I would like to thank all the people who responded to my original post of
one week ago, either here in public or through my private email address.
Thanks for all your careful descriptions of the respective algorithms
and for the references you pointed out to me. Thank you!
Here are my thoughts:
Joe Felsenstein is completely right if he says "My reading of Strimmer's
post is that he does NOT allow the tips to be renamed." because I am
not looking at trees from a mathematical point of view but from a
biological (phylogenetical) one. Joe Felsenstein's suggestion to
do multiple runs of his CONSENSE program with one of the sampled trees
and the target tree as input is in fact precisly the way I dealt with
the tree comparision problem up to now. His remark "But that is tedious"
indeed is very true, it takes me on a fast computer (Sparc 20/64 MB)
30 minutes to compare 1000 trees ... (slow I/O over the net). This can
only be (as James Foster writes) "a good heuristic workaround".
Ed Adams provided detailed algorithms to do direct NH-tree comparisons.
His very clear description seems to be quite straightforward to implement
(at least for an pointer-experienced programmer...). Finally I am happy
to
read that John Huelsenbeck has a functioning C routine designed precisely
for the purpose of tree comparison. The obstacle that "..the program is
not
very user friendly (I am currently the only user).." should be easily
removable. I haven't seen the code yet but if it is not too complicated
I would volunteer to add a menu from the sort of the PHYLIP menus to his
program to overcome this, so that everybody can do tree comparisons
with ease.
. .
Korbinian Strimmer ( ) ( )
|o|^|o|
| _ |
Laboratory of Molecular Evolution, University of Munich |_| |_|
---------------------------------------------------------------------
strimmer at zi.biologie.uni-muenchen.de
