I am conducting an analysis of how a certain mouthwash affects the oral microbiome over the space of a month, taking samples at 3 seperate timepoints from 9 individuals. At each timepoint, samples are taken before and after the mouthwash.
When using the qiime longitudinal plugin and calculating pairwise differences and distances, is inter-individual variability eliminated if the ‘states’ (e.g. before and after mouthwash) in the analysis contain samples belonging to more than one individual? If so, how? Clarifying this would be extremely useful for the progression of the analysis.
Yes, if you are setting the --p-individual-id-column appropriately (i.e., to identify individuals).
It uses a paired statistical test (e.g., signed-rank wilcoxon by default) to examine how each individual changes between state 1 and state 2.
So setting the states to before/after mouthwash and setting individual-id-column to a sample metadata column that identifies individuals who are sampled before/after would be a very appropriate way to use this action!
You can check out the help documentation (e.g., qiime longitudinal pairwise-differences --help) to read more about that option and other options for adjusting which statistical tests are used.