Relative rate test

[Joe Felsenstein]

In article <v01510104ae142735e266@[192.107.185.160]>,
Lee Weigt <weigt at FMPPR.FMNH.ORG> wrote:
>In article <4sgfu9$mcv at nntp3.u.washington.edu>,
>joe at evolution.genetics.washington.edu (Joe Felsenstein) writes:
>
>> But however that is to be done in a likelihood framework, the RRT has
>> more problems as it cannot tell you how to combine all the
>> three-taxon tests.
>
>In article <muse at kurtz.bio.psu.edu> writes
>>Agreed. BUT, that does not imply that doing many or all pairwise
>>comparisons is a useless thing to do. And, in fact, many of the tests
>>_are_ independent (this can be argued along the lines from Felsenstein's
>>1985 (?) article on independent contrasts).
>
>Which tests are independent?  When performing a multiple test correction,
>like the sequential Bonferroni, how do we decide what constitutes a family
>of tests that needs to be analyzed collectively?  Should a family be
>defined as all those tests that have two similar taxa (outgroup and one
>ingroup)?  Or should we combine all tests with a similar outgroup, which
>for 50 taxa and an alpha of .05 would make a "table-wide" alpha value of
>.00004!

Take one triple of taxa and draw the branches connecting them.  Then do the
same for another triple of taxa.  If there is any branch shared between them
(not just a point but a branch) then they are correlated.  How much they
are correlated could be worked out with a statistical model.

But it is easier just to do the Likelihood Ratio Test rather than do that --
it takes all correlations into account properly, subject to the adequacy
of the its statistical model of sequence change.

-- 
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 USA