P U Z Z L E 2.5: Announcement
15 September 1996
PUZZLE is (c) 1995-96 by Korbinian Strimmer and Arndt von Haeseler
Zoologisches Institut
Universitaet Muenchen
Luisenstrasse 14
D-80333 Muenchen
Germany
Voice: +49-89-5902-327
Fax: +49-89-5902-474
Email: strimmer at zi.biologie.uni-muenchen.de
Email: arndt at zi.biologie.uni-muenchen.de
INTRODUCTION
PUZZLE is a maximum likelihood program for reconstructing phylogenetic trees
from nucleotide and amino acid sequence data. It is available free of charge
over the Internet and runs on all popular systems. It is distributed by the
European Bioinformatics Institute:
ftp://ftp.ebi.ac.uk/pub/software/dos/puzzle (DOS version)
ftp://ftp.ebi.ac.uk/pub/software/mac/puzzle (MacOS version)
ftp://ftp.ebi.ac.uk/pub/software/unix/puzzle (UNIX version)
ftp://ftp.ebi.ac.uk/pub/software/vms/puzzle (VMS version)
PUZZLE is easy to install, easy to use, and implements a fast maximum likelihood
tree search algorithm (quartet puzzling) that allows analysis of large data
sets. Furthermore, this tree search automatically provides estimates of support
for each internal branch. PUZZLE also computes paiwise maximum likelihood
distances for all implemented models (TN, HKY, F84, and submodels; Dayhoff,
JTT, mtREV). It estimates the model parameters of the HKY/TN models (ratio
of transition-transversion ratio etc.) from the data. PUZZLE also estimates
maximum likelihood branch lengths for the quartet puzzling tree or a specifed
user tree (both for completely resolved and for multifurcating trees).
SUPPORTED FEATURES
Features include ($ marks features added in version 2.5):
* user menu PHYLIP "look and feel"
* analysis of both nucleotide and amino acid sequences
* wide range of models of sequence evolution
- nucleotides: Tamura-Nei 1993
Hasegawa et al. 1985
Felsenstein 1984 (PHYLIP DNAML model) $
and derivatives
- amino acids: Dayhoff 1978
Jones et al. 1992
Adachi and Hasegawa (proteins coded on mtDNA) 1995
- doublets: Schoeniger and von Haeseler 1994
* computes pairwise maximum likelihood distances for all models $
* automatic recognition of data type
* automatic choice of the optimal amino acid model
* deals with strictly bifurcating and multifurcating trees
* tree search is performed using quartet puzzling, a fast algorithm
based on quartet maximum likelihood trees that incorporates an
intrinsic replication scheme providing support values for each
internal branch. (Strimmer and von Haeseler 1996),
(Strimmer, Goldman, and von Haeseler 1996).
* as additional input a user tree can be supplied (can be multifurcating) $
* computes ML branch lengths along QP tree or a user tree $
* tree output in CLUSTAL W format, thus allowing to display branch lengths
and QP support values for each branch simultaneosly $
* check for homogenity of base composition between the sequences with
a chi-square test
* estimates TN/HKY model parameters (transition-transversion ratio etc.) $
* lists all unresolved quartets $
