结合基因条件关联检验和组织特异性表达谱解析全基因组关联信号的方法研究

Complex diseases are important causes of human health problem. Dissection of causal or susceptibility genes of complex diseases from genome-wide statistical association signals of sequence variants remains a fundamentally important mission in current post-genomic era. However, current approaches of isolating susceptibility genes from the statistical signals of sequence variants generally do not perform well mainly due to two limitations, 1) interruption by indirectly associated genes and 2) the usage of single type of information. The principle investigator (PI) published a series of gene-based association tests for prioritizing genes responsible for a complex disease by using genome-wide association study (GWAS) summary statistics of sequence variants. In the preliminary studies, the PI also develops a conditional gene-based association method which is able to remove genes of indirect association with a certain disease by using summary statistics. When applied to GWAS summary statistics of schizophrenia, this approach isolates directly associated genes which have significantly more intensive co-expression in a cortex region of human brain than other genes. In contrast, the indirectly associated genes isolated by this new method have no intensive co-expression. Based on what we have done, we will continue exploring methodological breakthroughs for addressing the two limitations. Specifically, we will combine the conditional gene-based association and tissue specific gene expression features to develop a more powerful framework for precisely interpreting association signals from GWAS. The new framework will be validated by real data of schizophrenia. Distinguished from existing approaches, this framework will be free of interruption of indirectly associated genes and will jointly use two types of complementary information, statistical significance and expression data. Therefore, it would be more powerful and will identify more susceptibility genes of schizophrenia and other complex diseases by using fewer samples, and then will contribute to understanding of the pathogenic mechanisms, development of diagnosis and treatment of complex diseases.

复杂疾病严重危害公众健康，从全基因组统计关联位点中进一步解析出其风险基因是后基因组时代的重要课题。但该领域面临两大方法学问题: ①难以克服基因之间的间接关联信号干扰；②利用信息比较单一。因此，解析效果不理想。在前期研发了一系列基因水平关联分析解析方法的基础上，申请人在预实验中提出了一项克服间接关联基因的“条件关联检验”新方法；用该方法分离出的精神分裂症条件关联基因在脑皮层中有大量的共表达，而间接关联基因则无此特征。我们拟在后继工作中进一步探索突破以上两大难题的方案，结合基因的条件统计关联与组织特异性表达特征，利用精神分裂症数据实例验证，深入研发精准解析全基因组关联信号的方法和软件包。相比于现有方法，该方法将有去除间接关联基因干扰和整合统计与表达功能两类互补信息的两大特色，将可能以更少的样本资源鉴定更多的精神分裂症易感基因，并可以扩展到其它复杂疾病的研究，促进复杂疾病分子机理解析与诊断治疗。

复杂疾病严重危害公众健康，从成千上万的全基因组统计关联位点中进一步解析和甄别出“真正”的疾病易感基因是目前精准医学领域的重要课题之一。本项目围绕这一问题，先后创建了三项结合基因条件关联检验和组织特异性表达谱解析全基因组关联信号的新方法，并应用新方法对精神分裂症等多种复杂疾病实例数据进行了系统性解析，为复杂疾病易感基因定位研究提供了有效的分析工具和应用范例。. 本项目首先按照计划书完成了基因水平条件关联检验新方法─有效卡方检验（ECS）的研发。该方法是首个直接利用GWAS统计量而进行条件关联检验的方法，结果发表于生物信息学权威专业期刊《Bioinformatics》。在此基础上，创建了一项全新的整合基因组织特异表达定位复杂疾病原发组织和易感基因的分析方法DESE。该方法不仅可以准确推断复杂疾病的原发组织和细胞类型，还可以帮助挖掘更多功能性的易感基因。结果发表于基因组与遗传学权威专业期刊《Genome Biology》。在DESE的基础上，进一步提出了整合eQTL数据更全面地发掘复杂疾病关联基因的新方法eDESE。eDESE提出了一个更准确的无需人为修正假阳性膨胀的有效卡方检验计算模型，还提出了一个不受限物理距离合并基因水平关联信号的策略使得该方法可以挖掘出更多在功能上有生物学意义的疾病关联基因。结果已被权威生物学综合期刊《eLife》正式接收。. 在本项目的支持下，我们还以精神分裂症、双向情感障碍为模板，探讨了新方法的实际应用效果，发现了多个重要的易感基因、生物学通路和相应的病理组织。此外，还将方法应用于乙型肝炎病毒介导肝癌的数据挖掘分析，发现了SLC39A8等潜在的肝癌风险相关基因。结果发表于癌症领域专业期刊《BMC Cancer》。. 本项目研发的ECS、DESE和eDESE方法已经实现在课题组的软件平台KGG（http://pmglab.top/kgg）和KGGSEE（http://pmglab.top/kggsee）上，已被多个课题组用于复杂疾病GWAS数据挖掘分析。这些方法学研究为解决如何通过GWAS数据更准确地发掘复杂疾病易感基因和病理机制的前沿难点问题提供了新思路和关键技术探索。