I want to clarify difference between below two codes for analysis of DEGs. I got two different answers with these two methods and would like to know which one is correct. Kindly help in this regards.
2
0
Entering edit mode
devi • 0
@20d1be8a
Last seen 7 weeks ago
Germany

Enter the body of text here

Code should be placed in three backticks as shown below

# Differential expression test- code1
et <- exactTest(yf)
topTags(et)

##other
fitq <- glmQLFit(yf,design)
qlf <- glmQLFTest(fitq,coef=2)
summary(decideTests(qlf))

edgeR • 228 views
1
Entering edit mode
@steve-lianoglou-2771
Last seen 2 days ago
United States

This isn't actually answering your question, but in brief: let's assume that you are using edgeR to analyze count based data (like some flavor of RNA-seq), and that all of the upstream code is correct between the two examples you present, these days you should almost certainly prefer to use the QLF framework.

Here's a post I found from Aaron Lun that speaks about to the rationale for that, which is a more direct answer to your question.

0
Entering edit mode

Thank you so much for the answer

1
Entering edit mode
@gordon-smyth
Last seen 19 minutes ago
WEHI, Melbourne, Australia

The first method is the "classic" edgeR whereas the second is the newer quasi-likelihood approach. Assuming that you've applied both methods in the right way, then both methods are correct and will give similar results. However we recommend the quasi approach because it gives more rigorous error rate control. The difference between the two is explained in the edgeR User's Guide.

BTW, decideTestsDGE and decideTests are the same function. The first is retained in edgeR only to avoid breaking legacy code.

0
Entering edit mode

Thank you so much for the nice explanation!