Pathway-based Analysis of the Hidden Genetic Heterogeneities in Cancers

Many cancers apparently showing similar phenotypes are actually distinct at the molecular level,leading to very different responses to the same treatment.It has been recently demonstrated that pathway-based approaches are robust and reliable for genetic analysis of cancers.Nevertheless,it remains unclear whether such function-based approaches are useful in deciphering molecular heterogeneities in cancers.Therefore,we aimed to test this possibility in the present study.First,we used a NCI60 dataset to validate the ability of pathways to correctly partition samples.Next,we applied the proposed method to identify the hidden subtypes in diffuse large B-cell lymphoma （DLBCL）.Finally,the clinical significance of the identified subtypes was verified using survival analysis.For the NCI60 dataset,we achieved highly accurate partitions that best fit the clinical cancer phenotypes.Subsequently,for a DLBCL dataset,we identified three hidden subtypes that showed very different 10-year overall survival rates （90％,46％ and 20％） and were highly significantly （P =0.008） correlated with the clinical survival rate.This study demonstrated that the pathwaybased approach is promising for unveiling genetic heterogeneities in complex human diseases.     

Network‐based global inference of human disease genes

Deciphering the genetic basis of human diseases is an important goal of biomedical research. On the basis of the assumption that phenotypically similar diseases are caused by functionally related genes, we propose a computational framework that integrates human protein–protein interactions, disease phenotype similarities, and known gene–phenotype associations to capture the complex relationships between phenotypes and genotypes. We develop a tool named CIPHER to predict and prioritize disease genes, and we show that the global concordance between the human protein network and the phenotype network reliably predicts disease genes. Our method is applicable to genetically uncharacterized phenotypes, effective in the genome‐wide scan of disease genes, and also extendable to explore gene cooperativity in complex diseases. The predicted genetic landscape of over 1000 human phenotypes, which reveals the global modular organization of phenotype–genotype relationships. The genome‐wide prioritization of candidate genes for over 5000 human phenotypes, including those with under‐characterized disease loci or even those lacking known association, is publicly released to facilitate future discovery of disease genes.     

Moving towards a new era of genomics in the neuronal ceroid lipofuscinoses

The neuronal ceroid lipofuscinoses (NCL) are a group of disorders defined by shared clinical and pathological features, including seizures and progressive decline in vision, neurocognition, and motor functioning, as well as accumulation of autofluorescent lysosomal storage material, or ‘ceroid lipofuscin’. Research has revealed thirteen distinct genetic subtypes. Precisely how the gene mutations lead to the clinical phenotype is still incompletely understood, but recent research progress is starting to shed light on disease mechanisms, in both gene-specific and shared pathways. As the application of new sequencing technologies to genetic disease diagnosis has grown, so too has the spectrum of clinical phenotypes caused by mutations in the NCL genes. Most genes causing NCL have probably been identified, underscoring the need for a shift towards applying genomics approaches to achieve a deeper understanding of the molecular basis of the NCLs and related disorders. Here, we summarize the current understanding of the thirteen identified NCL genes and the proteins they encode, touching upon the spectrum of clinical manifestations linked to each of the genes, and we highlight recent progress leading to a broader understanding of key pathways involved in NCL disease pathogenesis and commonalities with other neurodegenerative diseases.     

From phenotype to gene: Detecting disease-specific gene functional modules via a text-based human disease phenotype network construction

Currently, some efforts have been devoted to the text analysis of disease phenotype data, and their results indicated that similar disease phenotypes arise from functionally related genes. These related genes work together, as a functional module, to perform a desired cellular function. We constructed a text-based human disease phenotype network and detected 82 disease-specific gene functional modules, each corresponding to a different phenotype cluster, by means of graph-based clustering and mapping from disease phenotype to gene. Since genes in such gene functional modules are functionally related and cause clinically similar diseases, they may share common genetic origin of their associated disease phenotypes. We believe the investigation may facilitate the ultimate understanding of the common pathophysiologic basis of associated diseases.     

