I’ve got a comparison between control vs. treatment at four different sampling time over two weeks in a mesocosm-scale experiment (n=3). I applied lefse analysis for this whole scenario considering the condition (control/treatment) as the first class and the time (0, 3, 7, 14) as the subclass. However I’m thinking to look at biomarkers per each sampling point specifically, for example, just looking at time 7 (ctrl vs. treat). It will generate different biomarkers of course when we compared all the sampling time.
Do you believe this approach a wrongly one? just to check that our main investigation is to understand how the treatment affect the bacterial composition over time.
Thank you in advance!!
I think you need a more sophisticated design than lefse can provide. I suspect your have repeated measures since the data on consecutive days are dependent on each other. If you pick one time (e.g. 3) and then look across the treatments, that could work with lefse.
Hi Pat, thank you for replying me.
Yes, it was performed as a repeated measure design since the samples were taken from the respective mesocosms (those mesocosms for treatment and control one, being n=3) at 0, 3, 7 and 14th day.
What test do you suggest?
Then, do you believe there is no problem if I pick it at a specific time and have a look about these biomarkers that were different between treat and ctrl?
Thank you once again
It probably would be ok to pick a day and look across the treatments. I would encourage you to seek out a local statistical consultant to help with designing a repeated measures test in R. That’s a bit beyond the type of support we can provide here
Thank you very much, Pat!!