Abstract
The field of genetics has witnessed a tremendous increase in the amount of information in the past decade resulting from genotyping and resequencing technologies. There has been an explosion in the development, application and optimization of statistical and computational methods to analyze these data. In particular, genome-wide scanning of common or rare genetic variants and their associations to a phenotypic trait has been successfully applied in several studies. For complex diseases such as Autism Spectrum Disorders (ASD), however, interrogating single locus associations by genome-wide association studies (GWAS) has been unable to unambiguously identify disorder causing genes. In this dissertation, an analysis was conducted on one of the largest known collections of family-based ASD GWAS datasets. Subsequently, in order to exploit the fullest potential of such datasets, a novel pathway analysis method specifically empowered to handle family-based GWAS datasets was developed and tested. Finally, data reduction methods employing classification and clustering algorithms were tested and proposed as a useful supplement to improve our understanding of the complex genetic and phenotypic architecture of an incredibly heterogeneous disorder, ASD.