Expression signature of IFN/STAT1 signaling genes predicts poor survival outcome in glioblastoma multiforme in a subtype-specific manner

Previous reports have implicated an induction of genes in IFN/STAT1 (Interferon/STAT1) signaling in radiation resistant and prosurvival tumor phenotypes in a number of cancer cell lines, and we have hypothesized that upregulation of these genes may be predictive of poor survival outcome and/or treatment response in Glioblastoma Multiforme (GBM) patients. We have developed a list of 8 genes related to IFN/STAT1 that we hypothesize to be predictive of poor survival in GBM patients. Our working hypothesis that over-expression of this gene signature predicts poor survival outcome in GBM patients was confirmed, and in addition, it was demonstrated that the survival model was highly subtype-dependent, with strong dependence in the Proneural subtype and no detected dependence in the Classical and Mesenchymal subtypes. We developed a specific multi-gene survival model for the Proneural subtype in the TCGA (the Cancer Genome Atlas) discovery set which we have validated in the TCGA validation set. In addition, we have performed network analysis in the form of Bayesian Network discovery and Ingenuity Pathway Analysis to further dissect the underlying biology of this gene signature in the etiology of GBM. We theorize that the strong predictive value of the IFN/STAT1 gene signature in the Proneural subtype may be due to chemotherapy and/or radiation resistance induced through prolonged constitutive signaling of these genes during the course of the illness. The results of this study have implications both for better prediction models for survival outcome in GBM and for improved understanding of the underlying subtype-specific molecular mechanisms for GBM tumor progression and treatment response.     

Inferring gene-phenotype associations via global protein complex network propagation

Background

Phenotypically similar diseases have been found to be caused by functionally related genes, suggesting a modular organization of the genetic landscape of human diseases that mirrors the modularity observed in biological interaction networks. Protein complexes, as molecular machines that integrate multiple gene products to perform biological functions, express the underlying modular organization of protein-protein interaction networks. As such, protein complexes can be useful for interrogating the networks of phenome and interactome to elucidate gene-phenotype associations of diseases.

Methodology/Principal Findings

We proposed a technique called RWPCN (Random Walker on Protein Complex Network) for predicting and prioritizing disease genes. The basis of RWPCN is a protein complex network constructed using existing human protein complexes and protein interaction network. To prioritize candidate disease genes for the query disease phenotypes, we compute the associations between the protein complexes and the query phenotypes in their respective protein complex and phenotype networks. We tested RWPCN on predicting gene-phenotype associations using leave-one-out cross-validation; our method was observed to outperform existing approaches. We also applied RWPCN to predict novel disease genes for two representative diseases, namely, Breast Cancer and Diabetes.

Conclusions/Significance

Guilt-by-association prediction and prioritization of disease genes can be enhanced by fully exploiting the underlying modular organizations of both the disease phenome and the protein interactome. Our RWPCN uses a novel protein complex network as a basis for interrogating the human phenome-interactome network. As the protein complex network can capture the underlying modularity in the biological interaction networks better than simple protein interaction networks, RWPCN was found to be able to detect and prioritize disease genes better than traditional approaches that used only protein-phenotype associations.     

基于耦合双向聚类技术的DLBCL异质性分析

基因芯片技术为疾病异质性研究提供了有力的工具。当前基于传统聚类分析的方法一般利用芯片上大量基因作为特征来发现疾病的亚型, 因此它们没有考虑到特征中包含的大量无关基因会掩盖有意义的疾病样本的分割。为了避免这个缺点, 提出了基于耦合双向聚类的异质性分析方法(Heterogeneous Analysis Based on Coupled Two-Way Clustering, HCTWC)来搜索有意义的基因簇以便发现样本的内在分割。该方法被应用于弥漫性大B细胞淋巴瘤(diffuse large B-cell lymphoma DLBCL)芯片数据集, 通过识别的基因簇作为特征对DLBCL样本聚类发现生存期分别为55%和25%的两类DLBCL亚型(P<0.05), 因此, HCTWC方法在解决疾病异质性是有效的。     

Integrated analysis of multiple microarray datasets identifies a reproducible survival predictor in ovarian cancer

Background

Public data integration may help overcome challenges in clinical implementation of microarray profiles. We integrated several ovarian cancer datasets to identify a reproducible predictor of survival.

