Investigators have discovered thousands of genetic variants associated with various traits using genome-wide association studies (GWAS). These discoveries have substantially improved our understanding of the genetic architecture of many complex traits. Despite the striking success, these trait-associated loci collectively explain relatively little of disease risk. Many reasons for this unexplained heritability have been suggested and two understudied components are hypothesized to have an impact in complex disease etiology: rare variants and gene-environment (GE) interactions. Advances in next generation sequencing have offered the opportunity to comprehensively investigate the genetic contribution of rare variants on complex traits. Such diseases are multifactorial, suggesting an interplay of both genetics and environmental factors, but most GWAS have focused on the main effects of genetic variants and disregarded GE interactions. In this dissertation, we develop statistical methods to detect GE interactions for rare variant analysis for various types of outcomes in both independent and related samples. We leverage the joint information across a set of rare variants and implement variance component score tests to reduce the computational burden. First, we develop a GE interaction test for rare variants for binary and continuous traits in related individuals, which avoids having to restrict to unrelated individuals and thereby retaining more samples. Next, we propose a method to test GE interactions in rare variants for time-to-event outcomes. Rare variant tests for survival outcomes have been underdeveloped, despite their importance in medical studies. We use a shrinkage method to impose a ridge penalty on the genetic main effects to deal with potential multicollinearity. Finally, we compare different types of penalties, such as least absolute shrinkage selection operator and elastic net regularization, to examine the performance of our second method under various simulation scenarios. We illustrate applications of the proposed methods to detect gene x smoking interaction influencing body mass index and time-to-fracture in the Framingham Heart Study. Our proposed methods can be readily applied to a wide range of phenotypes and various genetic epidemiologic studies, thereby providing insight into biological mechanisms of complex diseases, identifying high-penetrance subgroups, and eventually leading to the development of better diagnostics and therapeutic interventions.

No description available.

No description available.

Background: Type 2 diabetes is thought to result from interactions between genetic and lifestyle factors, but few robust examples exist. The overarching aim of this thesis was to discover such interactions by studying cohorts of white youth and adults from northern Europe in which physical activity, genotypes, and diabetes-related traits or diabetes incidence had been ascertained.   Methods: The thesis includes four papers. In Paper I, we investigated associations and interactions between 35 common PPARGC1A polymorphisms and cardiovascular and metabolic disease traits in 2,101 Danish and Estonian children from the European Youth Heart Study (EYHS). Paper II used the same cohort to test associations and interactions on cardiometabolic traits for the diabetes-predisposing TCF7L2 polymorphism. In Paper III, we assessed associations for 17 type 2 diabetes gene polymorphisms on impaired glucose regulation (IGR) or incident type 2 diabetes, and tested whether these effects are modified by physical activity in a prospective cohort study of ~16,000 initially non-diabetic Swedish adults – the Malmö Preventive Project (MPP). Paper IV aimed to replicate main genetic effects and gene x physical activity interactions for an FTO polymorphism on obesity in 18,435 primarily non-diabetic Swedish (MPP) and Finnish (Prevalence, Prediction and Prevention of Diabetes in Botnia) adults. Results: In Paper I, nominally significant associations were observed for BMI (rs10018239, P=0.039), waist circumference (rs7656250, P=0.012; rs8192678 [Gly482Ser], P=0.015; rs3755863, P=0.02; rs10018239, P=0.043), systolic blood pressure (rs2970869, P=0.018) and fasting glucose concentrations (rs11724368, P=0.045). Stronger associations were observed for aerobic fitness (rs7656250, P=0.005; rs13117172, P=0.008) and fasting glucose concentrations (rs7657071, P=0.002). None remained significant after correcting for multiple statistical comparisons. We proceeded by testing for gene × physical activity interactions for the polymorphisms that showed statistical evidence of association (P<0.05) in the main effect models, but none was statistically significant. In Paper II, the minor T allele at the rs7903146 variant was associated with higher glucose levels in older (beta=–0.098 mmol/l per minor allele copy, P=0.029) but not in younger children (beta=–0.001 mmol/l per minor allele copy, P=0.972). A significant inverse association between the minor allele at rs7903146 and height was evident in boys (beta=–1.073 cm per minor allele copy, P=0.001), but not in girls. The test of interaction between the TCF7L2 rs7903146 variant and physical activity on HOMA-B was nominally statistically significant (beta=0.022, Pinteraction=0.015), whereby physical activity reduced the effect of the risk allele on estimated beta-cell function. In Paper III, tests of gene x physical activity interactions on IGR-risk for three polymorphisms were nominally statistically significant: CDKN2A/B rs10811661 (Pinteraction=0.015); HNF1B rs4430796 (Pinteraction=0.026); PPARG rs1801282 (Pinteraction=0.04). Consistent interactions were observed for the CDKN2A/B (Pinteraction=0.013) and HNF1B (Pinteraction=0.0009) variants on 2 hr glucose concentrations. Where type 2 diabetes was the outcome, only one statistically significant interaction effect was observed and this was for the HNF1B rs4430796 variant (Pinteraction=0.0004). The interaction effects for HNF1B on 2 hr glucose and incident diabetes remained significant after correction for multiple testing (Pinteraction=0.015 and 0.0068, respectively). In Paper IV, the minor A allele at rs9939609 was associated with higher BMI (P<0.0001). The tests of gene x physical activity interaction on BMI were not statistically significant in either cohort (Sweden: P=0.71, Finland: P=0.18). Conclusions: Variation at PPARGC1A is unlikely to have a major impact on cardiometabolic health in European children, but physical activity may modify the effect of the TFC7L2 variants on beta-cell function in this cohort. In Swedish adults, physical activity modifies the effects of common HNF1B and CDKN2A/B variants on risk of IGR and also modifies the effect of the HNF1B on type 2 diabetes risk. In Swedish and Finnish adults, we were unable to confirm previous reports of an interaction between FTO gene variation and physical activity on obesity predisposition.

