Exploring Disease Relationships In Terms Of Dysfunctional Regulation Mechanisms And Developing Differential Regulation Analysis Method

Currently rapid development of genetic and molecular biology techniques such as whole genome sequencing, GWAS, and multi-omics high-throughput data promote the studies for identification of disease related genes and revealing of disease pathogenesis. Simultaneously, scientists focused on complicated relationships between diseases. Because understanding how diseases are related to each other can provide biological insights into etiology and pathogenesis, and furthermore help to prioritize disease-related genes, perform drug repositioning and personalized therapy. Following this sense, we developed a computational approach to identify disease similarities in terms of similar dysfunctional regulation mechanisms based on microarry data by using updated DCGL v2. Furthermore, we implemented the computational approach as an R package-DSviaDRM for facilitating disease similarities study. Besides, based on another transcriptomic data, RNA-seq, we introduced and compared two methods for identification of allele-specific expression (ASE). This work provided the basis for further ASE study.The difference of transcription regulation mechanism may have caused diverse phenotypes. How to extract regulation mechanisms and compare their differences (i.e. differential regulation) become one of the most important things. Differential coexpression analysis which was designed for exploring gene interconnection changes has been considered more promising in unveiling differential regulation mechanisms of diseases. In order to promote differential coexpresison analysis to differential regulation analysis, we modified existing differential coexpression analysis and developed new differential regulation analysis method into DCGL v2. In DCGL v2, we first modified differential coexpression analysis for confirming a list of differentially coexpressed genes and links, and then identified differential regulation genes and links based on regulation knowledge, and visualized differential coexpression and differential regulation information, and finally prioritized transcription factors in terms of their potential relevance to the phenotype of interest. (This work is showed in Chapter 2).We performed differential coexpression analysis for gene expression profiles of 108 diseases (disease samples vs. health samples for each disease state) by using DCGL v2. Then differential coexpression values of genes were converged to differential coexpression values of biologic pathways. Partial Spearman correlation of pathways’differential coexpression values between two diseases was obtained as the disease similarity value. We identified 1,326 disease-disease links among 108 diseases and found disease-disease links based on differential coexpression method were more relevant to pathogeneses than differential expression based method. Simultaneously, we found both the type of disease and the type of affected tissue influenced the degree of disease similarity by estimating the disease relationships between diseases which originated from a same tissue or disease relationships between same diseases which were sampled from multiply tissues. (This work is showed in Chapter 3).Additionally, in order to facilitate researchers to expansion of applying range, we implemented the above strategy as DSviaDRM package. DSviaDRM contains three part:1) three data (exprs1、exprs2 and exprs3); 2) two knowledge library (pathways and tf2target); 3) five functionalities for identifiying disease similarity (DCEA, DCpathway^ DA> comDCGL �'� comDCGLplot). It is the first package for identification of disease relationships from the view of similar dysfunction regulation mechanisms, and thus provide usefull tool for doctors and biologists. (This work is showed in Chapter 4).Besides, in gene expression study domain, increasingly importance has been attached to allele specific expression (ASE). ASE involves in many important biological mechanisms, i.e., X-inactivation, autosomal imprinting, and cis regulation. It is essential for normal development and related cellular processes and impaired ASE can result in disease. With the continuously developed Next Generation Sequencing technology we introduced two methods for identification of ASE (pDNAar and mREF) based on RNA-seq data and then compared the two methods and found mREF theoretically outperformed than pDNAar. Our study provides useful information for ASE identification. (This work is showed in Chapter 5).Overall, we developed DCGL v2 for differential regulation analysis, and identified disease similarities based on similar dysfunction regulation mechanisms by using DCGL v2 and finally implemented an R package, DSviaDRM, for identification of disease-disease links. Besides, we compared two computational approaches for identification of allele specific expression based on next generation sequencing data and this comparison provided useful information for ASE identification.