Testing Association for a Quantitative Trait Locus Using Discordant Sibship Data

Schork et al. [2000] proposed an odds ratio test for association between alleles at quantitative trait locus and a single marker. The odds ratio test was based on sampling unrelated " case " and " control " subjects. Bur this test suffers from population mixture or stratification. Motivated by Risch and Zhang [1995] and Schork et al. [2000], we consider the sampling from discordant sibships and propose a simple test for QTL-marker allele association. The proposed test is not sensitive to population mixture or stratification since it is family-based. The test can be easily extended to multilocus case, and to involving in any sibships with at least one case sibling and one control sibling. Some statistical properties of the tests are studied by means of numerical calculations. We especially study how the choice of uper and lower percentiles, which define the discordant sibships, affects the power of the test. The results may be helpful in practice. Additional simulations show that the multiloci test can signifieantly improve the power. To compare the power of our test and that of Allison et al. [1999]. some simulations are performed. The simulation results show that our test performs better provided the uper and lower percenriles are properly chosen.