Chunlong Zhang

Subpathway-GM: identification of metabolic subpathways via joint power of interesting genes and metabolites and their topologies within pathways

Various ‘omics’ technologies, including microarrays and gas chromatography mass spectrometry, can be used to identify hundreds of interesting genes, proteins and metabolites, such as differential genes, proteins and metabolites associated with diseases. Identifying metabolic pathways has become an invaluable aid to understanding the genes and metabolites associated with studying conditions. However, the classical methods used to identify pathways fail to accurately consider joint power of interesting gene/metabolite and the key regions impacted by them within metabolic pathways. In this study, we propose a powerful analytical method referred to as Subpathway-GM for the identification of metabolic subpathways. This provides a more accurate level of pathway analysis by integrating information from genes and metabolites, and their positions and cascade regions within the given pathway. We analyzed two colorectal cancer and one metastatic prostate cancer data sets and demonstrated that Subpathway-GM was able to identify disease-relevant subpathways whose corresponding entire pathways might be ignored using classical entire pathway identification methods. Further analysis indicated that the power of a joint genes/metabolites and subpathway strategy based on their topologies may play a key role in reliably recalling disease-relevant subpathways and finding novel subpathways.

Topologically inferring risk-active pathways toward precise cancer classification by directed random walk

MOTIVATION The accurate prediction of disease status is a central challenge in clinical cancer research. Microarray-based gene biomarkers have been identified to predict outcome and outperform traditional clinical parameters. However, the robustness of the individual gene biomarkers is questioned because of their little reproducibility between different cohorts of patients. Substantial progress in treatment requires advances in methods to identify robust biomarkers. Several methods incorporating pathway information have been proposed to identify robust pathway markers and build classifiers at the level of functional categories rather than of individual genes. However, current methods consider the pathways as simple gene sets but ignore the pathway topological information, which is essential to infer a more robust pathway activity. RESULTS Here, we propose a directed random walk (DRW)-based method to infer the pathway activity. DRW evaluates the topological importance of each gene by capturing the structure information embedded in the directed pathway network. The strategy of weighting genes by their topological importance greatly improved the reproducibility of pathway activities. Experiments on 18 cancer datasets showed that the proposed method yielded a more accurate and robust overall performance compared with several existing gene-based and pathway-based classification methods. The resulting risk-active pathways are more reliable in guiding therapeutic selection and the development of pathway-specific therapeutic strategies. AVAILABILITY DRW is freely available at http://210.46.85.180:8080/DRWPClass/

Characterizing the Network of Drugs and Their Affected Metabolic Subpathways

A fundamental issue in biology and medicine is illustration of the overall drug impact which is always the consequence of changes in local regions of metabolic pathways (subpathways). To gain insights into the global relationship between drugs and their affected metabolic subpathways, we constructed a drug–metabolic subpathway network (DRSN). This network included 3925 significant drug–metabolic subpathway associations representing drug dual effects. Through analyses based on network biology, we found that if drugs were linked to the same subpathways in the DRSN, they tended to share the same indications and side effects. Furthermore, if drugs shared more subpathways, they tended to share more side effects. We then calculated the association score by integrating drug-affected subpathways and disease-related subpathways to quantify the extent of the associations between each drug class and disease class. The results showed some close drug–disease associations such as sex hormone drugs and cancer suggesting drug dual effects. Surprisingly, most drugs displayed close associations with their side effects rather than their indications. To further investigate the mechanism of drug dual effects, we classified all the subpathways in the DRSN into therapeutic and non-therapeutic subpathways representing drug therapeutic effects and side effects. Compared to drug side effects, the therapeutic effects tended to work through tissue-specific genes and these genes tend to be expressed in the adrenal gland, liver and kidney; while drug side effects always occurred in the liver, bone marrow and trachea. Taken together, the DRSN could provide great insights into understanding the global relationship between drugs and metabolic subpathways.

