I am trying to examine the change in abundance of OTUs along an elevational gradient. I have four 100m intervals of the gradient, with two sites per 100m interval. For example, sampled two sites around 700m, two sites around 800m, etc.
I am having a difficult time trying to find a way to fit linear models to this dataset in metagenomeseq.
A couple thoughts I had:
1) Make the elevation variable a factor, and bin intervals together (e.g., two sites around 700 m binned as 700 m) and do pairwise comparisons using fitFeatureModel. I am going to post a separate question about multiple group comparisons using fitzig. I got some results that seemed off when I used it comparing a different categorical variable, and I'm not very trusting off it right now.
2) Extract normalized counts and fit linear models on each OTU.
3) Figure out some way to use fitTimeSeries. I just briefly looked as this. I doubt I can use it, but thought since time is a continuous variable there is someway to alter the code to use it. But, I suppose what may be happening with the model is accounting for repeated measures over time, which wouldn't really help my case.
Any insight would be appreciated.