Scalable Feature Selection Applications for Genome-Wide Association Studies of Complex Diseases

Personalized medicine will revolutionize our capabilities to combat disease. Working toward this goal, a fundamental task is the deciphering of geneticvariants that are predictive of complex diseases. Modern studies, in the formof genome-wide association studies (GWAS) have afforded researchers with the opportunity to reveal new genotype-phenotype relationships through the extensive scanning of genetic variants. These studies typically contain over half a million genetic features for thousands of individuals. Examining this with methods other than univariate statistics is a challenging task requiring advanced algorithms that are scalable to the genome-wide level. In the future, next-generation sequencing studies (NGS) will contain an even larger number of common and rare variants. Machine learning-based feature selection algorithms have been shown to have the ability to effectively create predictive models for various genotype-phenotype relationships. This work explores the problem of selecting genetic variant subsets that are the most predictive of complex disease phenotypes through various feature selection methodologies, including filter, wrapper and embedded algorithms. The examined machine learning algorithms were demonstrated to not only be effective at predicting the disease phenotypes, but also doing so efficiently through the use of computational shortcuts. While much of the work was able to be run on high-end desktops, some work was further extended so that it could be implemented on parallel computers helping to assure that they will also scale to the NGS data sets. Further, these studies analyzed the relationships between various feature selection methods and demonstrated the need for careful testing when selecting an algorithm. It was shown that there is no universally optimal algorithm for variant selection in GWAS, but rather methodologies need to be selected based on the desired outcome, such as the number of features to be included in the prediction model. It was also demonstrated that without proper model validation, for example using nested cross-validation, the models can result in overly-optimistic prediction accuracies and decreased generalization ability. It is through the implementation and application of machine learning methods that one can extract predictive genotype–phenotype relationships and biological insights from genetic data sets.

Epidemiological studies of childhood and adolescence headache

The main purpose of this study was to examine the changes in the prevalence, incidence, and characteristics of headache in childhood and adolescence. In addition, the predictors of the change in the occurrence of childhood headache and the association between adolescent headache and behavior were studied. The occurrence and characteristics of headache were investigated as part of a prospective follow-up study, where 6-year-old children and their families (n=1132) were followed to the age of 12-years (n=1126). The study design entailed both a cohort and case-control group. The incidence of headache and the association between headache and behavior were studied in another cohort, consisting of 11-year-old twins (n=5393), who were followed to the age of 17 (n=4159). The prevalence rates of headache increased during the follow-up, especially in boys whose mothers suffered from frequent headache. The incidence rates of frequent headache changed the most in girls between ages of 11 and 14. Early-onset migraine and tension-type headache were equal predictors of migraine at age 12. The age-related changes observed in pain localization, concurrent symptoms and triggers were considerable. Headache frequency was significantly associated with externalizing and internalizing problem behaviors and adaptive behaviors as rated by parents, but only with externalizing problem behaviors as rated by teachers. Headache both in children and adolescents is characterized by its changing nature. Its prevention and treatment should take familial, environmental and psychosocial aspects into account.

Brain imaging studies in severe somatization

Somatization was described 4000 years ago but the pathophysiology of the, phenomenon is unknown. The aim of this investigation was to explore whether central nervous system (CNS) pathology is associated with severe somatization which was operationalized as somatization disorder (SD) and undifferentiated somatoform disorder. The study sample consisted of severely somatizing people who were included into the study after a multi-phase screening procedure in order to exclude psychiatric comorbidities and physical illnesses. Diagnosis of somatization disorder or undifferentiated sofatoform disorder were set according to Diagnostic and Statistical Manual of Mental Disorders 4th ed. (DSM-IV). The first study explored the regional cerebral metabolic rate of glucose (rCMRGlc) in severely somatizing females and found it to be reduced in several regions of the brain compared to healthy controls. The second study observed brain morphology with magnetic resonance imaging (MRI) based on the findings from the first study and showed enlarged caudate nuclei in somatizing women compared to healthy volunteers. The third study investigated temperament factors and brain metabolism, and their association with severe somatization. Low caudate and putamen metabolism, low novelty seeking as well as high harm avoidance were found to be associated with severe somatization in women, reduced caudate metabolism having the strongest association. The last study is a report of man with left-side gradient of multiple symptoms of unknown origin in the body. The examination revealed a hypermetabolic nucleus putamen on the contralateral side. All the main results reported in these four articles are original findings. The results suggest that CNS pathology is involved in the pathophysiology of severe somatization.

