Function-based exploration of genetic variation at genome-scale

项目摘要 全基因组关联研究发现了数千种与表型相关的遗传变异， 比如疾病风险。大多数相关的变异位于基因组的非编码区， 这些变异体的致病作用在很大程度上仍不清楚。关于相互作用的知识的稀疏性 基因组的编码和非编码调控部分之间的相互作用， 仅仅依靠基因组序列和位置是不可能的。我们建议通过实验来揭示功能 大规模遗传变异的相关性，通过干扰变异和含有变异的遗传元件， 并阅读出这些扰动对基因调控的直接影响。为此，我们建议 应用我们最近开发的CRISPR/Cas9功能基因组学筛选技术， 转录组学读出（简称为靶向Perturb-Seq或TAP-Seq），以实现对非转录组的系统询问。 编码区及其遗传变异。首先，我们将应用我们有针对性的Perturb-seq来破译 在整个人类染色体上编码的调节电路，通过系统地干扰所有主要的遗传基因， 元件（增强子、蛋白质编码和lncRNA基因）。这一广泛的数据集将使破译 控制所选染色体上基因表达的复杂调控网络。接下来，我们将揭开 通过将高通量精确基因组编辑与 同时进行单细胞基因组和转录组读出。使用这种新方法，我们将能够 破译遗传变异对基因表达的功能影响，并推导出遗传变异 干扰基因调控过程。我们将把生成的数据与现有的功能基因组学相结合 数据，如转录因子结合（ChIP-seq），染色质可及性（ATAC-seq，DNAse-seq）和 3D（Hi-C）中的交互，以便训练机器学习模型来导出所观察到的监管规则。 交互.这些模型将被应用于破译调控逻辑背后的分子机制， 并预测整个基因组和跨细胞类型的调控相互作用和变异。选择 预测将使用已建立的扰动技术进行实验验证，以验证临床 相关的预测，并提高预测模型的性能。总的来说，该项目将 回答基因调控的基本问题，揭示遗传变异影响的机制， 基因表达，并创建数据集和计算模型作为解释结果的有价值的工具， GWAS、eQTL和临床基因组研究。

PROJECT SUMMARY Genome-wide association studies have discovered thousands of genetic variants associated with phenotypic traits such as disease risk. Most of the associated variation lies within non-coding regions of the genome and the causative effects of those variants remain largely unknown. The sparsity of knowledge on interactions between the coding and non-coding regulatory parts of the genome makes the prediction of variant function solely from genome sequence and location impossible. We propose to experimentally uncover the functional relevance of genetic variants at a large scale, by perturbing variants and genetic elements containing variants, and reading out the direct consequences of those perturbations on gene regulation. To this end, we propose to apply our recently developed CRISPR/Cas9 functional genomics screening technology with targeted single-cell transcriptomic readouts (targeted Perturb-seq or TAP-Seq in short) to enable systematic interrogation of non- coding regions and genetic variation therein. First, we will apply our targeted Perturb-seq to decipher the regulatory circuitry encoded on an entire human chromosome by systematically perturbing all major genetic elements (enhancers, protein-coding and lncRNA genes). This extensive data set will enable to decipher the complex regulatory networks controlling gene expression on the selected chromosome. Next, we will uncover causal regulatory variants in these regions by coupling high-throughput precision genome editing to simultaneous single-cell genomic and transcriptomic readout. Using this novel approach, we will be able to decipher the functional impact of genetic variants on gene expression and derive rules by which genetic variation perturbs gene regulatory processes. We will integrate the generated data with available functional genomics data, such as transcription factor binding (ChIP-seq), chromatin accessibility (ATAC-seq, DNAse-seq) and interactions in 3D (Hi-C), in order to train machine learning models to derive rules of the observed regulatory interactions. These models will be applied to decipher the molecular mechanisms underlying the regulatory logic, and to predict regulatory interactions and variants throughout the genome and across cell types. Selected predictions will be experimentally validated using the established perturbation technologies, to verify clinically relevant predictions and improve the performance of the predictive models. Taken together, this project will answer fundamental questions in gene regulation, uncover the mechanisms by which genetic variation impacts gene expression, and create datasets and computational models as valuable tools for interpreting results from GWAS, eQTL and clinical genomic studies.