The quest for genes associated with different phenotypes and diseases is a central theme of research in human genetics. Several human phenotypes/diseases including height, myopia, diabetes, and others show complex pattern of inheritance where many genes together with environmental factors influence the intensity of a phenotype or disease. The genome-wide association study (GWAS) has been successful in mapping loci associated to variety of human traits and diseases. GWAS test for association of single nucleotide polymorphism (SNP) markers with a phenotype of interest at the genome-wide scale, this hence requires a severe multiple testing correction for statistical significance (p-value < 5 × 10-8). A single marker association approach has limited power in finding disease susceptibility genes especially in situations where the risk architecture of a gene is defined by many SNPs rather than one or two. This issue can be addressed through gene-based association approaches where the combined (or mean) statistics of a number of markers in a gene is tested for association with the trait of interest. Gene-based association approaches are established as a complementary approach to GWAS.
It is well established that genes work in concert as molecular networks and cellular pathways to perform biological processes. Individual tests for association of a gene (or a SNP in a gene) with phenotype may have limited power in identifying interacting genes with small effects. SNP or gene-based tests are limited in how much they can reveal of the biological mechanism behind phenotypic variation. Pathway-based association approaches, where gene-based tests statistics are combined into gene sets and tested for association, have been developed to overcome these limitations. Each gene set represents a biological pathway or a process, based on prior biological knowledge. Various techniques have been proposed to perform pathway-based association tests on GWAS data, including ALIGATOR, MAGENTA, and INRICH. These packages show several limitations, which can be addressed by the development of a new pathway-based association approach.
The aim of this thesis is to demonstrate the analytical approaches (GWAS and post-GWAS) to identify genes and pathways associated with complex ophthalmology traits and report methods for gene- and pathways-based analysis of GWAS summary data. Chapter 1 introduces the basic principles and strategies involved in GWAS analysis. Further, it extensively reviews the popular methods available for gene-based and pathway-based association tests.
Chapter 2 describes an initial GWAS on corneal curvature, a phenotype of interest to diseases such as myopia and keratoconus, in Australians and further meta-analysis of two cohorts comprising of 1788 Australian twins and their families and 1013 individuals from a birth cohort from Western Australia. It also reports on replication in Australians of Northern European ancestry of the previously reported association of SNPs in PDGFRA and FRAP1 genes with corneal curvature in an Asian population.
Chapter 3 is a demonstration of GWAS and pathway-based analysis. It reports the first GWAS and meta-analysis for corneal astigmatism in two Australian cohort studies, total sample size ~2,700. This chapter describes the application of Pathway-VEGAS, an extension of the gene-based analysis tool Versatile Gene-based Association Study (VEGAS) program for pathway-based analysis.
Chapter 4 reports the successful application of VEGAS and Pathway-VEGAS to GWAS meta-analysis data. As part of the international glaucoma genetics consortium (IGGC), I participated in genetic analysis of the vertical cup disc ratio (VCDR) phenotype. This manuscript describes the meta-analysis of GWAS on a discovery sample of 21,094 individuals from ten cohorts of northern European ancestry and a replication sample of 6,784 individuals from four Asian cohorts. Gene-based and pathway-based tests were performed using the VEGAS and Pathway-VEGAS programs separately for northern European ancestry samples and Asian ancestry samples, and respective results were meta-analysed. This chapter demonstrates the advantages of using different approaches such as single marker GWAS, gene-based association approach and pathway based association approach.
Chapter 5 presents the VEGAS2 software. VEGAS is one of the most popular gene-based association software. VEGAS however has some limitations such as the inability to perform a gene-based tests on the X chromosome, dependence on HapMap2 data to model the correlation between SNPs and inflexibility in the selection of gene boundary. The VEGAS2 software is an extension of VEGAS, redeveloped and upgraded to overcome these limitations.
Chapter 6 presents the VEGAS2Pathway approach for pathway analysis of GWAS summary data. This chapter describes the shortcomings of many popular pathway-based association approaches, and how VEGAS2-Pathway overcomes these limitations. Further it demonstrates the application of VEGAS2Pathway on the endometriosis GWAS summary data.
The last chapter provides the contributions this thesis makes to the literature and discusses its implications, limitations and future directions. In conclusion, my thesis provides the methodological and analytical approaches to further our knowledge in human genetics.