Bayesian Hierarchical Methods for Modeling Chromosomal Spatial Correlation

该奖项是根据2009年美国复苏和再投资法案（公法111-5）资助的。近年来，许多基因组学研究表明，许多生物体的基因表达存在显著的染色体空间相关性。在统计建模中忽略这种相关性会大大降低估计的效率和统计推断的能力。在本项目中，研究人员开发了一套新的贝叶斯层次模型，以解释染色体空间相关性在以下三个领域的方法和应用：(1)将与线性染色体结构相关的空间相关性纳入基因表达数据的分析；(2)利用基因表达数据定量推断体内染色体折叠结构；(3)探索染色体在体内的整体三维结构。该研究为探索三维染色体结构提供了统计工具，并直接解决了三维结构如何促进基因转录协调的重要生物学问题。它不仅可以对染色体的空间构型产生新的生物学见解，而且可以大大提高对生物体中染色体功能与结构关系的认识。它也有潜在的临床意义；例如，它提供了正式识别和比较肿瘤和正常样本中基因表达的染色体空间模式的工具，可以发现肿瘤基因组中微妙但协调的变化，并为癌症的潜在调节机制提供临床见解。除了具有科学的新颖性和意义外，该研究还促进了统计学、计算生物学、实验生物学和临床癌症研究等各个领域的研究人员之间的密切合作，并促进了参与机构之间的智力互动。

This award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5).Recently, many genomic studies have shown that significant chromosomal spatial correlation exists in gene expression of many organisms. Ignoring such correlation in statistical modeling can greatly reduce the efficiency of estimation and the power of statistical inference. In this project, the investigators develop a set of new Bayesian hierarchical models to account for chromosomal spatial correlation in the following three areas of methodologies and applications: (1) incorporating the spatial correlation associated with linear chromosome structures into the analysis of gene expression data; (2) quantifying and inferring chromosome folding structures in vivo using gene expression data; (3) probing the global three dimensional structure of a chromosome in vivo. This study provides statistical tools to explore three dimensional (3D) chromosome structures and directly addresses the important biological question that how the 3D structures facilitate the coordination in gene transcription. Not only can it generate new biological insights about spatial configuration of chromosomes, but also it can greatly improve the understanding of the relationship between the function and structure of chromosomes in living organisms. It also has potential clinical significance; for example, it provides tools to formally identify and compare chromosomal spatial patterns in gene expression from tumor and normal samples, which can discover subtle but coordinated changes in the tumor genome and lead to clinical insights on the underlying regulatory mechanism of cancer. Besides scientific novelty and significance, this research fosters intensive collaboration among researchers from various fields including statistics, computational biology, experimental biology and clinical cancer research, and promotes intellectual interactions among participating institutions.