OBJECTIVE: Joint analysis of multiple SNP markers can be informative, but studying joint effects of haplotypes and environmental exposures is challenging. Population structure can involve both genes and exposures and a case-control study is susceptible to bias from either source of stratification. We propose a procedure that uses case-parent triad data and, though not fully robust, resists bias from population structure. METHODS: Our procedure assumes that haplotypes under study have no influence on propensity to exposure. Then, under a no-interaction null hypothesis (multiplicative scale), transmission of a causative haplotype from parents to affected offspring might show distortion from Mendelian proportions but should be independent of exposure. We used this insight to develop a permutation test of no haplotype-by-exposure interaction. RESULTS: Simulations showed that our proposed test respects the nominal Type I error rate and provides good power under a variety of scenarios. We illustrate by examining whether SNP variants in GSTP1 modify the association between maternal smoking and oral clefting. CONCLUSION: Our procedure offers desirable features: no need for haplotype estimation, validity under unspecified genetic main effects, tolerance to Hardy-Weinberg disequilibrium, ability to handle missing genotypes and a relatively large number of SNPs. Simulations suggest resistance to bias due to exposure-related population stratification. Copyright 2010 S. Karger AG, Basel.
OBJECTIVE: Joint analysis of multiple SNP markers can be informative, but studying joint effects of haplotypes and environmental exposures is challenging. Population structure can involve both genes and exposures and a case-control study is susceptible to bias from either source of stratification. We propose a procedure that uses case-parent triad data and, though not fully robust, resists bias from population structure. METHODS: Our procedure assumes that haplotypes under study have no influence on propensity to exposure. Then, under a no-interaction null hypothesis (multiplicative scale), transmission of a causative haplotype from parents to affected offspring might show distortion from Mendelian proportions but should be independent of exposure. We used this insight to develop a permutation test of no haplotype-by-exposure interaction. RESULTS: Simulations showed that our proposed test respects the nominal Type I error rate and provides good power under a variety of scenarios. We illustrate by examining whether SNP variants in GSTP1 modify the association between maternal smoking and oral clefting. CONCLUSION: Our procedure offers desirable features: no need for haplotype estimation, validity under unspecified genetic main effects, tolerance to Hardy-Weinberg disequilibrium, ability to handle missing genotypes and a relatively large number of SNPs. Simulations suggest resistance to bias due to exposure-related population stratification. Copyright 2010 S. Karger AG, Basel.
Authors: Stijn Vansteelandt; Dawn L Demeo; Jessica Lasky-Su; Jordan W Smoller; Amy J Murphy; Matt McQueen; Kady Schneiter; Juan C Celedon; Scott T Weiss; Edwin K Silverman; Christoph Lange Journal: Biometrics Date: 2007-10-25 Impact factor: 2.571
Authors: Min Shi; Kaare Christensen; Clarice R Weinberg; Paul Romitti; Lise Bathum; Anthony Lozada; Richard W Morris; Michael Lovett; Jeffrey C Murray Journal: Am J Hum Genet Date: 2006-12-07 Impact factor: 11.025
Authors: Miriam Gjerdevik; Øystein A Haaland; Julia Romanowska; Rolv T Lie; Astanand Jugessur; Håkon K Gjessing Journal: Ann Hum Genet Date: 2017-11-02 Impact factor: 1.670