Identification of Risk Pathways and Functional Modules for Coronary Artery Disease Based on Genome-wide SNP Data

Coronary artery disease(CAD) is a complex human disease, involving multiple genes and their nonlinear interactions, which often act in a modular fashion. Genome-wide single nucleotide polymorphism(SNP) profiling provides an effective technique to unravel these underlying genetic interplays or their functional involvements for CAD. This study aimed to identify the susceptible pathways and modules for CAD based on SNP omics. First, the Wellcome Trust Case Control Consortium(WTCCC) SNP datasets of CAD and control samples were used to assess the jointeffect of multiple genetic variants at the pathway level, using logistic kernel machine regression model. Then, an expanded genetic network was constructed by integrating statistical gene–gene interactions involved in these susceptible pathways with their protein–protein interaction(PPI)knowledge. Finally, risk functional modules were identified by decomposition of the network. Of 276 KEGG pathways analyzed, 6 pathways were found to have a significant effect on CAD. Other than glycerolipid metabolism, glycosaminoglycan biosynthesis, and cardiac muscle contraction pathways, three pathways related to other diseases were also revealed, including Alzheimer's disease, non-alcoholic fatty liver disease, and Huntington's disease. A genetic epistatic network of 95 genes was further constructed using the abovementioned integrative approach. Of 10 functional modules derived from the network, 6 have been annotated to phospholipase C activity and cell adhesion molecule binding, which also have known functional involvement in Alzheimer's disease.These findings indicate an overlap of the underlying molecular mechanisms between CAD and Alzheimer's disease, thus providing new insights into the molecular basis for CAD and its molecular relationships with other diseases.     

Identification of susceptibility modules for coronary artery disease using a genome wide integrated network analysis

Although recent genome-wide association studies (GWAS) have identified a handful of variants with best significance for coronary artery disease (CAD), it remains a challenge to summarize the underlying biological information from the abundant genotyping data. Here, we propose an integrated network analysis that effectively combines GWAS genotyping dataset, protein–protein interaction (PPI) database, literature and pathway annotation information. This three-step approach was illustrated for a comprehensive network analysis of CAD as the following. First, a network was constructed from PPI database and CAD seed genes mined from the available literatures. Then, susceptibility network modules were captured from the results of gene-based association tests. Finally, susceptibility modules were annotated with potential mechanisms for CAD via the KEGG pathway database. Our network analysis identified four susceptibility modules for CAD including a complex module that consisted of 15 functional inter-connected sub-modules, AGPAT3–AGPAT4–PPAP2B module, ITGA11–ITGB1 module and EMCN–SELL module. MAPK10 and COL4A2 among the top-scored focal adhesion pathway related module were the most significant genes (MAPK10: OR = 32.5, P = 3.5 × 10^− 11; COL4A2: OR = 2.7, P = 2.8 × 10^− 10). The significance of the two genes were further validated by other two gene-based association tests (MAPK10: P = 0.009 and 0.007; COL4A2: P = 0.001 and 0.023) and another independent GWAS dataset (MAPK10: P = 0.001; COL4A2: P = 0.0004). Furthermore, 34 out of 44 previously reported CAD susceptibility genes were captured by our CAD PPI network and 17 of them were also significant genes. The susceptibility modules identified in our study might provide novel clues for the clarification of CAD pathogenesis in the future.     

Systematic identification of common functional modules related to heart failure with different etiologies

The development of heart failure (HF) is a complex process that can be initiated by multiple etiologies. Identifying common functional modules associated with HF is a challenging task. Here, we developed a systems method to identify these common functional modules by integrating multiple expression profiles, protein interactions from four species, gene function annotations, and text information. We identified 1439 consistently differentially expressed genes (CDEGs) across HF with different etiologies by applying three meta-analysis methods to multiple HF-related expression profiles. Using a weighted human interaction network constructed by combining interaction data from multiple species, we extracted 60 candidate CDEG modules. We further evaluated the functional relevance of each module by using expression, interaction network, functional annotations, and text information together. Finally, five functional modules with significant biological relevance were identified. We found that almost half of the genes in these modules are hubs in the weighted network, and that these modules can accurately classify HF patients from healthy subjects. We also identified many significantly enriched biological processes that contribute to the pathophysiology of HF, including two new ones, RNA splicing and vesicle-mediated protein transport. In summary, we proposed a novel framework to analyze common functional modules related to HF with different etiologies. Our findings provide important insights into the complex mechanism of HF. Further biological experimentations should be required to validate these novel biological processes.     

