paired samples

Theory behind mothur
1

Well, here is a question.

I have samples, sampled at different time points.

So I am wondering if it is ok to do the analysis like AMOVA and LEFSE on all my samples pooled together since some of them are paired and I do not think that these are good for paired samples.

On the other hands, they can also be considered different “conditions”.

In Mothur, my reflex would have been to split the shared file into 2 different shared file, one for each time, and do 2 separate analysis, one for T1 and one for T2 but one student in my lab has put some doubts in my mind about it. It is also a bugger if I want to know if my samples in different conditions have evolved the same way over time.

Any advice, apart from “go learn R”?

2

These more complex designs require more complex stats. I’m afraid that it means going to learn R. The vegan package has a more powerful version of amova (i.e. permanova) that i think will allow a paired analysis.

Pat

3

Alright thank you for the answers.

So Vegan-R it is!

not too sure how I feel about that!