PHYLIP versus PAUP?

Joe Felsenstein joe at evolution.genetics.washington.edu
Wed Dec 3 13:06:29 EST 1997


In article <6636p2$o02 at net.bio.net>,  <newsmgr at merrimack.edu> wrote:
>2. If I draw(gram) a Consense tree from PHYLIP it doesn=92t have any sense.
>However, if I have used SEQBOOT-DNADIST-NEIGHBOR and than rearranged a neig=
>hbor
><<treefile.>> with the Retree (midpoint rooting) the resulting tree has a l=
>ot
>of sense. The question is if I can transfer bootstrap values from a consens=
>e
>tree (SEQBOOT-DNADIST NEIGHBOR-CONSENSE) to a retreeted neighbor tree?

I think that if the one doesn't make send but the other does, then you need
to understand unrooted trees and what they mean.  The trees coming out of
CONSENSE in this case are unrooted, and are printed as if rooted in an
arbitrary place (species #1).  You can move the root anywhere you want
that will make you happy -- these PHYLIP programs are not specifying
where the root is.

The bootstrap values go along with the branches.  Thus if a group of three
species seems to have a 90% bootstrap value, it is actually the branch that
partitions the species in that group of 3 versus all the others that has the
90% bootstrap value.  When you reroot the tree you will find that this branch
is still there -- there is still a branch that separates those 3 from all the
others -- and it can maintain the 90% boostrap value.

I should add that the "branch lengths" on the consensus tree are just the
bootstrap values. To use it and get branch lengths that are more meaningful,
you can feed it into Fitch as a User tree (option U).

I'll let some experienced PAUP user answer your other question.  I am sure
PAUP (and PAUP*) do have boostrap values available.

-- 
Joe Felsenstein         joe at genetics.washington.edu
 Dept. of Genetics, Univ. of Washington, Box 357360, Seattle, WA 98195-7360 USA




More information about the Mol-evol mailing list