Sickle cell disease is a monogenic blood disorder in which the clinical course and disease severity vary widely among patients. In order for physicians to make more informed decisions regarding the treatment and management of disease, it would be useful to be able to predict disease severity. We focus on two primary modulators of disease severity in sickle cell patients, hemolysis and fetal hemoglobin (HbF). This dissertation evaluates methodology to identify genetic variants associated with severity of sickle cell disease and develops new methodology of genetic risk prediction to predict disease severity in sickle cell patients based on levels of HbF. Hemolysis is a trait that is influenced by multiple correlated phenotypes (lactate dehydrogenase, reticulocytes, bilirubin and aspartate transaminase). There are several approaches to statistical analyses of multiple correlated phenotypes. The first part of this dissertation evaluates the use of principal component analysis (PCA) and compares it to the alternative approach of examining the results of multiple univariate phenotypes individually. We will focus on the question of if and under what conditions we gain more power using a summarized phenotype from PCA in a genome wide association study (GWAS) rather than conducting multiple individual GWAS. We find that the there is more power gained from the PCA approach when there is a strong intercorrelation between the phenotypes. The second part of this dissertation proposes a novel method of genetic risk prediction for continuous traits using an ensemble of genetic models. We aim to show through a simulation and prediction of HbF that the proposed method is more robust to the inclusion of false positives and yields more stable predictions than computing a GRS and 10 fold cross validation. The third part of this dissertation introduces a Bayesian-based clustering approach to produce clusters of sickle cell anemia patients based on their "predicted genetic profiles" of HbF. We then examine the genetic profiles of individuals in the extreme clusters to determine which genes contribute more prominently to the genetic profile so that we may potentially identify genes that are highly influential in the regulation of extremely high and low values of HbF.

Background: The successful determination and analysis of phenotypes plays a key role in the diagnostic process, the evaluation of risk factors and the recruitment of participants for clinical and epidemiological studies. The development of computable phenotype algorithms to solve these tasks is a challenging problem, caused by various reasons. Firstly, the term 'phenotype' has no generally agreed definition and its meaning depends on context. Secondly, the phenotypes are most commonly specified as non-computable descriptive documents. Recent attempts have shown that ontologies are a suitable way to handle phenotypes and that they can support clinical research and decision making. The SMITH Consortium is dedicated to rapidly establish an integrative medical informatics framework to provide physicians with the best available data and knowledge and enable innovative use of healthcare data for research and treatment optimisation. In the context of a methodological use case 'phenotype pipeline' (PheP), a technology to automatically generate phenotype classifications and annotations based on electronic health records (EHR) is developed. A large series of phenotype algorithms will be implemented. This implies that for each algorithm a classification scheme and its input variables have to be defined. Furthermore, a phenotype engine is required to evaluate and execute developed algorithms. Results: In this article, we present a Core Ontology of Phenotypes (COP) and the software Phenotype Manager (PhenoMan), which implements a novel ontology-based method to model, classify and compute phenotypes from already available data. Our solution includes an enhanced iterative reasoning process combining classification tasks with mathematical calculations at runtime. The ontology as well as the reasoning method were successfully evaluated with selected phenotypes including SOFA score, socio-economic status, body surface area and WHO BMI classification based on available medical data. Conclusions: We developed a novel ontology-based method to model phenotypes of living beings with the aim of automated phenotype reasoning based on available data. This new approach can be used in clinical context, e.g., for supporting the diagnostic process, evaluating risk factors, and recruiting appropriate participants for clinical and epidemiological studies.