Search
Question: T-test in limma
0
gravatar for zhang.jianhai
3 months ago by
zhang.jianhai0 wrote:

Dear all,

I am confused about how t-test or F-test is implemented in limma/edgeR.

Let's say I have a small dataset: two conditions, each condition has 3 replicates, and there are 4 genes.

Con1 rep1 Con1 rep2 Con1 rep3 Con2 rep1 Con2 rep2 Con2 rep3
gene1          
gene2          
gene3          
gene4          

My questions is if I use limma/edgeR to analyze the DEGs across the two conditions, does limma first use the regular t-test to calculate if the mean of con1 is significantly different from the mean of con2 for each gene independently? And gets a regular t-statistic? Then use eBayes to adjust the regular t statistic and get the moderated t?

Look forward to your reply.

Regards.

ADD COMMENTlink modified 3 months ago by Gordon Smyth34k • written 3 months ago by zhang.jianhai0
0
gravatar for Gordon Smyth
3 months ago by
Gordon Smyth34k
Walter and Eliza Hall Institute of Medical Research, Melbourne, Australia
Gordon Smyth34k wrote:

No, that is not how it works.

If you want to know how the mathematics works, then read the relevant published papers. Type ?eBayes and look at the list of references. The relevant papers are either Smyth (2004) or Phipson et al (2016).

The bottom line though is that you don't need to understand any of the maths to use the software.

ADD COMMENTlink modified 3 months ago • written 3 months ago by Gordon Smyth34k

If it does not work like I said, is the DEG calculation process in limma/edgeR essentially based on t-test/F-test? If not, what test is the foundation?

Regards.

ADD REPLYlink written 3 months ago by zhang.jianhai0
2

Yes, the tests in limma are closely analogous to ordinary t-tests and F-tests and can be thought of as equivalent to them when you make conclusions.

That is all explained in the limma documentation and in the references that I listed. See for example Chapter 13 of the limma User's Guide. Even just reading the ?eBayes help page would give you some pointers as to the relationship between ordinary t-tests and moderated t-tests.

ADD REPLYlink modified 3 months ago • written 3 months ago by Gordon Smyth34k

Thank you very much, got it!

Regards.

ADD REPLYlink written 3 months ago by zhang.jianhai0
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.2.0
Traffic: 233 users visited in the last hour