How to see if changes of microbiota and compound over time are related

General Discussion
1

Hello,

I'am new to QIIME2. I want to test if changes in microbiota over time are related to changes in abundance of a certain compound which increases over time. I have samples taken at specific timepoints and also a column in my metadata file which indicates the abundance of the compound at those timepoints.

Is there any specific test I can use for this problem?

Thank you all!

2

Welcome to the forums, @Lexy,

Check out q2-longitudinal, especially the part about Linear mixed effects (LME) models:
https://docs.qiime2.org/2021.11/tutorials/longitudinal/#linear-mixed-effect-models

And of course, let us know if you have any questions!

2 Likes
3

Dear Colin,

Thank you for your response. I was reading the documentation and I see this line:
"observations are made across dependent samples, e.g., in repeated-measures sampling experiments"
In my case, there are no dependent samples; I have biological and technical replicates for different timepoints and also the abundance of a compound at that specific timepoint.

It is not clear to me what should I use as --p-individual-id-column: Time or Run column? Also, should I use the compound_abundance column as --p-metric then?

Thank you in advance

5

Hello Lexy,

Dig a little deeper into the notes and example provided in that section, and see if you can apply these to your study.

To start, the --p-state-column sounds like a good fit for your Time column.

The --p-individual-id-column tracks the same individuals/experimental units as they are sampled again and again over time, like if the same single :pig2: was sampled three times over three months. (Do you have samples from the same individuals/experimental units, or do you have totally different individuals at each timepoint, maybe due to destructive sampling? :bacon:)

You mentioned in your DM

If that compound was added to the samples by you, that's probably a fixed effect (input variable), but if it's being created by the samples it's probably a random effect (output variable). I think we need more context to understand that part of the experiment.

Specifically, I think we need to figure out what it the dependent variable that you expected to change based on your experimental inputs.