Question

What is difference between classic and GLM edgeR

1

Entering edit mode

adrianna-christina ▴ 10

@adrianna-christina-7353

Last seen 9.8 years ago

Australia

Hi,

I have a RNA-seq experiment data and I want to extract differentially expressed miRNAs for the data. The data is from two groups(cell type1 and cell type 2; both have biological duplicates) and I want to know which miRNAs are differentially expressed between these two groups.

Which approach of edgeR should I use?

The classic one with exactTest function or the GLM approach. What's the difference between the exactTest and GLM results?

Thank you for your reply

r edger de • 8.9k views

ADD COMMENT • link 9.9 years ago adrianna-christina ▴ 10

0

Entering edit mode

Thanks a lot James and Aaron!!

ADD REPLY • link 9.9 years ago adrianna-christina ▴ 10

score 2 · Answer 1 · 2015-02-11

2

Entering edit mode

James W. MacDonald 67k

@james-w-macdonald-5106

Last seen 4 days ago

United States

The edgeR User's manual describes the differences between the two tests, in a relatively heuristic manner, starting on page 5:

http://bioconductor.org/packages/release/bioc/vignettes/edgeR/inst/doc/edgeRUsersGuide.pdf

But the take-home message is that you should use exactTest() if you only have two groups, and fit a GLM if you have a more complicated design.

ADD COMMENT • link 9.9 years ago James W. MacDonald 67k

3

Entering edit mode

Of course, the GLM approach will work just fine with the two-group design. Even though exactTest works for such designs, using GLMs has the benefit of allowing access to other functionality within edgeR, e.g., quasi-likelihood methods, fold change testing with TREAT. I generally use the GLM approach as my default, as it means I can avoid having to change my analysis pipeline for new datasets and designs.

ADD REPLY • link 9.9 years ago Aaron Lun ★ 28k