Different Beta Diversity Results on the Same Samples

[Sierra_Jaramillo]

Hi All,

I recently used QIIME2 to run beta diversity analyses on fecal samples. Because we had 4 sequencing runs over a course of a year I used multiple versions of QIIME2(QIIME2-2018 through QIIME2-2019.10) for demultiplexing, filtering and combining feature tables, and running phylogenetic core metrics (labeled as Try1 in screenshots). I recently went back to re-run all analyses in QIIME2 version 2019.10 (labeled as Try2 in screenshots). However, I am getting a very different result in Beta diversity compared to my previous analysis. I have attached photos here of beta diversity results and the provenance files. In try 1, day 25 samples are clustering with the other days. In try 2, day 25 samples are clustering out separately from all other samples. I do not think I did anything differently in the second try, but do you see any major changes that I may have overlooked? Any ideas on why these are so different?

[colinbrislawn (Colin J Brislawn)]

Good morning Sierra,

@Sierra_Jaramillo

Any ideas on why these are so different?

I wonder if the ordination plot is making things look more different than they are. Because in beta diversity, all the samples are measured against all other samples, even small changes can make the final graph look different, though underlying clustering is similar.

Can you share some alpha diversity graphs as well? These should not be changed by addition of new samples.

Also, I think some of the defaults changed between qiime-2018 and qiime-2019.10. Would you be willing to post some of the summary stats and maybe the full artifacts so we can take a look at providence?

Thanks,
Colin