I am using DeSeq2 in Macqiime and R studio to analyse differential abundance between my samples at OTU level. I would also like to examine changes between higher taxonomic levels eg. genus, family etc.
The following links suggests not using DeSeq2 to look at anything other than OTU level, which seems reasonable and suggests using Mann Whitney U. https://groups.google.com/forum/#!searchin/qiime-forum/using$20DeSeq2$20for$20class%7Csort:relevance/qiime-forum/13YdbVfEKUY/eNbO2Oh2sroJ However I was thinking I could still use it as an indicator.
My problem is that I have <10 samples per group and Mann Whitney U requires at least 20. Also when looking at levels higher than OTU is it still recommended to manipulate the data i.e using the negative binomial or log ratio?
My data is 16s data, collected from birds with different lesion scores (0, 1, 2, 3, 4) and uninfected controls, for each I have <10 biological replicates per group.
I have found a thread in the phloseq forum with someone asking to do this in R https://github.com/joey711/phyloseq/issues/683, however I am having problems replicating the data input in the first place (I have posted about this: https://github.com/joey711/phyloseq/issues/700).
I was also thinking I could use something like the group_significance test in Qiime, howver this is used to compare OTU frequency. I haven't come across a corresponding test for other levels, am I missing something simple?
Going round in circles with this, Cheers, Sarah