FMT Tutorial Question 2c

[YinXun]

Is this pattern stronger based on unweighted UniFrac or Bray-Curtis distance? Based on how you know about these metrics, what does this suggest to you about what is changing in the microbiome with fecal microbiota transplant?

I found a stronger pattern for the Bray-Curtis distance compared to unweighted UniFrac. Unlike unweighted UniFrac, I am aware that Bray-Curtis does not take phylogeny into account. But I am struggling to understand what this suggests about the change in microbiome with the FMT treatment. Any help please? Thank you!

[jwdebelius (Justine Debelius)]

Hi @YinXun,

If you know that Bray-Curtis also takes abundance into account with unweighted UniFrac doesn't, does that help you answer the question?

I find this post really helpful when considering alpha and beta diversity

Alpha and Beta Diversity Explanations and Commands

About the Alpha and Beta Diversity Analysis Tutorial This Alpha and Beta Diversity Community Tutorial (run using QIIME 2017.12) walks you through analyzing the alpha and beta diversity of a sample dataset. Below you will find a link to a small test dataset to download and use in this tutorial. Files used in tutorial The following files, derived from the Moving Pictures tutorial, are used in this document. table.qza rooted-tree.qza Alpha Diversity Analysis The alpha and alpha-phylogenetic …

(I'm also really surprised at your results, are you sure they're correctly labeled? How do they look compared to the tutorial results?)

Best,
Justine

PS I moved this to the general discussion, since it's more theory about metrics.

[colinbrislawn (Colin J Brislawn)]

Like Justine suggested, Bray-Curtis is a waited metric (uses abundance values) and Unweighted UniFrac is a unwaited metric (uses only presence-absence values).

Because the main difference is weighted vs unweighted, this is driving the finding. In the weighted metric, the most abundant microbes will contribute more to the differance. In the unweighted metric, all microbes contribute equally, i.e. less abundant microbes will be measured more by the differance.

@YinXun

I found a stronger pattern for the Bray-Curtis distance compared to unweighted UniFrac.

So there is not a strong phylogenetic pattern (unweighted UniFrac), but the composition of more abundant microbes does change between groups (Bray-Curtis).

Colin

P.S. Weighted UniFrac is more comparable to Bray-Curtis as it's also waited. So when you compare them, Bray-Curtis vs Weighted UniFrac, the main question is composition vs phylogenetic composition. Now that's interesting!