Faith_PD scale units

Hello, I am trying to write up some results for a paper, so I wanted to describe Faith_PD results... I have seen all papers describe it as dimensionless... however, this comes from a summation of branch lengths of a phylogenetic tree... Units' scale in a phylogenetic tree are usually "number of substitutions per site", so, the sum of these lengths should be same scale... is that true?

1 Like

I'm not sure... While the MSA can include a tree, it's often fed into another program using a reference tree, or a maximum likelihood calculation like FastTree2. After this step, I'm not sure what the units represent.

The units of Faith's PD and UniFrac don't seem to matter as much as the difference in branch length.

But maybe they matter to you! What are you hoping to use the branch length units to report? Maybe there's a better way we can get the unit you are looking for.

1 Like

Hello Colin, thanks for your answer... I am writing a paper and I want to be precise in my description, I mean, Faith_PD is described everywhere like dimensionless but based on the definition is not... so, I wonder someone here could know the units of the tree, which I guess should be kind of "number of substitutions per site"... or something like that...

Thanks anyway...

1 Like

Hi @Alejandro_Marquez, welcome to :qiime2:!

To echo @colinbrislawn point here:

Faith's phylodiversity is somewhat agnostic to the units associated with the branches (edge weights) of the tree. For example, if you performed maximum likelihood analysis then your PD value will be represented as "substitutions per site per unit branch length". If you used parsimony, your branch lengths might be the number of parsimonious changes (or there may not be any branch lengths). If you have a tree without branch lengths, i.e. only topological / hierarchical dendrogram, then the branches are often set to some arbitrary value like 1, and then summed to calculate phylodiversity. Ultimately, you define the units based on how the tree was generated.


Hey Mike thanks by your answer!! That's exactly what I want to know... and your comment was perfect to guide the discussion... I would like to know what method QIIME is using to build that tree... I used "qiime phylogeny align-to-tree-mafft-fasttree" to build the tree and after that "qiime diversity core-metrics-phylogenetic" with the rooted tree as input to get the Faith_PD values... Based in what you said I looked for MAFFT and found that this uses several methods such as NJ or UPGMA, but in any case, branches' lengths are in "substitution/site". So, based on that Faith_PD index is expressed also in "substitution/site".



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