Testing Models against a tree (help)

Joe Felsenstein joe at evolution.genetics.washington.edu
Thu Jul 20 18:09:56 EST 1995


In article <1995Jul20.174815.39942 at ac.dal.ca>,  <aroger at ac.dal.ca> wrote:
>I have two alternative models of character state
>change that I wish to test
>
>Model A:
>1-->0 (where 1 is ancestral for all characters)
>
>Model B:
>1-->0 and 0-->1 (where 0 is ancestral for all characters)
>
>I can assume a tree which accurately describes the organismal
>relationships, what I wish to do is ask: which model better
>fits the data given the tree?  How should I do this using simple 
>methods (for instance using a parsimony framework)?

Inevitably, if one uses parsimony, model B will fit better (or at any rate no
worse) than model A.  That is because it allows all the events in model A.
One could use the Templeton "paired sites test" as to whether the sum of
changes in model A is significantly worse than in model B.

I am not sure this is doable in PHYLIP, however, as I think one cannot
change character model in MIX in mid-run.

-----
Joe Felsenstein         joe at genetics.washington.edu     (IP No. 128.95.12.41)
 Dept. of Genetics, Univ. of Washington, Box 357360, Seattle, WA 98195-7360



More information about the Mol-evol mailing list