jwdebelius
May 19, 2025, 2:41am
2
Hi @takahitokawaguchi ,
There is a known disprepancy between QIIME 2 and the phyloseq weighted UniFrac implementation. The QIIME versions are tested against hte original pycogent implementations from Cathy Lozupone in 2005 and 2007; Im not sure about the phyloseq tests.
See
opened 07:00AM - 16 Jun 18 UTC
A couple of members of my lab and I have been getting very different results whe… n using qiime2 versus phyloseq for calculating weighted unifrac values. It seems that the default parameters used by these tools (qiime diversity core-metrics-phylogenetic versus phyloseq::distance) generate very different weighted unifrac values, even when pre-rarefying the dataset (since qiime2 automatically rarefies).
I've already posted this issue with a reproducible example with the `GlobalPatterns` dataset on the [qiime2 forum](https://forum.qiime2.org/t/very-different-weighted-unifrac-results-for-qiime2-versus-phyloseq/4591).
Any idea why there's such a big difference between the qiime2 and phyloseq defaults for calculating weighted unifrac values? Moreover, which method is generating "correct" weighted unifrac values? At least from analyzing my own microbiome dataset, the conclusions are completely different depending on if I use the qiime2- versus phyloseq-generated weighted unifrac values.
and
https://forum.qiime2.org/t/very-different-weighted-unifrac-results-for-qiime2-versus-phyloseq/4591/16
Best,
2 Likes