> In article <1994Apr26.193206.1597 at honte.uleth.ca> std_dickout at hg.uleth.ca writes:
>>I am taking a course in cladistics and am interested in information
>>concerning the use of Maximum Parsimony and Distance analyses in the
>>treatment of molecular data. Any reasons why you would choose one
>>method over the other would be greatly appreciated.
The generally accepted methods have been outlined by Bill Pearson
earlier. The note to add to this discussion is the paper by Hillis
and colleagues in the most recent Science (4/29). In this paper, the
authors use combinations of simulations and know phylogenies to test
the performance of UPGMA, NJ (Both distance methods) and Parsimony.
The results show that weighted parsimony has the best performance
curve, but in the tests used, all methods can work.
So it is a philosophical and practical issue as to which method or
methods to use. Some have said that they do not trust any single
method, and that situations where significantly different results are
obtained, make to problem too difficult to approach.
I myself am heartened as we have used a form of weighted parsimony we
call "Accepted Mutation Parsimony" (AMP) to look at a complex 125
taxon problem. This is a protein sequence method which utilizes a
weighting scheme derived from the aligned dataset.
What do other people think about the MP vs NJ vs ML problem?
Thanks
--
Frank Kolakowski
=======================================================================
O Email: lfk at receptor.mgh.harvard.edu O
O kolakowski at helix.mgh.harvard.edu O
O lfk at eastman1.mit.edu O
O US Mail: Lee F. Kolakowski Renal Unit, 8th Flr. O
O Massachusetts General Hospital Charlestown Navy Yard O
O 149 13th Street FAX : 1-617-726-5669 O
O Charlestown, MA 02129 Phone AT&T: 1-617-726-5666 O
=======================================================================
O The home of the G-Protein-Coupled Receptor Database (GCRDb) O
=======================================================================