Disease-driven detection of differential inherited SNP modules from SNP network

Detection of the synergetic effects between variants, such as single-nucleotide polymorphisms (SNPs), is crucial for understanding the genetic characters of complex diseases. Here, we proposed a two-step approach to detect differentially inherited SNP modules (synergetic SNP units) from a SNP network. First, SNP-SNP interactions are identified based on prior biological knowledge, such as their adjacency on the chromosome or degree of relatedness between the functional relationships of their genes. These interactions form SNP networks. Second, disease-risk SNP modules (or sub-networks) are prioritised by their differentially inherited properties in IBD (Identity by Descent) profiles of affected and unaffected sibpairs. The search process is driven by the disease information and follows the structure of a SNP network. Simulation studies have indicated that this approach achieves high accuracy and a low false-positive rate in the identification of known disease-susceptible SNPs. Applying this method to an alcoholism dataset, we found that flexible patterns of susceptible SNP combinations do play a role in complex diseases, and some known genes were detected through these risk SNP modules. One example is GRM7, a known alcoholism gene successfully detected by a SNP module comprised of two SNPs, but neither of the two SNPs was significantly associated with the disease in single-locus analysis. These identified genes are also enriched in some pathways associated with alcoholism, including the calcium signalling pathway, axon guidance and neuroactive ligand-receptor interaction. The integration of network biology and genetic analysis provides putative functional bridges between genetic variants and candidate genes or pathways, thereby providing new insight into the aetiology of complex diseases.     

LncRNA and mRNA integration network reconstruction reveals novel key regulators in esophageal squamous-cell carcinoma

Many experimental and computational studies have identified key protein coding genes in initiation and progression of esophageal squamous cell carcinoma (ESCC). However, the number of researches that tried to reveal the role of long non-coding RNAs (lncRNAs) in ESCC has been limited. LncRNAs are one of the important regulators of cancers which are transcribed dominantly in the genome and in various conditions. The main goal of this study was to use a systems biology approach to predict novel lncRNAs as well as protein coding genes associated with ESCC and assess their prognostic values. By using microarray expression data for mRNAs and lncRNAs from a large number of ESCC patients, we utilized “Weighted Gene Co-expression Network Analysis” (WGCNA) method to make a big coding-non-coding gene co-expression network, and discovered important functional modules. Gene set enrichment and pathway analysis revealed major biological processes and pathways involved in these modules. After selecting some protein coding genes involved in biological processes and pathways related to cancer, we used “LncTar”, a computational tool to predict potential interactions between these genes and lncRNAs. By combining interaction results with Pearson correlations, we introduced some novel lncRNAs with putative key regulatory roles in the network. Survival analysis with Kaplan-Meier estimator and Log-rank test statistic confirmed that most of the introduced genes are associated with poor prognosis in ESCC. Overall, our study reveals novel protein coding genes and lncRNAs associated with ESCC, along with their predicted interactions. Based on the promising results of survival analysis, these genes can be used as good estimators of patients' survival, or even can be analyzed further as new potential signatures or targets for the therapy of ESCC disease.     

Mining functional gene modules linked with rheumatoid arthritis using a SNP-SNP network

The identification of functional gene modules that are derived from integration of information from different types of networks is a powerful strategy for interpreting the etiology of complex diseases such as rheumatoid arthritis (RA). Genetic variants are known to increase the risk of developing RA. Here, a novel method, the construction of a genetic network, was used to mine functional gene modules linked with RA. A polymorphism interaction analy-sis (PIA) algorithm was used to obtain cooperating single nucleotide polymorphisms (SNPs) that contribute to RA disease. The acquired SNP pairs were used to construct a SNP-SNP network. Sub-networks defined by hub SNPs were then extracted and turned into gene modules by mapping SNPs to genes using dbSNP database. We per-formed Gene Ontology (GO) analysis on each gene module, and some GO terms enriched in the gene modules can be used to investigate clustered gene function for better understanding RA pathogenesis. This method was applied to the Genetic Analysis Workshop 15 (GAW 15) RA dataset. The results show that genes involved in func-tional gene modules, such as CD160 (rs744877) and RUNX1 (rs2051179), are especially relevant to RA, which is supported by previous reports. Furthermore, the 43 SNPs involved in the identified gene modules were found to be the best classifiers when used as variables for sample classification.     

