Integrating genome structure information and statistical analysis of co-segregation of rare genetic variants with phenotypes in families

在家族遗传学研究中，具有感兴趣表型的亲属共享罕见遗传变异（RV）是推断RV参与表型的关键信息。我率先发展RV共享概率之间的远亲科目的基础上连接RV的表型。这种方法在全基因组测序研究中的应用提出了许多挑战：1）目前没有有效的方法来对预期对基因调控具有类似影响的基因间变异进行分组; 2）RV的丰度意味着多个RV出现在需要精确描绘的小基因组区域内的相同单倍型上，以及3）RV共享概率可能被低估，由于未知的关系之间显然无关的家庭成员。我建议通过以下具体目标来解决我的研究计划中的这些挑战： 1.基于3D染色体接触定义罕见变异的聚类，并开发基于家族的统计数据，以测试这些聚类与全基因组序列数据中的二分表型之间的联系。 为此，RV的簇将源自通过高通量染色体构象捕获实验产生的3D接触矩阵，因为紧密接触的区域参与相同的基因调控过程。RV共享统计数据将在3D联系人领域内和不同领域的集群上开发。 2.精确推断远亲之间的重组事件，以改善罕见变异单倍型推断。 这一目标将通过结合来自家庭遗传传递和来自群体单倍型频率的信息来实现。 3.更准确地模拟家庭成员之间未知的关系。 由合作者开发的远亲关系估计值将被整合到RV共享方法中。 4.开发一个软件工具，将罕见变异统计和3D接触数据库连接起来。 所提出的方法将作为Bioconductor项目的一个包来实施，扩展现有的R包RV共享。 实际的DNA序列、表型和家族结构数据将用于训练模型、测试方法和软件以及校准模拟以评估统计特性。该研究计划将为进行家族遗传研究的研究人员提供新的工具，并在涉及这些研究人员的环境中培养生物统计学和生物信息学的研究生。

In familial genetic studies, sharing of rare genetic variants (RVs) by relatives with a phenotype of interest is a key piece of information to infer the involvement of the RVs in the phenotype. I have spearheaded the development of RV sharing probabilities among distantly related subjects as basis to link RVs to a phenotype. The application of this approach to whole genome sequencing studies presents a number of challenges: 1) there is currently no validated approach to group intergenic variants expected to have a similar impact on gene regulation; 2) the abundance of RVs implies that multiple RVs occur on the same haplotype within small genomic regions which need to be precisely delineated and 3) RV sharing probabilities may be underestimated due to unknown relationships among apparently unrelated family members. I propose to address these challenges in my research program through the following specific objectives: 1. Define clusters of rare variants based on 3D chromosomal contacts and develop family-based statistics to test the link between these clusters and dichotomous phenotypes in whole genome sequence data. To this end, clusters of RVs will be derived from 3D contact matrices produced by high-throughput chromosome conformation capture experiments, as regions in close contact are involved in the same gene regulation processes. RV sharing statistics will be developed over clusters within domains of 3D contacts and across distinct domains. 2. Accurately infer recombination events between distant relatives to improve rare variant haplotype inference. This aim will be achieved by combining information from genetic transmission in families and from population haplotype frequencies. 3. Model more accurately unknown relationships among family members. Estimates of distant relatedness developed by collaborators will be integrated in the RV sharing approach. 4. Develop a software tool linking rare variant statistics and 3D contact databases. The proposed methods will be implemented as a package for the Bioconductor project expanding the existing R package RVsharing. Actual DNA sequence, phenotype and family structure data will be used for training models, testing methods and software and calibrating simulations for evaluating statistical properties. This research program will provide new tools for researchers conducting familial genetic studies and train graduate students in biostatistics and bioinformatics in an environment involving such researchers.