Identical p-values from Classical test and Limma analysis for paired data

Dear MetaboAnalyst Support Team,

I am analyzing targeted lipidomics data using the Statistical Analysis (one factor) module in MetaboAnalyst. My data have a paired design (samples collected from the same subjects before and after treatment).

When I perform the analysis using both Classical test and Limma, I obtain exactly the same results, including identical p-values for all lipid features.

I would like to know whether this is expected for paired data, or whether there is any specific setting required for applying Limma in a paired analysis.

Thank you very much for your help.

Best regards,

Lee

Thank you for the note.

Paired design is currently implemented only for classical t-tests. The support has now been added for limma (using block design).

Please wait for the server update (this weekend) and try again.

Hello Xia Lab,

Thank you very much for your reply and for adding the paired design support for limma.

I tried the analysis again after the update, but the results are still exactly the same as before. The p-values obtained from the Classical test and Limma remain identical for all lipid features.

For the analysis, I selected:

Statistical Analysis → One factor → Samples in rows (paired)

and used the paired sample design option.

Could you please confirm whether this is the expected behavior for paired analysis, or whether there is any additional setting required to enable the block design implementation for Limma?

Thank you very much for your help.

Best regards,
Lee

Dear Lee,
I have updated MetaboAnalyst server with the issue fixed. Please double-check and let’s know.
Bests,
Zhiqiang

This topic was automatically closed after 12 days. New replies are no longer allowed.