Methodology/Principal Findings

Four microarray datasets from different institutions comprising 265 advanced stage tumors were uniformly reprocessed into a single training dataset, also adjusting for inter-laboratory variation (“batch-effect”). Supervised principal component survival analysis was employed to identify prognostic models. Models were independently validated in a 61-patient cohort using a custom array genechip and a publicly available 229-array dataset. Molecular correspondence of high- and low-risk outcome groups between training and validation datasets was demonstrated using Subclass Mapping. Previously established molecular phenotypes in the 2^nd validation set were correlated with high and low-risk outcome groups. Functional representational and pathway analysis was used to explore gene networks associated with high and low risk phenotypes. A 19-gene model showed optimal performance in the training set (median OS 31 and 78 months, p<0.01), 1^st validation set (median OS 32 months versus not-yet-reached, p = 0.026) and 2^nd validation set (median OS 43 versus 61 months, p = 0.013) maintaining independent prognostic power in multivariate analysis. There was strong molecular correspondence of the respective high- and low-risk tumors between training and 1^st validation set. Low and high-risk tumors were enriched for favorable and unfavorable molecular subtypes and pathways, previously defined in the public 2^nd validation set.

Conclusions/Significance

Integration of previously generated cancer microarray datasets may lead to robust and widely applicable survival predictors. These predictors are not simply a compilation of prognostic genes but appear to track true molecular phenotypes of good- and poor-outcome.     

Systematic screens for human disease genes, from yeast to human and back

Systematic screens for human disease genes have emerged in recent years, due to the wealth of information provided by genome sequences and large scale datasets. Here we review how integration of genomic data in yeast and human is helping to elucidate the genetic basis of mitochondrial diseases. The identification of nearly all yeast mitochondrial proteins and many of their functional interactions provides insight into the role of mitochondria in cellular processes. This information enables prioritization of the candidate genes underlying mitochondrial disorders. In an iterative fashion, the link between predicted human candidate genes and their disease phenotypes can be experimentally tested back in yeast.     

Subcellular alterations that lead to diarrhea during bacterial pathogenesis

Pathogenic microorganisms routinely exploit host cellular functions for their benefit. These alterations often enhance the survival and/or dissemination of the pathogen. However, these effects on the host can be quite debilitating. Consequently, an in-depth understanding of the molecular mechanisms employed by pathogens to manipulate their hosts is crucial. One of the common host phenotypes elicited by enteric pathogens is the generation of diarrhea. Here, we overview the current advances in understanding strategies used by bacterial pathogens to cause diarrheal diseases and discuss how the coordination of various subcellular events can influence disease progression.     

Harnessing the power of yeast to unravel the molecular basis of neurodegeneration

Several neurodegenerative diseases, such as Parkinson's disease (PD), Alzheimer's disease (AD), Huntington's disease (HD), amyotrophic lateral sclerosis (ALS), or prion diseases, are known for their intimate association with protein misfolding and aggregation. These disorders are characterized by the loss of specific neuronal populations in the brain and are highly associated with aging, suggesting a decline in proteostasis capacity may contribute to pathogenesis. Nevertheless, the precise molecular mechanisms that lead to the selective demise of neurons remain poorly understood. As a consequence, appropriate therapeutic approaches and effective treatments are largely lacking. The development of cellular and animal models that faithfully reproduce central aspects of neurodegeneration has been crucial for advancing our understanding of these diseases. Approaches involving the sequential use of different model systems, starting with simpler cellular models and ending with validation in more complex animal models, resulted in the discovery of promising therapeutic targets and small molecules with therapeutic potential. Within this framework, the simple and well‐characterized eukaryote Saccharomyces cerevisiae, also known as budding yeast, is being increasingly used to study the molecular basis of several neurodegenerative disorders. Yeast provides an unprecedented toolbox for the dissection of complex biological processes and pathways. Here, we summarize how yeast models are adding to our current understanding of several neurodegenerative disorders.     

