Development of bioinformatic tools and application of novel statistical methods in genome-wide analysis

A genome-wide association study is a method to discover potential genes involved in complex traits by scanning the entire genome. These studies traditionally employ brute force over sophistication: very simple analytic methods are applied to very large samples. In this thesis, we examined methods to improve the quality of these studies. This was accomplished in three ways. Firstly, we developed software that enables automatic and comprehensive quality control of the results of such studies. Secondly, we employed survival analysis, which allows more accurate modelling of both time-to-event phenotypes (here: the age at which someone starts using cannabis) and of biomarkers with a limit of detection. Thirdly, we used a new, more comprehensive genome reference, yielding 10 new genes associated with kidney disease. The second part of this thesis aimed at dissecting trait heritability. To this end, we investigated in 32 complex traits (such as height, blood pressure and cholesterol level) the amount of heritability that can be attributed to known genetic variants, and the upper limit that can be explained by all common genetic variants. This revealed, amongst other things, that even up-to-date genetic predictions of complex traits are not yet accurate enough to be used for personalized predictive medicine. Furthermore, we attempted to elucidate the heritability of neuroticism by analysing it in a variety of novel ways. In conclusion, the methods developed and applied in this thesis allow for more sophisticated genome-wide analysis and analysis of heritability.