relating variables to OTUs

Question

[Furor]

Hello all,

I don't know if this is the exact purpose of this forum, but perhaps you could redirect me.

I have an OTU x sample table and a sample x environmental variables table. I'm looking for ways to see which OTUs (or taxa) are significantly linked to environmental variables.

Taking this a step further, when I do a network analysis (co-occurrence) of my OTUs, I'd also like to see which environmental parameters are significantly associated or correlated with the observed clusters.

Anybody who could set me on my way? I'm basically doing my analysis in R, currently using phyloseq for the network analysis, but I'm stuck with the superimposing of the variables ...

Thanks!

Answer 1

[bioinfoSeeker]

You could add the environmental variables in the metadata file, and then produce the OTU table again. Then you can produce PCoA plots etc that reflect the environmental variables added.You can also produce subsets of OTU tables for those variables and specific samples and do diversity analysis on them separately, if required.

Regarding network (co-occurence) analysis, the R package Mixomics could be an option. But I'm unsure if they have yet extended it for 16S and metagenomics datasets. I have previously used it for RNAseq data.

Answer 2

[J06262016]

I used QIIME to analyze my OTU data and I believe you are looking for the following two scripts if you are interested in using QIIME. The tutorials are really good.

Below are exact descriptions from the QIIME website.

This script is used to compare OTU frequencies in sample groups and to ascertain whether or not there are statistically significant differences between the OTU abundance in the different sample groups. (http://qiime.org/1.9.0/scripts/group_significance.html)

This script allows for the analysis of the strength and statistical significance of sample groupings using a distance matrix as the primary input. (http://qiime.org/1.9.0/scripts/compare_categories.html)