High content screening in neurodegenerative diseases

The functional annotation of genomes, construction of molecular networks and novel drug target identification, are important challenges that need to be addressed as a matter of great urgency. Multiple complementary 'omics' approaches have provided clues as to the genetic risk factors and pathogenic mechanisms underlying numerous neurodegenerative diseases, but most findings still require functional validation. For example, a recent genome wide association study for Parkinson's Disease (PD), identified many new loci as risk factors for the disease, but the underlying causative variant(s) or pathogenic mechanism is not known. As each associated region can contain several genes, the functional evaluation of each of the genes on phenotypes associated with the disease, using traditional cell biology techniques would take too long. There is also a need to understand the molecular networks that link genetic mutations to the phenotypes they cause. It is expected that disease phenotypes are the result of multiple interactions that have been disrupted. Reconstruction of these networks using traditional molecular methods would be time consuming. Moreover, network predictions from independent studies of individual components, the reductionism approach, will probably underestimate the network complexity. This underestimation could, in part, explain the low success rate of drug approval due to undesirable or toxic side effects. Gaining a network perspective of disease related pathways using HT/HC cellular screening approaches, and identifying key nodes within these pathways, could lead to the identification of targets that are more suited for therapeutic intervention. High-throughput screening (HTS) is an ideal methodology to address these issues. but traditional methods were one dimensional whole-well cell assays, that used simplistic readouts for complex biological processes. They were unable to simultaneously quantify the many phenotypes observed in neurodegenerative diseases such as axonal transport deficits or alterations in morphology properties. This approach could not be used to investigate the dynamic nature of cellular processes or pathogenic events that occur in a subset of cells. To quantify such features one has to move to multi-dimensional phenotypes termed high-content screening (HCS). HCS is the cell-based quantification of several processes simultaneously, which provides a more detailed representation of the cellular response to various perturbations compared to HTS. HCS has many advantages over HTS, but conducting a high-throughput (HT)-high-content (HC) screen in neuronal models is problematic due to high cost, environmental variation and human error. In order to detect cellular responses on a 'phenomics' scale using HC imaging one has to reduce variation and error, while increasing sensitivity and reproducibility. Herein we describe a method to accurately and reliably conduct shRNA screens using automated cell culturing and HC imaging in neuronal cellular models. We describe how we have used this methodology to identify modulators for one particular protein, DJ1, which when mutated causes autosomal recessive parkinsonism. Combining the versatility of HC imaging with HT methods, it is possible to accurately quantify a plethora of phenotypes. This could subsequently be utilized to advance our understanding of the genome, the pathways involved in disease pathogenesis as well as identify potential therapeutic targets.     

Identifying core gene modules in glioblastoma based on multilayer factor-mediated dysfunctional regulatory networks through integrating multi-dimensional genomic data

The driver genetic aberrations collectively regulate core cellular processes underlying cancer development. However, identifying the modules of driver genetic alterations and characterizing their functional mechanisms are still major challenges for cancer studies. Here, we developed an integrative multi-omics method CMDD to identify the driver modules and their affecting dysregulated genes through characterizing genetic alteration-induced dysregulated networks. Applied to glioblastoma (GBM), the CMDD identified a core gene module of 17 genes, including seven known GBM drivers, and their dysregulated genes. The module showed significant association with shorter survival of GBM. When classifying driver genes in the module into two gene sets according to their genetic alteration patterns, we found that one gene set directly participated in the glioma pathway, while the other indirectly regulated the glioma pathway, mostly, via their dysregulated genes. Both of the two gene sets were significant contributors to survival and helpful for classifying GBM subtypes, suggesting their critical roles in GBM pathogenesis. Also, by applying the CMDD to other six cancers, we identified some novel core modules associated with overall survival of patients. Together, these results demonstrate integrative multi-omics data can identify driver modules and uncover their dysregulated genes, which is useful for interpreting cancer genome.     

Network-assisted investigation of combined causal signals from genome-wide association studies in schizophrenia

