can I do that ?

This was my DE code :

fit<- glmQLFit(y, design, robust = TRUE)
tr <- glmQLFTest(fit, coef = "StatusUnstimulated")    

I would be appreciating any answer !

R edgeR • 2.5k views
ADD COMMENT
1
Entering edit mode
@steve-lianoglou-2771
Last seen 22 months ago
United States

My question is why the design matrix named its last column ' StatusControl' and not 'statustreated' ?

It's because the Intercept is the first level of the factor. If you define your factor like so:

Status <- factor(Sample_info$Status, levels = c('Control', 'Treated'))

You'll get what you're after.

ADD COMMENT
0
Entering edit mode

Thanks a lot, will try this and see.

ADD REPLY
1
Entering edit mode

When I answered your original question, I assumed you would use the simplest possible code

Status <- factor(Sample_info$Status)

which would have given the correct result. By default, the factor levels are ordered in alphabetical order, and "Control" comes before "Treated" alphabetically.

ADD REPLY

Login before adding your answer.

Traffic: 572 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