Identification of miRNA-Mediated Core Gene Module for Glioma Patient Prediction by Integrating High-Throughput miRNA, mRNA Expression and Pathway Structure

The prognosis of glioma patients is usually poor, especially in patients with glioblastoma (World Health Organization (WHO) grade IV). The regulatory functions of microRNA (miRNA) on genes have important implications in glioma cell survival. However, there are not many studies that have investigated glioma survival by integrating miRNAs and genes while also considering pathway structure. In this study, we performed sample-matched miRNA and mRNA expression profilings to systematically analyze glioma patient survival. During this analytical process, we developed pathway-based random walk to identify a glioma core miRNA-gene module, simultaneously considering pathway structure information and multi-level involvement of miRNAs and genes. The core miRNA-gene module we identified was comprised of four apparent sub-modules; all four sub-modules displayed a significant correlation with patient survival in the testing set (P-values≤0.001). Notably, one sub-module that consisted of 6 miRNAs and 26 genes also correlated with survival time in the high-grade subgroup (WHO grade III and IV), P-value = 0.0062. Furthermore, the 26-gene expression signature from this sub-module had robust predictive power in four independent, publicly available glioma datasets. Our findings suggested that the expression signatures, which were identified by integration of miRNA and gene level, were closely associated with overall survival among the glioma patients with various grades.

The Identification of Specific Methylation Patterns across Different Cancers

Abnormal DNA methylation is known as playing an important role in the tumorgenesis. It is helpful for distinguishing the specificity of diagnosis and therapeutic targets for cancers based on characteristics of DNA methylation patterns across cancers. High throughput DNA methylation analysis provides the possibility to comprehensively filter the epigenetics diversity across various cancers. We integrated whole-genome methylation data detected in 798 samples from seven cancers. The hierarchical clustering revealed the existence of cancer-specific methylation pattern. Then we identified 331 differentially methylated genes across these cancers, most of which (266) were specifically differential methylation in unique cancer. A DNA methylation correlation network (DMCN) was built based on the methylation correlation between these genes. It was shown the hubs in the DMCN were inclined to cancer-specific genes in seven cancers. Further survival analysis using the part of genes in the DMCN revealed high-risk group and low-risk group were distinguished by seven biomarkers (PCDHB15, WBSCR17, IGF1, GYPC, CYGB, ACTG2, and PRRT1) in breast cancer and eight biomarkers (ZBTB32, OR51B4, CCL8, TMEFF2, SALL3, GPSM1, MAGEA8, and SALL1) in colon cancer, respectively. At last, a protein-protein interaction network was introduced to verify the biological function of differentially methylated genes. It was shown that MAP3K14, PTN, ACVR1 and HCK sharing different DNA methylation and gene expression across cancers were relatively high degree distribution in PPI network. The study suggested that not only the identified cancer-specific genes provided reference for individual treatment but also the relationship across cancers could be explained by differential DNA methylation.

ESEA: Discovering the Dysregulated Pathways based on Edge Set Enrichment Analysis

Pathway analyses are playing an increasingly important role in understanding biological mechanism, cellular function and disease states. Current pathway-identification methods generally focus on only the changes of gene expression levels; however, the biological relationships among genes are also the fundamental components of pathways, and the dysregulated relationships may also alter the pathway activities. We propose a powerful computational method, Edge Set Enrichment Analysis (ESEA), for the identification of dysregulated pathways. This provides a novel way of pathway analysis by investigating the changes of biological relationships of pathways in the context of gene expression data. Simulation studies illustrate the power and performance of ESEA under various simulated conditions. Using real datasets from p53 mutation, Type 2 diabetes and lung cancer, we validate effectiveness of ESEA in identifying dysregulated pathways. We further compare our results with five other pathway enrichment analysis methods. With these analyses, we show that ESEA is able to help uncover dysregulated biological pathways underlying complex traits and human diseases via specific use of the dysregulated biological relationships. We develop a freely available R-based tool of ESEA. Currently, ESEA can support pathway analysis of the seven public databases (KEGG; Reactome; Biocarta; NCI; SPIKE; HumanCyc; Panther).

Subpathway-GMir: identifying miRNA-mediated metabolic subpathways by integrating condition-specific genes, microRNAs, and pathway topologies.

