Computational modeling of spatial genome organization and gene regulation

项目总结/摘要 基因组的三维（3D）组织在基因组稳定性、基因调控、 和许多疾病，包括癌症。高通量染色质构象捕获技术的研究进展 （Hi-C）及其变体为研究更高级的染色质组织提供了前所未有的机会。 尽管研究3D基因组组织的资源迅速积累，但我们对3D基因组组织的理解仍然是有限的。 基因组组织的调节机制和功能在很大程度上仍然不完整。Hi-C分析和 3D基因组研究仍处于早期阶段，面临着一些挑战。第一，高分辨率染色质 接触图谱需要极深的测序，因此仅在少数细胞系中实现。第二、 用一维（1D）基因组来补充3D基因组结构在计算上具有挑战性， 表观基因组特征第三，最近的研究才刚刚开始推断染色质相互作用之间的关联 和遗传变异体，并在全基因组范围内鉴定这些变异体的潜在靶基因。给定 这些挑战和我独特的多学科训练，我的长期研究目标是发展创新 计算和统计方法来揭示3D基因组结构和功能之间的相互作用。 具体来说，在未来五年，我将i）开发计算方法，以提高现有的分辨率 Hi-C数据和研究细尺度3D基因组结构及其时空动态，以及ii）构建 可扩展和可解释的机器学习模型，利用1D表观基因组数据预测细胞类型特异性 三维染色质相互作用与基因表达，阐明三维基因组结构在基因表达中的作用 法规和人类疾病。该工作的完成将加深我们对三维基因组的认识 结构及其在基因调控和疾病中的功能。

PROJECT SUMMARY/ABSTRACT The three-dimensional (3D) organization of the genome plays an essential role in genome stability, gene regulation, and many diseases, including cancer. The recent development of high-throughput chromatin conformation capture (Hi-C) and its variants provide an unprecedented opportunity to investigate higher-order chromatin organization. Despite the rapidly accumulating resources for investigating 3D genome organization, our understanding of the regulatory mechanisms and functions of the genome organization remain largely incomplete. Hi-C analyses and 3D genome research are still in their early stage and face several challenges. First, high-resolution chromatin contact maps require extremely deep sequencing and hence have been achieved only for a few cell lines. Second, it is computationally challenging to complement 3D genome structure with one-dimensional (1D) genomic and epigenomic features. Third, recent studies have just begun to infer associations between chromatin interactions and genetic variants and to identify potential target genes of those variants at the genome-wide scale. Given these challenges and my unique multi-disciplinary training, my long-term research goal is to develop innovative computational and statistical methods to uncover the interplay between 3D genome structure and function. Specifically, in the next five years, I will i) develop computational approaches to enhance the resolution of existing Hi-C data and investigate fine-scale 3D genome architecture as well as its spatiotemporal dynamics and ii) build scalable and interpretable machine learning models that leverage 1D epigenomic data to predict cell type-specific 3D chromatin interactions and gene expression and elucidate the function of 3D genome organization in gene regulation and human diseases. The completion of the proposed work will deepen our knowledge of 3D genome architecture as well as its functions in gene regulation and disease.