Risky alcohol use increases the risk of certain crimes such as drunk driving, spousal abuse and fighting. Around 60% of an individual’s alcohol use is attributable to genetic influences, however little is known regarding the specific genes that are involved in increasing the risk of risky alcohol use. Recent theories posit that some genes are so called susceptibility genes, meaning that carriers of certain genes or alleles are more susceptible both to positive and negative environments. The aim of the present study was to examine main and interaction effects of a possible susceptibility gene (ANKK1, which in previous research has been found to be related to risky alcohol use), and victimization as a negative environmental factor and parent-child relationship as a positive factor. Data were drawn from the RESUME project, and in the present study, 1.800 participants were included (47% males; 53% females; mean age of 22.15 years). Results showed no statistically significant main or interaction effect for ANKK1, but a statistically significant main effect was found for victimization and parent-child relationship. In conclusion, future research should include a larger sample size and use participants diagnosed with alcohol dependency. In addition, the susceptibility properties of ANKKI needs to be further examined, as the results from the present study indicate that ANKK1 is not a susceptibility gene. / Riskfylld alkoholkonsumtion kan öka risken för att vissa brott begås, såsom rattfylleri, partnervåld och slagsmål. Forskning har visat att omkring 60% av en individs alkoholkonsumtion kan förklaras av genetik, men det finns lite forskning kring just vilka gener som ökar risken för alkoholism. Nya teorier tror att vissa gener är så kallade sårbarhetsgener, vilket innebära att individer som bär på dessa gener eller särskilda alleler är mer sårbara för både positiva och negativa miljöfaktorer. Syftet med den nuvarande studien var att undersöka huvud-och interaktionseffekter av en potentiell sårbarhetsgen (ANKK1, som tidigare forskning visat har en påverkan på riskfylld alkoholkonsumtion), med utsatthet som den negativa faktorn och barn-föräldrarelation som den positiva faktorn. Data i form av enkätsvar och DNA prov från RESUME studien användes för den nuvarande studien, där 1,800 deltagare inkluderades (47% män; 53% av kvinnor; medelålder = 22.15). Resultaten visade inte på någon statistiskt signifikant huvud-eller interaktionseffekt för ANKKI, men det fanns en statistiskt signifikant huvudeffekt för utsatthet och barn-föräldrarelation. Slutsatsens som dras är att framtida studier bör fokusera på att ha ett större urval och använda sig av deltagare som är diagnosticerade med alkoholmissbruk. Utöver det bör framtida forskning fortsätta undersöka om ANKK1 är en sårbarhetsgen, eftersom resultaten från denna studie visar att den inte är en sårbarhetsgen.

Diss. (sammanfattning) Umeå : Umeå universitet, 2010. / Härtill 4 uppsatser. Även tryckt utgåva.