I have a question regarding alpha group significance testing and the possibility to carry it out without setting a sampling depth (or another similar way to do so). I know this way of doing is not advised at all but I have a trial with a very specific configuration and I would like to answer a particular question.
For one sample group (treatment), we recovered a very low number of sequences (average ~ 40 reads, which equals ~ 5 different ASVs), which was expected. Whereas for other groups, the amount of reads and corresponding ASVs were much higher (~ 80,000 reads and ~80 ASVs, respectively).
If I follow the traditional workflow to carry out the alpha group significance testing, I should rarefy my data by setting the 'p-sampling-depth' parameter of the
qiime diversity core-metrics-phylogenetic command to ~ 40 reads (but it does not make sense to analyze diversity in other groups with only 40 seqs/samples whereas several tens of thousands are available) or to a much higher number of reads (but I will thus loose the information regarding samples in the first group).
My idea is to show that, in addition to raw numbers that are pretty obvious (40 reads vs 80,000 reads, etc.), there are also statistical evidences showing a very strong shift in diversity within the first group of samples. To do so, I would need for instance to statistically compare between groups the total number of reads (or ASVs, or faith_pd, or evenness, or ...) when all reads are taken into account in every sample. However, I didn't find a way to do it with
qiime diversity core-metrics-phylogenetic since a sampling depth is mandatory.
Does anyone have an idea on how I could do it?
Thanks in advance,