Using a Network-Based Analysis Approach to Investigate the Involvement of S. aureus in the Pathogenesis of Granulomatosis with Polyangiitis.

Chronic nasal carriage of Staphylococcus aureus (SA) has been shown to be significantly higher in GPA patients when compared to healthy subjects, as well as being associated with increased endonasal activity and disease relapse. The aim of this study was to investigate SA involvement in GPA by applying a network-based analysis (NBA) approach to publicly available nasal transcriptomic data. Using these data, our NBA pipeline generated a proteinase 3 (PR3) positive ANCA associated vasculitis (AAV) disease network integrating differentially expressed genes, dysregulated transcription factors (TFs), disease-specific genes derived from GWAS studies, drug-target and protein-protein interactions. The PR3+ AAV disease network captured genes previously reported to be dysregulated in AAV associated. A subnetwork focussing on interactions between SA virulence factors and enriched biological processes revealed potential mechanisms for SA's involvement in PR3+ AAV. Immunosuppressant treatment reduced differential expression and absolute TF activities in this subnetwork for patients with inactive nasal disease but not active nasal disease symptoms at the time of sampling. The disease network generated identified the key molecular signatures and highlighted the associated biological processes in PR3+ AAV and revealed potential mechanisms for SA to affect these processes.

Target Identification Using Homopharma and Network-Based Methods for Predicting Compounds Against Dengue Virus-Infected Cells.

Drug target prediction is an important method for drug discovery and design, can disclose the potential inhibitory effect of active compounds, and is particularly relevant to many diseases that have the potential to kill, such as dengue, but lack any healing agent. An antiviral drug is urgently required for dengue treatment. Some potential antiviral agents are still in the process of drug discovery, but the development of more effective active molecules is in critical demand. Herein, we aimed to provide an efficient technique for target prediction using homopharma and network-based methods, which is reliable and expeditious to hunt for the possible human targets of three phenolic lipids (anarcardic acid, cardol, and cardanol) related to dengue viral (DENV) infection as a case study. Using several databases, the similarity search and network-based analyses were applied on the three phenolic lipids resulting in the identification of seven possible targets as follows. Based on protein annotation, three phenolic lipids may interrupt or disturb the human proteins, namely KAT5, GAPDH, ACTB, and HSP90AA1, whose biological functions have been previously reported to be involved with viruses in the family Flaviviridae. In addition, these phenolic lipids might inhibit the mechanism of the viral proteins: NS3, NS5, and E proteins. The DENV and human proteins obtained from this study could be potential targets for further molecular optimization on compounds with a phenolic lipid core structure in anti-dengue drug discovery. As such, this pipeline could be a valuable tool to identify possible targets of active compounds.

Why imaging data alone is not enough: AI-based integration of imaging, omics, and clinical data.

Artificial intelligence (AI) is currently regaining enormous interest due to the success of machine learning (ML), and in particular deep learning (DL). Image analysis, and thus radiomics, strongly benefits from this research. However, effectively and efficiently integrating diverse clinical, imaging, and molecular profile data is necessary to understand complex diseases, and to achieve accurate diagnosis in order to provide the best possible treatment. In addition to the need for sufficient computing resources, suitable algorithms, models, and data infrastructure, three important aspects are often neglected: (1) the need for multiple independent, sufficiently large and, above all, high-quality data sets; (2) the need for domain knowledge and ontologies; and (3) the requirement for multiple networks that provide relevant relationships among biological entities. While one will always get results out of high-dimensional data, all three aspects are essential to provide robust training and validation of ML models, to provide explainable hypotheses and results, and to achieve the necessary trust in AI and confidence for clinical applications.

Identification of cancer omics commonality and difference via community fusion.

The analysis of cancer omics data is a "classic" problem; however, it still remains challenging. Advancing from early studies that are mostly focused on a single type of cancer, some recent studies have analyzed data on multiple "related" cancer types/subtypes, examined their commonality and difference, and led to insightful findings. In this article, we consider the analysis of multiple omics datasets, with each dataset on one type/subtype of "related" cancers. A Community Fusion (CoFu) approach is developed, which conducts marker selection and model building using a novel penalization technique, informatively accommodates the network community structure of omics measurements, and automatically identifies the commonality and difference of cancer omics markers. Simulation demonstrates its superiority over direct competitors. The analysis of TCGA lung cancer and melanoma data leads to interesting findings.

Brain galanin system genes interact with life stresses in depression-related phenotypes.

