I have used q2-gemelli with my datasets because the view on feature rankings and feature loadings via q2-qurro were quite helpful with some datasets.
I have now seen the preprint of Pat Schloss at https://www.biorxiv.org/content/10.64898/2026.01.06.697977v2 raising some questions about Aitchison PCA and pointing to advantages of using rarefaction instead. I will certainly follow these discussions, but I do have two questions for now:
@cmartino : Is there an update of q2-gemelli available with an revised simulation code?
@gregcaporaso : Is q2-boots already able to perform rarefaction analyses as suggested by Pat Schloss? I assume yes via --p-replacement / --p-no-replacement, right?
Thank you all for your great tools! I would love to see that the best from different ‘worlds’ would merge sometime…
Yes, the methods in q2-boots were motivated by supporting rarefaction as described in thesetwo Schloss papers. This is discussed in the q2-boots paper. core-metrics is its highest-level action for traditional microbiome diversity metrics so is a good place to start. The --p-no-replacement supports rarefaction and --p-replacement supports bootstrapping.
I don't think the pre-print you're linking recommends any updates to the rarefaction procedure - but let me know if I'm missing that.
Correct, the pre-print does not mention q2-boots. Great to have q2-boots as a plugin, core-metrics is one of our standard analysis steps in metabarcoding analyses.
Glad it's been useful for you! We designed qiime boots core-metrics to be a drop-in replacement for qiime diversity core-metrics - it does of course take longer to run because it's doing a lot more work, but it should be very straight-forward to adapt your workflows to use it.
q2-boots core-metrics default output distances are bray-curtis, jaccard, weighted and unweighted unifrac. I prepared also Aitchison distance by q2-boots beta --p-n 100 -–p-metric ‘aitchison’, followed by q2-diversity beta-group-significance on the output to run permanova tests on Aitchison distance. The results are very interesting and I will investigate further.
Is there a way to add Aitchison metric to the core-metrics pipeline? I had to run q2-boots a second time to obtain Aitchison distance, and I am wondering whether there is a more easier way to use the resampled tables from the core-metrics pipeline?
Surprisingly, we are missing a corollary action that would take the collection of FeatureTable.
It wouldn’t be very hard to add additional metrics to core-metrics via a parameter (the outputs being collections help support this), but I do feel like there should be an action that would let you use those resampled-tables directly.
@gregcaporaso do you have any recollection here? It sure looks like we had intended to support this on some level, but there’s nothing that seems to take the output of boots resample or the tables from core-metrics.
Thank you for your comments. As described above, I found a way to calculate Aitchison distance, but q2-boots rarefaction takes some time to complete.
I have seen in some plugins under ‘Miscellaneous’ the –-parallel option. I have not yet found an explanation how to use this parameter. Is there somewhere a documentation/tutorial how to use this with q2 plugins (if available) and a linux machine with 12 to 40 cpus (not HPC)?
See the following docs for information on this, probably reading these in the order that I'm listing them. This functionality is a little under-documented at the moment - just let us know if you have questions. The Pipelines in q2-boots do support the --parallel option.