The Genome Aggregation Database (gnomAD)

项目摘要 基因组聚合数据库（gnomAD）是基础研究和临床研究的普遍资源。 解释。作为世界上最大的遗传变异资源，gnomAD数据集几乎用于所有临床 该网站是全球基因诊断管道的一部分，迄今网页浏览量已超过2 000万次。这里我们 概述了一项提案，该提案将把gnomAD资源扩展到全球不同地区的数百万个样本。 人口。我们的建议将扩大变异调用和质量控制，以匹配此样本量，整合 统计工具和其他对临床解释至关重要的基因组资源，并确保我们 aggregate将继续与生物医学界自由分享。为此，我们将应用一个 高计算效率的策略来调用所有类别的变化（包括SNV，小插入缺失，和 结构变异的突变谱）在数百万个测序样品中富集， 代表祖先群体。我们将部署基于云的框架，以实现高效存储和自动化 使用大规模并行Hail对这些非常大且异质的序列数据集进行质量控制 架构我们将利用gnomAD的规模来提供越来越高分辨率的地图， 跨基因组区域的功能变异的耗尽（突出显示天然存在的基因组区域） 选择约束DNA变化），并提供定量评估是否 变异的群体频率与致病性一致，将这一信息与来自 ClinVar资源。我们将继续尽可能迅速和公开地与 早在出版之前，生物医学界。我们将支持和扩展我们的广泛的功能 访问的数据浏览器，以及创建可扩展和公开访问的数据集，集成我们的变化 具有临床和功能基因组注释的数据，可通过API框架访问， 数据集的应用。我们还将提供资源和培训，以改进gnomAD的使用 临床遗传学和更广泛的生物医学社区的资源。

Project Summary The Genome Aggregation Database (gnomAD) is a ubiquitous resource for basic research and clinical interpretation. The world’s largest genetic variation resource, the gnomAD dataset is used in virtually all clinical genetic diagnostic pipelines worldwide, and the website has over 20 million page views to date. Here we outline a proposal that will expand the gnomAD resource to millions of samples across diverse global populations. Our proposal will scale variant-calling and quality control to match this sample size, integrate statistical tools and other genomic resources critical to clinical interpretation, and ensure that the data we aggregate will continue to be shared freely with the biomedical community. To accomplish this we will apply a highly computationally efficient strategy to call all classes of variation (including SNVs, small indels, and the mutational spectrum of structural variants) across millions of sequenced samples enriched for under- represented ancestry groups. We will deploy a cloud-based framework for the efficient storage and automated quality control of these very large and heterogeneous sequence data sets using the massively parallel Hail architecture. We will leverage the scale of gnomAD to provide increasingly high-resolution maps of the depletion of functional variation across regions of the genome (highlighting genome regions where natural selection constrains DNA change) and provide statistical frameworks for quantitatively assessing whether the population frequency of a variant is consistent with pathogenicity, linking this information with evidence from the ClinVar resource. We will continue to share all of this data as rapidly and openly as possible with the biomedical community, long before publication. We will support and expand functionality in our widely accessed data browser as well as create scalable and publicly accessible datasets that integrate our variation data with clinical and functional genomic annotations, accessible through API frameworks to empower novel applications of the datasets. We will also provide resources and training to improve the use of gnomAD resources by the clinical genetics and wider biomedical communities.