Question

EdgeR two condition problem

0

Entering edit mode

bioinformatics ▴ 10

@bioinformatics-11531

Last seen 6.9 years ago

Hello !

I have used DESeq and EdgeR to get differentially expressed genes.

For DESeq it is possible to put a design with more conditions.

This is important because the data i used has

for example treated and untreated columns. But also differences in

cells types. And also the way of sequencing used differs. So there are

more things to account for then simple the differences of

treated and untreated data.

In DESeq this can be done by :

dds <-

DESeqDataSetFromMatrix(

countData = countdata,

colData = coldata,

design = ~ condition + cellType + MethodsUsed

)

In EdgeR the syntax is :

dataDGEList <-

DGEList(

counts = countdata,

group = factor(condition)

)

So the group sould account for the differences in EdgeR but

i might be mistaken. When reading the documentation of edgeRUsersGuide it sais

i have to do something with a glm model. But i do not understand the

rest of the manual and the approch used there. Could someone give me

an example of how to perform the de with EdgeR using three conditions :

condition + cellType + MethodsUsed?

edger • 623 views

ADD COMMENT • link updated 7.4 years ago by Aaron Lun ★ 28k • written 7.4 years ago by bioinformatics ▴ 10

score 1 · Answer 1 · 2016-12-01

If you want to use an additive design, you'll have to use the GLM framework. Your design matrix would look like:

design <- model.matrix(~ condition + cellType + MethodsUsed, coldata)

... which can then be used in glmFit, estimateDisp, etc. Comparisons between treatment conditions or cells types or methods can be performed by using makeContrasts to specify the coefficients that you want to compare. The edgeR user's guide contains more details about this, especially in the case studies. If there's something you don't understand, you'll have to be more specific if you want (good) help.