Protml

Thorsten Burmester thorsten at erfurt.thur.de
Mon Jun 14 12:47:28 EST 1999


Luis wrote:
 
> While checking the .doc files from the PHYLIP package I came across with
> the documentation file of a pascal program called PROTML (Maximum
> Likelihood Inference of Protein Phylogeny) by Jun Adachi and  Masami
> Hasegawa which is not included in the package, can ayone tell me how to
> get this program?

You need to compile it yourself. PROTML.PAS is included in the package,
but you need to find someone with a Pascal-compiler. Perhaps it is a
good idea to use PUZZLE for the ML analysis instead. 
 
> By the way, another question, does anyone know how to get branch lenght
> estimates using the parsimony or maximum likelihood methods in th PHYLIP
> package?

PROTPARS does not estimate branch lengths because the number of
evolutionary events in a particular branch can not be *exactly*
calculated. I do not know about PROTML and DNAML, but IIRC at least
DNAML does estimate branch lengths. PAUP provides such values, but IIRC
implements other assumptions not included in the parsimony method, and
the branch lengths are actually something like the "mean values". 

If you prefer the PHYLIP PROTPARS assumption (I do), I would suggest to
re-run the analysis with FITCH, invoking the "user tree" constraints
(see PHYLIP manual). This will give you the branch lengths of your
original PROTPARS tree. 

HTH

Thorsten

-- 
Thorsten Burmester
thorsten at erfurt.thur.de




More information about the Mol-evol mailing list