joe@uw-evolution.UUCP (Joe Felsenstein) (01/11/89)
The person known as BIO252 asked about how to test spatial heterogeneity of mtDNA haplotypes. The method cited, by DeSalle, looks at one haplotype at a time over r populations, using an arc-sine transform and a between-populations weighted sum of squares. BIO252 is right, that this leaves us uncertain how to combine information from different haplotypes. Probably the simplest way, easier than what DeSalle did, is to simply make up a haplotypes x populations table, with n[ij] being the number of haplotype i observed in the sample from population j. The heterogeneity of frequencies across populations can then be tested by an (alleles x populations) heterogeneity chi-square (treating the table as a contingency table). There are (alleles-1)x(populations-1) degrees of freedom. I think that for the two-allele case this will get almost the same result as DeSalle did, and it is a correct way of combining information from multiple alleles. But notice what all these approaches leave out -- the genealogy of the haplotypes. We are treating them as alleles with arbitrary frequencies. There is more information there to be extracted by making a tree of haplotypes. No one has yet found a nice way to combine this genealogical information with frequencies of haplotypes. --- Joe Felsenstein, Dept. of Genetics SK-50, Univ. of Washington, Seattle WA 98195 BITNET: FELSENST@UWALOCKE INTERNET: uw-evolution!joe@entropy.ms.washington.edu UUCP: ... uw-beaver!uw-entropy!uw-evolution!joe