An integrative omics approach to investigate gene-environment interaction in colorectal cancer risk

项目概要/摘要 结直肠癌（CRC）仍然是世界各地癌症相关死亡的主要原因之一。 许多环境风险因素和200多个遗传风险变异已被确定为这个复杂的， 多因素疾病然而，尽管有强大的生物学理由的重要性和丰富的 基因-环境（GxE）相互作用，环境风险因素（在此广义定义为 生活方式、饮食、肥胖、药物使用和中间生物标志物）调节遗传风险因素的机制知之甚少。 为了实现精准预防的承诺，我们迫切需要对GxE有更深入的了解 CRC风险的相互作用。了解哪些可改变的风险因素调节遗传风险，而遗传风险是固定的， 为新的预防干预战略提供生物学见解和可操作的目标。加快 发现GxE相互作用在CRC风险中的作用，并采取重要的下一步翻译，我们提出了一个 综合创新方法，结合单细胞多组学数据，个体水平协调 流行病学和临床数据，以及来自大型、充分表征、多样化研究的全基因组数据 人口，与新的计算和统计方法。单电池的显著改进 多模态组学技术，结合基于强大深度学习的新计算工具 建模方法现在允许我们预测遗传变异对细胞类型中基因调控的影响， 具体的整体方式。因为同时测量单细胞基因表达（scRNA-seq）和 正常结直肠粘膜组织的染色质可及性（scATAC-seq）数据缺乏种族和人种差异。 我们在目标1中建议，通过对环境风险因素进行详细评估， 50个人的数据。这一资源，连同其他单细胞多组学纲要结直肠组织 (like HTAN），将被用来开发跨基因组的遗传变异的功能预测评分。在 目标2，我们将使用这些功能预测分数来提高发现新GxE的统计能力 交互.我们将在超过230，000例种族和民族多样化的CRC病例中进行全基因组GxE扫描 并控制关键的环境风险因素，包括肥胖、糖尿病、吸烟、酗酒、吸毒， 与代谢失调和慢性炎症相关的饮食因素和中间生物标志物。到 为了扩大我们可以评估的关键风险因素的数量，我们将利用现有的遗传工具。在目标3中， 将全面表征和翻译GxE相互作用。为此，我们将通过以下方式对GxE结果进行分层： 临床因素，包括发病年龄、种族和民族、性别和肿瘤亚型。此外，我们将 将GxE相互作用和遗传预测生物标志物纳入综合跨祖先风险 预测模型，以改善预测，并提供可操作的信息，以减少CRC的负担。我们 社区顾问强调了包括遗传和环境之间相互作用的重要性 风险预测模型中的风险因素，以提高社会对风险预测模型的接受程度。

PROJECT SUMMARY/ABSTRACT Colorectal cancer (CRC) remains one of the leading causes of cancer-related deaths around the world. Many environmental risk factors and over 200 genetic risk variants have been identified for this complex, multifactorial disease. However, despite the strong biological rationale for the importance and abundance of gene-environment (GxE) interactions, the extent to which environmental risk factors (broadly defined here as lifestyle, diet, obesity, drug use and intermediate biomarkers) modulate genetic risk factors is poorly understood. To achieve the promise of precision prevention, we urgently need to gain a deeper understanding of GxE interactions in CRC risk. Understanding which modifiable risk factors modulate genetic risk, which is fixed, provides biological insights and actionable targets for new prevention intervention strategies. To accelerate the discovery of GxE interactions in CRC risk and to take an important next step towards translation, we propose a comprehensive innovative approach that combines single-cell multi-omics data, individual-level harmonized epidemiological and clinical data, and genome-wide data from large, well-characterized, diverse study populations, with novel computational and statistical approaches. Dramatic improvements in single-cell multimodal omics technologies, combined with new computational tools based on powerful deep-learning modeling approaches now allow us to predict the impact of genetic variants on gene regulation in a cell-type- specific holistic manner. Because simultaneously measured single-cell gene expression (scRNA-seq) and chromatin accessibility (scATAC-seq) data for normal colorectal mucosa tissue is lacking for racially and ethnically diverse samples with detailed assessment of environmental risk factors, we propose in Aim 1 to generate such data for 50 individuals. This resource, together with other single cell multi-omics compendia for colorectal tissue (like HTAN), will be leveraged to develop functional prediction scores for genetic variants across the genome. In Aim 2, we will use these functional prediction scores to boost statistical power for discovery of novel GxE interactions. We will perform genome-wide GxE scans in over 230,000 racially and ethnically diverse CRC cases and controls across key environmental risk factors, including obesity, diabetes, smoking, alcohol, drug use, dietary factors and intermediate biomarkers linked to metabolic dysregulation and chronic inflammation. To expand the number of key risk factors we can evaluate, we will utilize existing genetic instruments. In Aim 3, we will comprehensively characterize and translate GxE interactions. To do so, we will stratify GxE findings by clinical factors, including age of onset, racial and ethnic group, sex, and tumor subtypes. Additionally, we will incorporate GxE interactions and genetically predicted biomarkers in a comprehensive trans-ancestral risk prediction model to improve prediction and provide actionable information to reduce the burden of CRC. Our community advisors have stressed the importance of including the interplay between genetic and environmental risk factors in risk prediction modeling to enhance the acceptance of risk prediction models in the community.