A haplotype-based haplotype relative risk (HHRR) analysis of simulated
data for 200 affected offspring and their parents (Genetic Analysis W
orkshop 9, Problem 1) detected linkage disequilibrium at 2 of 360 mark
er loci. An additive model was suggested but not proven by haplotypes
of affected vs. unaffected offspring. These findings were consistent w
ith the generating model. Affected sib pair analysis failed to detect
additional loci. Discussion among workshop participants suggested that
the chi-square test used here (2 [transmitted vs. nontransmitted] x n
[alleles] for each locus) was invalid because of the nonindependence
of proportions of transmitted alleles. In post-workshop analyses, tran
smission disequilibrium tests (TDTs) for each allele at each locus det
ected only the true associations if p values were corrected by one of
two methods: Bonferroni correction for 2,035 TDTs, or correcting each
test for n (number of tests at the locus) minus 1 and then for the num
ber of loci tested. Screening loci for linkage disequilibrium requires
careful attention to correction for multiple comparisons. (C) 1995 Wi
ley-Liss, Inc.