Interrogation of the Impact of Selection on the Evolution of Human Pancreatic Cancer Precursor Lesions

项目摘要 胰腺癌是由癌前病变引起的，如果及早发现和治疗，这些病变是可以治愈的。 胰腺癌前病变导管内乳头状瘤的多区域基因组分析 粘液性肿瘤(IPMN)，提示IPMN与 浸润性癌症，强调了选择在癌前病变进展中的潜在重要性 损伤。我们建议使用综合分子来表征选择性作用力在IPMN中的作用 人类IPMN组织样本和有机培养物的分析和计算数据整合。我们 将使用多区域确定人类IPMN样本中进展的亚克隆的分子特征 全基因组测序和RNA测序。此外，我们还将开发和应用新的多组学 整合DNA和RNA测序数据以描绘人类选择性作用力的计算方法 IPMN。我们将通过全基因组DNA测序来确定进展克隆的功能 在一个新的半监督框架中从大量转录数据中推断出基因签名。到时候我们会的 采用人IPMN细胞的三维体外器官培养模型来表征相对分子质量 随着时间的推移，选择性压力的贡献。我们将使用组合的DNA测序和单细胞RNA- 测序以确定在我们的器官模型中进行的亚克隆的基因表达特征，进一步 使我们的计算框架适用于单细胞RNA测序数据。总而言之，拟议的 研究结合了对人体组织样本的直接分析和对人类癌前细胞的处理 三维文化与新颖的多组学计算集成，极大地扩展了我们对 选择在胰腺癌前病变中的作用。

Project Summary Pancreatic cancer arises from precancerous lesions that are curable if detected and treated early enough. Recent multi-region genomic analyses of one type of precancerous pancreatic lesion, intraductal papillary mucinous neoplasm (IPMN), suggest unique evolutionary and selective pressures in IPMNs compared to invasive cancers, underscoring the potential importance of selection in the progression of precancerous lesions. We propose to characterize the role of selective forces in IPMNs using comprehensive molecular analyses and computational data integration of both human IPMN tissue samples and organoid cultures. We will determine the molecular features of progressed subclones in human IPMN samples using multi-region whole genome sequencing and RNA-sequencing. In addition, we will develop and apply novel multi-omics computational methods to integrate DNA and RNA-sequencing data to delineate selective forces in human IPMNs. We will determine the function of progressed clones identified by whole genome DNA sequencing with gene signatures inferred from bulk transcriptional data in a new semi-supervised framework. We will then employ a three-dimensional in vitro organoid culture model of human IPMN cells to characterize the relative contributions of selective pressures over time. We will use combined DNA sequencing and single-cell RNA- sequencing to identify gene expression signatures of progressed subclones in our organoid model, further adapting our computational framework to single cell RNA-sequencing data. Taken together, the proposed studies combine direct analysis of human tissue samples and manipulation of human precancerous cells in three-dimensional culture with novel multi-omics computational integration to greatly expand our knowledge of the role of selection in pancreatic cancer precursor lesions.