Joe Felsenstein joe at
Wed Jan 24 05:47:16 EST 1996

In article <4e18ub$1dqu at>,
Ron Kagan  <rkagan at> wrote:
>I have a question for any of you who are familiar with the use of maximum
>likelihood analysis in Felsenstein's PHYLIP package:  How should I set
>the "categories" option for defining how many categories of substitution
>sites there are and the relative rates for each?  I'm doing a DNAML and
>DNAMLK analysis of coding regions, so I was thinking that there should be
>2 categories of rates, one for the synonymous substitutions and one for
>the non-synonymous substitutions.  An average value for this ratio (for
>42 genes, Nei, 1987, p.80) is about 5.3:1.  Should I use this ratio, or
>should I empirically determine what it is for my sequences.  If I decide
>to determine this, what program might I use?

In principle one can determine the relative rates in the two regions by
maximum likelihood for the particular gene, by doing a series of DNAML runs
with different ratios of the rates (only the ratio matters, not the absolute
rate which gets rescaled) and choosing that ratio which achieves the highest

Actually, in DNAML 3.5 (the current release) one cannnot pre-assign
categories to sites.  But in directory "tempdir" on our ftp area
(on there is a pre-alpha version of
DNAML 4.0 which can do it (sorry, C source code only, no executables yet).

Joe Felsenstein         joe at     (IP No.
 Dept. of Genetics, Univ. of Washington, Box 357360, Seattle, WA 98195-7360 USA

More information about the Mol-evol mailing list