Mothur 1.27, possible unifrac.unweighted problem


I Think there may be a bug in the weighted unifrac command in mothur 1.27.0. With my data I found significant differences between all groups with both parsimony and unifrac.weighted. In order to check my results I randomized all my sequences with regard to the treatment groups, the parsimony p-values were all between .04 and 1.00 but the unifrac.weighted results all came back as highly significant. Maybe it is something I am doing wrong but the exact same group and tree files worked with parsimony. There are a lot of duplicate sequences in my file but isn’t that exactly what weighted is meant to deal with. Anyways, maybe someone should check it out.


Well, you know the three methods do three totally different things, right? If you can send your tree, name, and group files, we can take a look.

Note: the problem I am having is with unifrac.weighted, not unifrac.unweighted, my mistake. I did not test the unweighted unifrac.