| The purpose of this dissertation is to evaluate, modify, and develop bioinformatic tools that can be applied to extended pedigrees for the localization of genes involved in complex diseases, specifically focusing on Tourette Syndrome (TS) and Autism Spectrum Disorder (ASD). Three analytically techniques are examined -- linkage, association, and shared genomic segments (SGS). With respect to linkage, MCMC methods were applied to a large TS pedigree to conduct a parametric linkage analysis. This was the first time that a linkage analysis was performed on the full pedigree in its entirety. We found linkage peaks for a qualitative analysis of TS on chromosome 3p and for a quantitative analysis of tic severity on chromosome 1p (LOD = 3.1 and 3.3, respectively). With respect to association, we developed a new weighting algorithm to perform association analyses in pedigrees. The algorithm considers all relationships simultaneously in arbitrary-structured pedigrees and assigns weights to pedigree members that can be used in subsequent association analyses to address relatedness. This new method outperformed a previous weighting approach. However, limitations were also evident. Further examination of the validity and power of this weighting approach, in addition to other association analyses, variance correction (VC) and a naive (ignore relatedness), was performed. The major findings were that: the weighting method was valid only when independent controls were used; the naive analysis could be conservative or anti conservative based on how cases and controls were selected; and power was increased with independent controls for small to moderate selectively ascertained pedigrees, but for randomly ascertained pedigrees and extended pedigrees controls selected from within the pedigrees increased the power. The findings will help the design of association studies using pedigrees. Finally, the power of SGS approach to identify dominant rare risk variants was examined. We amended the SGS procedure to assess each specific marker position in the genome. It was found that for a single extended pedigree the power of this SGS statistic ranged widely conditional on the underlying disease model considered. But, that ten or fewer pedigrees was sufficient to gain >80% in the majority of models. This power clearly out-performed genome-wide association studies for detecting rare risk variants. We implemented SGS in the analysis of a large high risk ASD pedigree and identified genomewide significant sharing on chromosome 1q. In conclusion, this dissertation illustrates the depth in versatility of extended pedigrees in genetic research and provides insight on ways for researchers to better use extended pedigree resources for gene-finding. |
