The genetics of sleep patterns and their relationship to obesity and Type 2 diabetes

睡眠过多、过少或质量差与几种人类疾病有关，尤其是代谢紊乱，如肥胖和2型糖尿病。例如，每晚睡眠不足6小时的人肥胖的风险增加75%。睡眠模式的另一个方面是我们个人的昼夜节律——激素、体温和调节我们清醒和困倦感觉的大多数身体系统的24小时变化周期。扰乱我们的昼夜节律与疾病密切相关。例如，倒班工人患心脏病的风险增加了40%。虽然这些关联很强，但这些关联的本质意味着我们不能说睡眠模式是否会导致疾病，疾病是否会影响睡眠模式，或者是否有其他与两者相关的因素（例如，社会经济地位）可以解释这种关联。解决因果方向的一种方法是使用遗传学——因为一个人的遗传学在其一生中不会改变，我们可以使用遗传变异作为因果“锚”。现在被广泛使用的一种技术叫做孟德尔随机化，它使用与感兴趣的特征（如生物钟）相关的遗传变异，让我们能够测试它是否会导致疾病风险增加（如肥胖），反之亦然。识别与睡眠模式正常变化相关的遗传变异也将为睡眠生物学和昼夜节律提供新的见解，并为治疗睡眠障碍提供新的靶点。在这项提议中，我们首先旨在确定与睡眠模式相关的遗传变异。为了做到这一点，我们将首先测试来自英国生物银行研究的48万人的1亿2000万个基因变异，这些变异与睡眠时间、睡眠效率和昼夜节律测量等特征有关。我们将使用自我报告和基于活动监测的睡眠估计来确定每个人的睡眠变量。使用加速度计对睡眠的估计是很重要的，因为自我报告的测量可能会有不准确的报告。在英国生物银行，所有50万人的睡眠时间、睡眠效率和生物钟的自我报告测量将可用，我们将能够通过活动监测数据验证10万人子集中的关联。我们将使用来自bbb200000的数据复制从UK Biobank中确定的关联，包括23andMe、CHARGE和chronogen consortium等国际研究。这个复制阶段对于确保遗传关联的健壮性非常重要。我们将以两种方式使用相关的遗传变异信号。首先，我们将使用一系列计算机方法提供每个个体变异的生物学见解，以获得对睡眠和昼夜节律重要的个体基因的见解。我们将寻找关联信号之间的联系，以突出在这些表型中重要的途径、生物系统和组织。其次，我们将进行孟德尔随机化分析，以测试睡眠模式与代谢性疾病之间的流行病学关联的因果方向。我们将使用强相关变异，并从独立的、非常大规模的全基因组关联结果中测试它们是否与BMI、2型糖尿病和心脏病相关。我们还将进行反向分析，并测试BMI和2型糖尿病变体是否与睡眠模式相关。我们将使用最新的孟德尔随机化技术，如艾格回归来克服潜在的偏差。这项工作将为睡眠生物学和昼夜节律提供重要的新见解，并有助于确定代谢性疾病和睡眠中断之间的因果关系。

Too much, too little or poor quality sleep is associated with several human diseases, in particular metabolic disorders such as obesity and Type 2 diabetes. For example, individuals who sleep < 6 hours per night have a 75% increased risk of obesity. Another aspect of sleep patterns is our individual circadian rhythm - the 24 hour cycle of changes in hormones, body temperature and most body systems which regulate our feelings of wakefulness and sleepiness. Disrupting our circadian rhythms is strongly associated with disease. For example, shift-workers have a >40% increased risk of heart disease. While these associations are strong and robust, the nature of these associations means that we can't say whether sleep patterns are causing disease, whether the diseases affect sleep patterns or if something else associated with both (for example, socioeconomic status) explains the association. One way of addressing the causal direction is to use genetics - because an individual's genetics doesn't change over their lifetime we can use genetic variants as causal "anchors". A now widely-used technique called Mendelian Randomisation uses genetic variants associated with a trait of interest (e.g. chronotype) to allow us to test whether it causes an increased risk of disease (e.g. obesity) or vice versa. Identifying genetic variants associated with normal variation in sleep patterns will also provide new insights in the biology of sleep and circadian rhythms and provide new targets for therapeutics to treat sleep disorders.In this proposal we first aim to identify genetic variants associated with sleep patterns. We will do this by initially testing >20,000,000 genetic variants in 480,000 individuals from the UK Biobank study against traits such as sleep duration, sleep efficiency and measures of circadian rhythms. We will determine these sleep variables for each individual using both self-reported and activity-monitor based estimates of sleep. Using accelerometer derived estimates of sleep will be important because there may be reporting inaccuracies from self-reported measures. In UK Biobank self-reported measures of sleep duration, sleep efficiency and chronotype will be available in all 500,000 individuals and we will be able to validate the associations in a subset of 100,000 individuals with activity monitor data. We will replicate the associations identified from UK Biobank using data from >200,000 including international studies such as 23andMe, the CHARGE and chronogen consortia. This replication stage is important to ensure the genetic associations are robust.We will use the associated genetic variants signals in two ways. First, we will provide biological insights at each of the individual variants using a range of in silico approaches to gain insights into individual genes important in sleep and circadian rhythms. We will look for connections across association signals to highlight pathways, biological systems and tissues that are important in these phenotypes. Second, we will perform Mendelian Randomisation analyses to test the causal direction of the epidemiological association between sleep patterns and metabolic disease. We will use robustly associated variants and test whether they are associated with BMI, Type 2 diabetes and heart disease from independent very large-scale genome-wide association results. We will also perform the reverse analyses and test whether robustly associated BMI and Type 2 diabetes variants are associated with sleep patterns. We will use the latest Mendelian Randomisation techniques such as Egger's regression to overcome potential biases. This work will lead to important new insights into the biology of sleep and circadian rhythms and help determine the causal nature of the association between metabolic disease and disrupted sleep.