Path: utzoo!utgpu!news-server.csri.toronto.edu!bonnie.concordia.ca!nstn.ns.ca!news.cs.indiana.edu!bionet!crc.ac.uk!mbishop
From: mbishop@crc.ac.uk (Martin Bishop)
Newsgroups: bionet.molbio.evolution
Subject: Some thoughts on what to do
Message-ID: <4413.9101251737@crc.ac.uk>
Date: 25 Jan 91 17:37:11 GMT
Sender: daemon@genbank.bio.net
Lines: 39

WHAT TO DO NEXT IN MOLECULAR PHYLOGENY RESEARCH

It is artificial to separate an alignment step and a phylogeny
program (eg. dnapars) step in making phylogenetic reconstructions
as has already been pointed out by another contributor.

We attempted to make some contribution in this direction a few
years ago:
M.J.Bishop, A.E.Friday and E.A.Thompson
Inference of evolutionary relationships.
In M.J.Bishop and C.J.Rawlings 1987.
Nucleic acid and protein sequence analysis
a practical approach. IRL Press, Oxford.

I dont think it is referenced in the Phylip documentation,
so that is why people may be unaware of it.

Even more fundamental is the question of whether the simplistic
models of molecular evolution which these programs use can
be justified in the face of suspected processes of molecular
evolution such as gene conversion (proven processes in some
organisms - fungi).  More worrying needs to be done about the
appropriateness of the models.

I would suggest turning the problem on its head. Instead of trying
to estimate trees and times from sequence data take a group
for which a plausible tree and times can be written down.
Now write a program to tell us about the most likely pathways of
sequence change and relate these to functional constraints on a
variety of groups of macromolecules.

I think a grant committee should be prepared to fund some work along
these lines, at least to see if there is any mileage in it.
Regretably, I am too busy doing other things to have a go at it myself.

Worry more about the processes by which molecular sequences change and
less about getting the most parsimonious (but incorrect tree).

Martin Bishop.