Genetic Association Mapping Based on
Discordant Sib Pairs:
The Discordant-Alleles Test
Michael Boehnke and Carl D. Langefeld
Department of Biostatistics, University of Michigan, Ann Arbor
American Journal of Human Genetics,
62:950-961 (April 1998)
Abstract
Family-based tests of association provide the opportunity to test
for an association between a disease and a genetic marker. Such tests
avoid false-positive results produced by population stratification, so
that evidence for association may be interpreted as evidence for
linkage or causation. Several methods that use family-based controls
have been proposed, including the haplotype relative risk, the
transmission-disequilibrium test, and affected familybased controls.
However, because these methods require genotypes on affected
individuals and their parents, they are not ideally suited to the study of
late-onset diseases. In this paper, we develop several family-based
tests of association that use discordant sib pairs (DSPs) in which one
sib is affected with a disease and the other sib is not. These tests are
based on statistics that compare counts of alleles or genotypes or that
test for symmetry in tables of alleles or genotypes. We describe the
use of a permutation framework to assess the significance of these
statistics. These DSP-based tests provide the same general
advantages as parent-offspring triobased tests, while being
applicable to essentially any disease; they may also be tailored to
particular hypotheses regarding the genetic model. We compare the
statistical properties of our DSP-based tests by computer simulation
and illustrate their use with an application to Alzheimer disease and the
apolipoprotein E polymorphism. Our results suggest that the
discordant-alleles test, which compares the numbers of nonmatching
alleles in DSPs, is the most powerful of the tests we considered, for a
wide class of disease models and marker types. Finally, we discuss
advantages and disadvantages of the DSP design for genetic
association mapping.
Association Studies