MicroRNAs (miRNAs) regulate disease-relevant metabolic pathways. However, most current pathway identification methods fail to consider miRNAs in addition to genes when analyzing pathways. We developed a powerful method called Subpathway-GMir to construct miRNA-regulated metabolic pathways and to identify miRNA-mediated subpathways by considering condition-specific genes, miRNAs, and pathway topologies. We used Subpathway-GMir to analyze two liver hepatocellular carcinomas (LIHC), one stomach adenocarcinoma (STAD), and one type 2 diabetes (T2D) data sets. Results indicate that Subpathway-GMir is more effective in identifying phenotype-associated metabolic pathways than other methods and our results are reproducible and robust. Subpathway-GMir provides a flexible platform for identifying abnormal metabolic subpathways mediated by miRNAs, and may help to clarify the roles that miRNAs play in a variety of diseases. The Subpathway-GMir method has been implemented as a freely available R package.

Identifying disease related sub-pathways for analysis of genome-wide association studies

Most methods for genome-wide association studies (GWAS) focus on discovering a single genetic variant, but the pathogenesis of complex diseases is thought to arise from the joint effect of multiple genetic variants. Information about pathway structure, such as the interactions and distances between gene products within pathways, can help us learn more about the functions and joint effect of genes associated with disease risk. We developed a novel sub-pathway based approach to study the joint effect of multiple genetic variants that are modestly associated with disease. The approach prioritized sub-pathways based on the significance values of single nucleotide polymorphisms (SNPs) and the interactions and distances between gene products within pathways. We applied the method to seven complex diseases. The result showed that our method can efficiently identify statistically significant sub-pathways associated with the pathogenesis of complex diseases. The approach identified sub-pathways that may inform the interpretation of GWAS data.

Topologically inferring pathway activity toward precise cancer classification via integrating genomic and metabolomic data: prostate cancer as a case

Precise cancer classification is a central challenge in clinical cancer research such as diagnosis, prognosis and metastasis prediction. Most existing cancer classification methods based on gene or metabolite biomarkers were limited to single genomics or metabolomics, and lacked integration and utilization of multiple ‘omics’ data. The accuracy and robustness of these methods when applied to independent cohorts of patients must be improved. In this study, we propose a directed random walk-based method to evaluate the topological importance of each gene in a reconstructed gene–metabolite graph by integrating information from matched gene expression profiles and metabolomic profiles. The joint use of gene and metabolite information contributes to accurate evaluation of the topological importance of genes and reproducible pathway activities. We constructed classifiers using reproducible pathway activities for precise cancer classification and risk metabolic pathway identification. We applied the proposed method to the classification of prostate cancer. Within-dataset experiments and cross-dataset experiments on three independent datasets demonstrated that the proposed method achieved a more accurate and robust overall performance compared to several existing classification methods. The resulting risk pathways and topologically important differential genes and metabolites provide biologically informative models for prostate cancer prognosis and therapeutic strategies development.

Chromatin modifications and genomic contexts linked to dynamic DNA methylation patterns across human cell types

DNA methylation is related closely to sequence contexts and chromatin modifications; however, their potential differences in different genomic regions across cell types remain largely unexplored. We used publicly available genome-scale DNA methylation and histone modification profiles to study their relationships among different genomic regions in human embryonic stem cells (H1), H1-derived neuronal progenitor cultured cells (NPC), and foetal fibroblasts (IMR90) using the Random forests classifier. Histone modifications achieved high accuracy in modelling DNA methylation patterns on a genome scale in the three cell types. The inclusion of sequence features helped improve accuracy only in non-promoter regions of IMR90. Furthermore, the top six feature combinations obtained by mean decrease Gini were important indicators of different DNA methylation patterns, suggesting that H3K4me2 and H3K4me3 are important indicators that are independent of genomic regions and cell types. H3K9me3 was IMR90-specific and exhibited a genomic region-specific correlation with DNA methylation. Variations of essential chromatin modification signals may effectively discriminate changes of DNA methylation between H1 and IMR90. Genes with different co-variations of epigenetic marks exhibited genomic region-specific biological relevance. This study provides an integrated strategy to identify systematically essential epigenetic and genetic elements of genomic region-specific and cell type-specific DNA methylation patterns.