With the recent success of genome-wide association studies (GWAS), a wealth of association data has been accomplished for more than 200 complex diseases/traits, proposing a strong demand for data integration and interpretation. A combinatory analysis of multiple GWAS datasets, or an integrative analysis of GWAS data and other high-throughput data, has been particularly promising. In this study, we proposed an integrative analysis framework of multiple GWAS datasets by overlaying association signals onto the protein-protein interaction network, and demonstrated it using schizophrenia datasets. Building on a dense module search algorithm, we first searched for significantly enriched subnetworks for schizophrenia in each single GWAS dataset and then implemented a discovery-evaluation strategy to identify module genes with consistent association signals. We validated the module genes in an independent dataset, and also examined them through meta-analysis of the related SNPs using multiple GWAS datasets. As a result, we identified 205 module genes with a joint effect significantly associated with schizophrenia; these module genes included a number of well-studied candidate genes such as DISC1, GNA12, GNA13, GNAI1, GPR17, and GRIN2B. Further functional analysis suggested these genes are involved in neuronal related processes. Additionally, meta-analysis found that 18 SNPs in 9 module genes had P _meta<1×10⁻⁴, including the gene HLA-DQA1 located in the MHC region on chromosome 6, which was reported in previous studies using the largest cohort of schizophrenia patients to date. These results demonstrated our bi-directional network-based strategy is efficient for identifying disease-associated genes with modest signals in GWAS datasets. This approach can be applied to any other complex diseases/traits where multiple GWAS datasets are available.     

Epistatic interaction maps relative to multiple metabolic phenotypes

An epistatic interaction between two genes occurs when the phenotypic impact of one gene depends on another gene, often exposing a functional association between them. Due to experimental scalability and to evolutionary significance, abundant work has been focused on studying how epistasis affects cellular growth rate, most notably in yeast. However, epistasis likely influences many different phenotypes, affecting our capacity to understand cellular functions, biochemical networks adaptation, and genetic diseases. Despite its broad significance, the extent and nature of epistasis relative to different phenotypes remain fundamentally unexplored. Here we use genome-scale metabolic network modeling to investigate the extent and properties of epistatic interactions relative to multiple phenotypes. Specifically, using an experimentally refined stoichiometric model for Saccharomyces cerevisiae, we computed a three-dimensional matrix of epistatic interactions between any two enzyme gene deletions, with respect to all metabolic flux phenotypes. We found that the total number of epistatic interactions between enzymes increases rapidly as phenotypes are added, plateauing at approximately 80 phenotypes, to an overall connectivity that is roughly 8-fold larger than the one observed relative to growth alone. Looking at interactions across all phenotypes, we found that gene pairs interact incoherently relative to different phenotypes, i.e. antagonistically relative to some phenotypes and synergistically relative to others. Specific deletion-deletion-phenotype triplets can be explained metabolically, suggesting a highly informative role of multi-phenotype epistasis in mapping cellular functions. Finally, we found that genes involved in many interactions across multiple phenotypes are more highly expressed, evolve slower, and tend to be associated with diseases, indicating that the importance of genes is hidden in their total phenotypic impact. Our predictions indicate a pervasiveness of nonlinear effects in how genetic perturbations affect multiple metabolic phenotypes. The approaches and results reported could influence future efforts in understanding metabolic diseases and the role of biochemical regulation in the cell.     

Bioinformatics role of the WGCNA analysis and co-expression network identifies of prognostic marker in lung cancer

Lung cancer is the most talked about cancer in the world. It is also one of the cancers that currently has a high mortality rate. The aim of our research is to find more effective therapeutic targets and prognostic markers for human lung cancer. First, we download gene expression data from the GEO database. We performed weighted co-expression network analysis on the selected genes, we then constructed scale-free networks and topological overlap matrices, and performed correlation modular analysis with the cancer group. We screened the 200 genes with the highest correlation in the cyan module for functional enrichment analysis and protein interaction network construction, found that most of them focused on cell division, tumor necrosis factor-mediated signaling pathways, cellular redox homeostasis, reactive oxygen species biosynthesis, and other processes, and were related to the cell cycle, apoptosis, HIF-1 signaling pathway, p53 signaling pathway, NF-κB signaling pathway, and several cancer disease pathways are involved. Finally, we used the GEPIA website data to perform survival analysis on some of the genes with GS > 0.6 in the cyan module. CBX3, AHCY, MRPL12, TPGB, TUBG1, KIF11, LRRC59, MRPL17, TMEM106B, ZWINT, TRIP13, and HMMR was identified as an important prognostic factor for lung cancer patients. In summary, we identified 12 mRNAs associated with lung cancer prognosis. Our study contributes to a deeper understanding of the molecular mechanisms of lung cancer and provides new insights into drug use and prognosis.