Offshore Sourcing in Software Development: Case Studies of Finnish-Russian Cooperation

The study examines international cooperation in product development in software development organisations. The software industry is known for its global nature and knowledge-intensity, which makes it an interesting setting to examine international cooperation in. Software development processes are increasingly distributed worldwide, but for small or even medium-sized enterprises, typical for the software industry, such distribution of operations is often possible only in association with crossing the company’s boundaries. The strategic decision-making of companies is likely to be affected by the characteristics of the industry, and this includes decisions about cooperation or sourcing. The objective of this thesis is to provide a holistic view on factors affecting decisions about offshore sourcing in software development. Offshore sourcing refers to a cooperative mode of offshoring, where a firm does not establish its own presence in a foreign country, but utilises a local supplier. The study examines product development activities that are distributed across organisational and geographical boundaries. The objective can be divided into two subtopics: general reasons for international cooperation in product development and particular reasons for cooperation between Finnish and Russian companies. The focus is on the strategic rationale at the company level, in particular in small and medium-sized enterprises. The theoretical discourse of the study builds upon the literature on international cooperation and networking, with particular focus on cooperation with foreign suppliers and within product development activities. The resource-based view is also discussed, as heterogeneity and interdependency of the resources possessed by different firms are seen as factors motivating international cooperation. Strategically, sourcing can be used to access resources possessed by an industrial network, to enhance the product development of a firm, or to optimise its cost structure. In order to investigate the issues raised by the theoretical review, two empirical studies on international cooperation in software product development have been conducted. The emphasis of the empirical part of the study is on cooperation between Finnish and Russian companies. The data has been gathered through four case studies on Finnish software development organisations and four case studies on Russian offshore suppliers. Based on the material from the case studies, a framework clarifying and grouping the factors that influence offshore sourcing decisions has been built. The findings indicate that decisions regarding offshore sourcing in software development are far more complex than generally assumed. The framework provides a holistic view on factors affecting decisions about offshore sourcing in software development, capturing the multidimensionality of motives for entering offshore cooperation. Four groups of factors emerged from the data: A) strategy-related aspects, B) aspects related to resources and capabilities, C) organisation-related aspects, and D) aspects related to the entrepreneur or management. By developing a holistic framework of decision factors, the research offers in-depth theoreticalunderstanding of offshore sourcing rationale in product development. From the managerial point of view, the proposed framework sums up the issues that a firm should pay attention to when contemplating product development cooperation with foreign suppliers. Understanding different components of sourcing decisions can lead to improved preconditions for strategising and engaging in offshore cooperation. A thorough decisionmaking process should consider all the possible benefits and risks of product development cooperation carefully.

Metabolic syndrome in young adults – prevalence, childhood predictors and association with subclinical atherosclerosis

Background: Metabolic syndrome (MetS) is a cluster of cardiovascular risk factors including central obesity, insulin resistance, impaired glucose tolerance, hypertension and dyslipidemia. The prevalence of MetS is increasing worldwide in all age groups. MetS is associated with increased risk of cardiovascular disease and type 2 diabetes mellitus. Aims: The aim of the present study was to investigate the prevalence, secular trends and childhood predictors of MetS in young adults. Furthermore, the relations between MetS and subclinical atherosclerosis were studied and whether apolipoproteins (apo) B and A-I, C-reactive protein (CRP) and type II secretory phospholipase A2 (sPLA2) were associated with MetS, and to what extent the atherogenicity of MetS was explained by these factors. Participants and Methods: The present thesis is part of the large scale population-based, prospective study, the Cardiovascular Risk in Young Finns Study. The first cross-sectional study was conducted in 1980 and included 3,596 participants aged 3-18 years. Carotid and brachial ultrasound studies were performed for 2,283 of these participants in 2001 and 2,200 of these participants in 2007. Results: The overall prevalence of MetS in young adults aged 24-39 years in 2001 was 10-15 % and 6 years later in 30-45 year-old adults it was 15-23 % depending on the MetS definition used. Between the years 1986 and 2001, MetS prevalence increased from 1.0 % to 7.5 % (p<0.0001) in 24-year-old participants that was mostly driven by the increased central obesity. Participants with MetS had increased carotid intima-media thickness (cIMT) and decreased carotid elasticity compared to those without the syndrome. Impaired brachial flow-mediated dilatation (FMD) was not related to MetS but it modified the relationship between MetS and cIMT (P for interaction 0.023). High levels of apoB, CRP, sPLA2 and low levels of apoA-I associated with MetS in young adults. In prospective analysis both MetS and high apoB predicted (P<0.0001) incident high cIMT, defined as cIMT>90th percentile and/or plaque. The association between MetS and incident high cIMT was attenuated by ~40 % after adjustment with apoB. Conclusions: MetS is common in young adults and increases with age. Screening for risk factors, especially obesity, at an early life stage could help identify children and adolescents at increased risk of developing MetS and cardiovascular disease later in life. MetS identifies a population of young adults with evidence of increased subclinical atherosclerosis. Impaired brachial endothelial response is not a hallmark of MetS in young adults, but the status of endothelial function modifies the association between metabolic risk factors and atherosclerosis. In addition, the atherogenicity of MetS in this population assessed by incident high cIMT appears to be substantially mediated by elevated apoB.