X chromosome studies of Parkinson's disease and HapReg: A weighted regression program for family-based haplotype association studies

Parkinson disease (PD), the second most common neurodegenerative disorder, is caused by degeneration of dopaminergic neurons in the substantia nigra and other areas of the brain. Mutations in five genes have been identified to cause PD in a minority of families. Four independent linkage analyses provide compelling evidence for additional PD susceptibility genes on the X-chromosome. In this thesis, a linkage analysis for X-chromosome was conducted using 283 affected sib-pairs from the GenePD Study and modest evidence for linkage (multipoint LOD = 1.07) to PD affection status was found at 41 Mb. Thirty-seven SNPs for possible PD candidate genes located from 40 to 47 Mb were evaluated for association to PD using multiple methods. Significant evidence for association was observed for c6231643 in two independent samples: familial PD (p = 0.018) and random idiopathic PD ( p = 0.016). C6231643 is 257bp 3' of the type-1 protein phosphatase inhibitor 4 (I-4) gene. Presence of the C-A-G haplotype for three I-4 SNPs ( C2462683-C2462691-C6231643) is significantly increased ( p = 0.019) among men in the familial PD sample. As a potentially important regulatory subunit of protein phosphatase 1, I-4 may influence phosphorylation processes triggered by dopamine, resulting in the development of PD.; Haplotype analysis has become increasingly important for disease gene or quantitative trait locus (QTL) mapping. However, case-control studies are susceptible to spurious associations due to population admixture. In this thesis, a novel computer program HapReg has been developed to estimate the haplotype frequencies by an EM algorithm and evaluate haplotype effect sizes for quantitative traits by a weighted regression model using family data. Simulations showed that when parental genotypes were available, HapReg protected against population admixture and provided a robust test for haplotype effects. In a homogeneous population with missing parental genotypes, HapReg significantly improved the power for detecting association when compared to the popular family-based association software haploFBAT. HapReg allows testing using different genetic models, adjustment for non-genetic covariates, and X-linked marker analysis. HapReg was applied to the PARK3 region and identified a haplotype A-G-G associated with older age at onset of PD (p = 0.023).