Mapping Nucleotide Sequences that Encode Complex Binary Disease Traits with HapMap

Yuehua      Cui; Wenjiang      Fu; Kelian      Sun; Roberto      Romero; Rongling      Wu

Abstract

Detecting the patterns of DNA sequence variants across the human genome is a crucial step for unraveling the genetic basis of complex human diseases. The human HapMap constructed by single nucleotide polymorphisms (SNPs) provides efficient sequence variation information that can speed up the discovery of genes related to common diseases. In this article, we present a generalized linear model for identifying specific nucleotide variants that encode complex human diseases. A novel approach is derived to group haplotypes to form composite diplotypes, which largely reduces the model degrees of freedom for an association test and hence increases the power when multiple SNP markers are involved. An efficient two-stage estimation procedure based on the expectation-maximization (EM) algorithm is derived to estimate parameters. Non-genetic environmental or clinical risk factors can also be fitted into the model. Computer simulations show that our model has reasonable power and type I error rate with appropriate sample size. It is also suggested through simulations that a balanced design with approximately equal number of cases and controls should be preferred to maintain small estimation bias and reasonable testing power. To illustrate the utility, we apply the method to a genetic association study of large for gestational age (LGA) neonates. The model provides a powerful tool for elucidating the genetic basis of complex binary diseases.

Keywords: Nucleotide sequence, complex disease, EM algorithm, logistic regression, haplotype

« Previous Next »