smallm at PBS.DFO.CA wrote:
> Has anyone had conflicting results when comparing GENEPOP to CHIHW.
> Has anyone had populations out of equilibrium test out within
> equilibrium with CHIHW?
I did a number of simulations (with one and more loci) where
conditional probability test was consistently exhibiting slightly
greater power than conditional X2. However, my simulations were
limited to the case of drift with migration, and equal starting allele
frequencies. It is easy to generate a particular case, where either
Cond_X2 or Cond_Pr would give much smaller p-value than the other.
In Genetica, 96:169-178 we have a table illustrating the opposite
case, where Cond_X2 p-value is about 0.01, but Cond_Pr is 1.
Together with CHIHW
(ftp://brooks.statgen.ncsu.edu/pub/zaykin/chihw.exe)
I have another program, that performs a test for disequilibrium for
many loci (so that HW is its special case). The test is described in
Zaykin D., L. Zhivotovsky, B.S. Weir. 1995. Exact tests for
association between alleles at arbitrary number of loci. Genetica,
96:169-178, or in Bruce Weir's book (GDA 2, 1996). It does both
conditional X2 (like CHIHW), and the test based on the conditional
probability 1/Cond_Pr (like GENEPOP). The program can analyze each
locus separately, all pairs at once, triples, etc. When each locus
contributes a little to HWD, the power is greatest if all loci are
treated simultaneously.
Self-extracting executables are on:
ftp://brooks.statgen.ncsu.edu/pub/zaykin/xaloci16.exe [Win3.1]
ftp://brooks.statgen.ncsu.edu/pub/zaykin/xaloci32.exe [Win95]
Source code is available upon request. It is portable enough to be
compiled under UNIX without modifications.
(I'm finishing putting a similar code into the GDA package maintained
by Paul Lewis: http://biology.unm.edu/~lewisp/gda.html)
Dmitri