Statistical analyses for beta-rarefaction output

Good afternoon!

I realize there is a post similar to this Beta-rarefaction diversity matrix export or mean distance?, but was hoping bringing it up again to see if anything has been changed since or could be produced. I have been using the script below:

qiime diversity beta-rarefaction
--i-table feat-table.qza
--i-phylogeny 2022.10.phylogeny.asv.nwk.qza
--p-metric weighted_normalized_unifrac
--p-clustering-method upgma
--m-metadata-file metadata_mouse.txt
--p-sampling-depth 100
--p-iterations 100
--p-correlation-method spearman
--p-color-scheme RdYlBu_r
--o-visualization Visualization/JK_w_norm_unif.qzv

My samples are so low biomass and are fairly unevenly sequenced, so I don't like to rarefy and loose samples or information. The plug-in above has been wonderful in visualizing the variability that can come from my samples. I appreciate the jackknifing technique since, as stated in multiple papers, rarefying once can lead to incorrect interpretation. I was wondering, though, if there was a way to use a "mean distance matrix" to perform the rest of my diversity analyses with. Essentially, I would like a matrix constituting of the means of each sample centroid displayed in the beta-rarefaction output PCoA so that I could do statistical analysis. Theoretically, the feature table has already been rarefied 100 times and distance matrices have been made for each iteration. I get that the code outputs temp files since it would be too much to store, but could it not output one final matrix that is a mean of all 100 matrices, for example?

As stated in the topic linked above, I had also thought of manually creating 100 rarefied tables and then making distance matrices for each of them. That would essentially mean I would have to put my computer through running the exact same process all over again (and my poor computer has already been a trooper, I don't want to strain it with more hours of work) just to get to the data one step before the PCoA output. I wanted to know if there was now a qiime2 plug-in that would do that for me or if beta-rarefaction could add that as an output.

Since I have already made the plots, is there any way to extract the distance matrix that made the PCoA plot? I would like to be able to make heat-maps and run stats on them.

Thank you so much for your time! I really appreciate the work that's been done to make these plug-ins in the first place and I'm excited to implement them in my research.


Hello @MarlydEMejia. We still do not have a built in way to calculate the mean distance matrix you are referring to. You will need to follow the steps indicated in the other post to get the raw data then calculate the mean matrix yourself.

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.