Accurate and robust inference of mutational bias across complex traits and diseases

项目总结/摘要 人类遗传学中一个长期存在的问题是为什么常见疾病继续在人群中持续存在 尽管有潜在的健身后果。天真地，我们会期望选择将疾病从 人口，这表明它必须由一个抗衡力量来维持。一个自然的零假设是， 疾病至少部分地通过对疾病状态的突变偏向而在群体中得以维持。到 迄今为止，复杂疾病中的突变程度仍未得到探索。然而，在过去的十年里， 半全基因组关联研究（GWAS）已经确定了与复杂的 疾病提供深入了解复杂疾病的潜在遗传结构。的发展 如果有合适的工具，大量GWAS数据的涌入将使我们能够测量跨性状的突变偏差。这里我 我建议开发出准确和可靠地推断突变偏倚的方法，并将其应用于GWAS 数据集。群体遗传学理论表明，因果风险等位基因的平均频率是一个敏感的指标， 突变偏差的衡量标准在目标1中，我将首先证明计算 在GWAS中鉴定的相关风险等位基因也是突变偏倚的准确量度。然后我将 对泛英国研究中可用的7，221种表型进行广泛的突变偏倚证据筛查， 生物银行在目标2中，我将首先证明GWAS汇总统计中的人群分层可以 潜在地模拟突变偏差的信号，然后开发一种方法来校正分层的影响， 使用等位基因的进化状态作为工具变量。最后，在目标3中，我将扩展尖端 基于LD的统计方法，用于估计和划分遗传力，以提供第一个全基因组 在复杂性状和功能基因组类别内的突变偏倚程度的估计。 总之，我的研究结果将提供深入了解驱动复杂性状进化的进化力量， 新方法的应用超出了我的直接研究问题。更好地理解进化论 驱动疾病病因学的机制不仅有助于回答疾病持续性的问题， 揭示了驱动疾病易感性的生物过程。

PROJECT SUMMARY/ABSTRACT A long-standing question in human genetics is why common diseases continue to persist in the population despite potential fitness consequences. Naively, we would expect selection to remove disease from the population, suggesting that it must be maintained by a countervailing force. A natural null hypothesis is that disease is maintained in the population at least partially by a mutational bias towards the disease state. To date, the degree of mutational across complex diseases remains unexplored. However, over the past decade and a half genome-wide association studies (GWAS) have identified genetic loci associated with complex diseases providing insight into the underlying genetic architecture of complex diseases. With the development of the right tools, this large influx of GWAS data will allow us to measure mutational bias across traits. Here I propose to develop methods that accurately and robustly infer mutational bias and apply them to GWAS datasets. Population genetic theory indicates that the average frequency of causal risk alleles is a sensitive measure of mutational bias. In Aim 1, I will first demonstrate that calculating the average frequency of associated risk alleles, as identified in GWAS, is also an accurate measure of mutational bias. I will then conduct a broad screen for evidence of mutational bias across the 7,221 phenotypes available in the Pan-UK Biobank. In Aim 2, I will first demonstrate that population stratification in GWAS summary statistics can potentially mimic signals of mutational bias and then develop a method to correct for the effects of stratification using the evolutionary status of an allele as an instrumental variable. Finally, in Aim 3 I will extend cutting-edge LD based statistical methods for estimating and partitioning heritability to provide the first genome-wide estimates of the degree of mutational bias across complex traits and within functional genomic categories. Together my results will provide insight into the evolutionary forces driving complex trait evolution and generate novel methods with application beyond my direct research question. A better understanding of the evolutionary mechanisms driving disease etiology will not only help answer the question of disease persistence but also uncover biological processes driving disease susceptibility.