Convergence and divergence of genetic and modular networks between diabetes and breast cancer

Diabetes mellitus (DM) and breast cancer (BC) can simultaneously occur in the same patient populations, but the molecular relationship between them remains unknown. In this study, we constructed genetic networks and used modularized analysis approaches to investigate the multi‐dimensional characteristics of two diseases and one disease subtype. A text search engine (Agilent Literature Search 2.71) and MCODE software were applied to validate potential subnetworks and to divide the modules, respectively. A total of 793 DM‐related genes, 386 type 2 diabetes (T2DM) genes and 873 BC‐related genes were identified from the Online Mendelian Inheritance in Man database. For DM and BC, a total of 99 overlapping genes, 9 modules, 29 biological processes and 7 pathways were identified. Meanwhile, for T2DM and BC, 56 overlapping genes, 5 modules, 20 biological processes and 12 pathways were identified. Based on the Gene Ontology functional enrichment analysis of the top 10 non‐overlapping modules of the two diseases, 10 biological functions and 5 pathways overlapped between them. The glycosphingolipid and lysosome pathways verified molecular mechanisms of cell death related to both DM and BC. We also identified new biological functions of dopamine receptors and four signalling pathways (Parkinson's disease, Alzheimer's disease, Huntington's disease and long‐term depression) related to both diseases; these warrant further investigation. Our results illustrate the landscape of the novel molecular substructures between DM and BC, which may support a new model for complex disease classification and rational therapies for multiple diseases.     

Highly interconnected genes in disease-specific networks are enriched for disease-associated polymorphisms

Background

Complex diseases are associated with altered interactions between thousands of genes. We developed a novel method to identify and prioritize disease genes, which was generally applicable to complex diseases.

Results

We identified modules of highly interconnected genes in disease-specific networks derived from integrating gene-expression and protein interaction data. We examined if those modules were enriched for disease-associated SNPs, and could be used to find novel genes for functional studies. First, we analyzed publicly available gene expression microarray and genome-wide association study (GWAS) data from 13, highly diverse, complex diseases. In each disease, highly interconnected genes formed modules, which were significantly enriched for genes harboring disease-associated SNPs. To test if such modules could be used to find novel genes for functional studies, we repeated the analyses using our own gene expression microarray and GWAS data from seasonal allergic rhinitis. We identified a novel gene, FGF2, whose relevance was supported by functional studies using combined small interfering RNA-mediated knock-down and gene expression microarrays. The modules in the 13 complex diseases analyzed here tended to overlap and were enriched for pathways related to oncological, metabolic and inflammatory diseases. This suggested that this union of the modules would be associated with a general increase in susceptibility for complex diseases. Indeed, we found that this union was enriched with GWAS genes for 145 other complex diseases.

Conclusions

Modules of highly interconnected complex disease genes were enriched for disease-associated SNPs, and could be used to find novel genes for functional studies.     

Integrative Genomics Reveals Novel Molecular Pathways and Gene Networks for Coronary Artery Disease

The majority of the heritability of coronary artery disease (CAD) remains unexplained, despite recent successes of genome-wide association studies (GWAS) in identifying novel susceptibility loci. Integrating functional genomic data from a variety of sources with a large-scale meta-analysis of CAD GWAS may facilitate the identification of novel biological processes and genes involved in CAD, as well as clarify the causal relationships of established processes. Towards this end, we integrated 14 GWAS from the CARDIoGRAM Consortium and two additional GWAS from the Ottawa Heart Institute (25,491 cases and 66,819 controls) with 1) genetics of gene expression studies of CAD-relevant tissues in humans, 2) metabolic and signaling pathways from public databases, and 3) data-driven, tissue-specific gene networks from a multitude of human and mouse experiments. We not only detected CAD-associated gene networks of lipid metabolism, coagulation, immunity, and additional networks with no clear functional annotation, but also revealed key driver genes for each CAD network based on the topology of the gene regulatory networks. In particular, we found a gene network involved in antigen processing to be strongly associated with CAD. The key driver genes of this network included glyoxalase I (GLO1) and peptidylprolyl isomerase I (PPIL1), which we verified as regulatory by siRNA experiments in human aortic endothelial cells. Our results suggest genetic influences on a diverse set of both known and novel biological processes that contribute to CAD risk. The key driver genes for these networks highlight potential novel targets for further mechanistic studies and therapeutic interventions.     

Identification of EPHX2 and RMI2 as two novel key genes in cervical squamous cell carcinoma by an integrated bioinformatic analysis

Cervical cancer is the fourth most common malignancy in women worldwide and cervical squamous cell carcinoma (CESC) is the most common histological type of cervical cancer. The dysregulation of genes plays a significant role in cancer. In the present study, we screened out differentially expressed genes (DEGs) of CESC in the GSE63514 data set from the Gene Expression Omnibus database. An integrated bioinformatics analysis was used to select hub genes, as well as to investigate their related prognostic signature, functional annotation, methylation mechanism, and candidate molecular drugs. As a result, a total of 1907 DEGs were identified (944 were upregulated and 963 were downregulated). In the protein–protein interaction network, three hub modules and 30 hub genes were identified. And two hub modules and 116 hub genes were screened out from four CESC-related modules by the weighted gene coexpression network analysis. The gene ontology term enrichment analysis and Kyoto encyclopedia of genes and genomes pathway analysis were performed to better understand functions and pathways. Genes with a significant prognostic value were found by prognostic signature analysis. And there were five genes (EPHX2, CHAF1B, KIAA1524, CDC45, and RMI2) identified as significant CESC-associated genes after expression validation and survival analysis. Among them, EPHX2 and RMI2 were noted as two novel key genes for the CESC-associated methylation and expression. In addition, four candidate small molecule drugs for CESC (camptothecin, resveratrol, vorinostat, and trichostatin A) were defined. Further studies are required to explore these significant CESC-associated genes for their potentiality in diagnosis, prognosis, and targeted therapy.     

LinkNMF: Identification of histone modification modules in the human genome using nonnegative matrix factorization

Histone modifications are ubiquitous processes involved in various cellular mechanisms. Systemic analysis of multiple chromatin modifications has been used to characterize various chromatin states associated with functional DNA elements, gene expression, and specific biological functions. However, identification of modular modification patterns is still required to understand the functional associations between histone modification patterns and specific chromatin/DNA binding factors. To recognize modular modification patterns, we developed a novel algorithm that combines nonnegative matrix factorization (NMF) and a clique-detection algorithm. We applied it, called LinkNMF, to generate a comprehensive modification map in human CD4 + T cell promoter regions. Initially, we identified 11 modules not recognized by conventional approaches. The modules were grouped into two major classes: gene activation and repression. We found that genes targeted by each module were enriched with distinguishable biological functions, suggesting that each modular pattern plays a unique functional role. To explain the formation of modular patterns, we investigated the module-specific binding patterns of chromatin regulators. Application of LinkNMF to histone modification maps of diverse cells and developmental stages will be helpful for understanding how histone modifications regulate gene expression. The algorithm is available on our website at biodb.kaist.ac.kr/LinkNMF.     

Disease gene interaction pathways: a potential framework for how disease genes associate by disease-risk modules

Background

Disease genes that interact cooperatively play crucial roles in the process of complex diseases, yet how to analyze and represent their associations is still an open problem. Traditional methods have failed to represent direct biological evidences that disease genes associate with each other in the pathogenesis of complex diseases. Molecular networks, assumed as ‘a form of biological systems’, consist of a set of interacting biological modules (functional modules or pathways) and this notion could provide a promising insight into deciphering this topic.

Methodology/Principal Findings

In this paper, we hypothesized that disease genes might associate by virtue of the associations between biological modules in molecular networks. Then we introduced a novel disease gene interaction pathway representation and analysis paradigm, and managed to identify the disease gene interaction pathway for 61 known disease genes of coronary artery disease (CAD), which contained 46 disease-risk modules and 182 interaction relationships. As demonstrated, disease genes associate through prescribed communication protocols of common biological functions and pathways.

Conclusions/Significance

Our analysis was proved to be coincident with our primary hypothesis that disease genes of complex diseases interact with their neighbors in a cooperative manner, associate with each other through shared biological functions and pathways of disease-risk modules, and finally cause dysfunctions of a series of biological processes in molecular networks. We hope our paradigm could be a promising method to identify disease gene interaction pathways for other types of complex diseases, affording additional clues in the pathogenesis of complex diseases.     

Network-Based Elucidation of Human Disease Similarities Reveals Common Functional Modules Enriched for Pluripotent Drug Targets

Current work in elucidating relationships between diseases has largely been based on pre-existing knowledge of disease genes. Consequently, these studies are limited in their discovery of new and unknown disease relationships. We present the first quantitative framework to compare and contrast diseases by an integrated analysis of disease-related mRNA expression data and the human protein interaction network. We identified 4,620 functional modules in the human protein network and provided a quantitative metric to record their responses in 54 diseases leading to 138 significant similarities between diseases. Fourteen of the significant disease correlations also shared common drugs, supporting the hypothesis that similar diseases can be treated by the same drugs, allowing us to make predictions for new uses of existing drugs. Finally, we also identified 59 modules that were dysregulated in at least half of the diseases, representing a common disease-state “signature”. These modules were significantly enriched for genes that are known to be drug targets. Interestingly, drugs known to target these genes/proteins are already known to treat significantly more diseases than drugs targeting other genes/proteins, highlighting the importance of these core modules as prime therapeutic opportunities.