Galanin is a stress-inducible neuropeptide and cotransmitter in serotonin and norepinephrine neurons with a possible role in stress-related disorders. Here we report that variants in genes for galanin (GAL) and its receptors (GALR1, GALR2, GALR3), despite their disparate genomic loci, conferred increased risk of depression and anxiety in people who experienced childhood adversity or recent negative life events in a European white population cohort totaling 2,361 from Manchester, United Kingdom and Budapest, Hungary. Bayesian multivariate analysis revealed a greater relevance of galanin system genes in highly stressed subjects compared with subjects with moderate or low life stress. Using the same method, the effect of the galanin system genes was stronger than the effect of the well-studied 5-HTTLPR polymorphism in the serotonin transporter gene (SLC6A4). Conventional multivariate analysis using general linear models demonstrated that interaction of galanin system genes with life stressors explained more variance (1.7%, P = 0.005) than the life stress-only model. This effect replicated in independent analysis of the Manchester and Budapest subpopulations, and in males and females. The results suggest that the galanin pathway plays an important role in the pathogenesis of depression in humans by increasing the vulnerability to early and recent psychosocial stress. Correcting abnormal galanin function in depression could prove to be a novel target for drug development. The findings further emphasize the importance of modeling environmental interaction in finding new genes for depression.

Gene signature-based mapping of immunological systems and diseases.

BACKGROUND: The immune system is multifaceted, structured by diverse components that interconnect using multilayered dynamic cellular processes. Genomic technologies provide a means for investigating, at the molecular level, the adaptations of the immune system in host defense and its dysregulation in pathological conditions. A critical aspect of intersecting and investigating complex datasets is determining how to best integrate genomic data from diverse platforms and heterogeneous sample populations to capture immunological signatures in health and disease. RESULT: We focus on gene signatures, representing highly enriched genes of immune cell subsets from both diseased and healthy tissues. From these, we construct a series of biomaps that illustrate the molecular linkages between cell subsets from different lineages, the connectivity between different immunological diseases, and the enrichment of cell subset signatures in diseased tissues. Finally, we overlay the downstream genes of drug targets with disease gene signatures to display the potential therapeutic applications for these approaches. CONCLUSION: An in silico approach has been developed to characterize immune cell subsets and diseases based on the gene signatures that most differentiate them from other biological states. This modular 'biomap' reveals the linkages between different diseases and immune subtypes, and provides evidence for the presence of specific immunocyte subsets in mixed tissues. The over-represented genes in disease signatures of interest can be further investigated for their functions in both host defense and disease.

Network-based analysis for identification of candidate genes for colorectal cancer progression.

Although high-throughput biological technologies have been producing a vast amount of multi-omics data regarding cancer genomics and several disease susceptible genes have been reported, many of these genes are likely to be irrelevant for the cancer process because only one feature of the tumor pathway could be focused on. By identifying 'CpG core', which was extracted from CpG sites in genomic DNA by our newly developed method, we performed integrated analysis using gene expression and DNA methylation profiles of 116 colorectal cancer samples. First, based on gene expression values, colorectal cancer samples were divided into three clusters (Cluster-1, -2, and -3) by k-means clustering. The 5-year overall survival rates of colorectal cancer patients were 74.8%, 29.2%, and 29.4% in Cluster-1, -2, and -3, respectively, and the prognosis of Cluster-2 was significantly poorer than that of the other two clusters owing to liver metastasis (P < 0.001). Second, each cluster was divided into two subgroups based on methylation status, and the 5-year overall survival rate of Cluster-1H (36.8%) was significantly shorter than that of Cluster-1L (96.1%) due to the accumulation of aberrant DNA methylation (P = 0.014). Third, network-based analysis using expression and methylation profiles demonstrated that nucleoporin family genes were downregulated in Cluster-2 and that the PTX3 gene was highly methylated in Cluster-1H. These combined data indicate that integrated analysis can identify disease characteristics that would be missed using single comprehensive analysis, and that multiple pathways would play pivotal roles in the liver metastasis of colorectal cancer.

TIST: Transcriptome and Histopathological Image Integrative Analysis for Spatial Transcriptomics.

Sequencing-based spatial transcriptomics (ST) is an emerging technology to study in situ gene expression patterns at the whole-genome scale. Currently, ST data analysis is still complicated by high technical noises and low resolution. In addition to the transcriptomic data, matched histopathological images are usually generated for the same tissue sample along the ST experiment. The matched high-resolution histopathological images provide complementary cellular phenotypical information, providing an opportunity to mitigate the noises in ST data. We present a novel ST data analysis method called transcriptome and histopathological image integrative analysis for ST (TIST), which enables the identification of spatial clusters (SCs) and the enhancement of spatial gene expression patterns by integrative analysis of matched transcriptomic data and images. TIST devises a histopathological feature extraction method based on Markov random field (MRF) to learn the cellular features from histopathological images, and integrates them with the transcriptomic data and location information as a network, termed TIST-net. Based on TIST-net, SCs are identified by a random walk-based strategy, and gene expression patterns are enhanced by neighborhood smoothing. We benchmark TIST on both simulated datasets and 32 real samples against several state-of-the-art methods. Results show that TIST is robust to technical noises on multiple analysis tasks for sequencing-based ST data and can find interesting microstructures in different biological scenarios. TIST is available at http://lifeome.net/software/tist/ and https://ngdc.cncb.ac.cn/biocode/tools/BT007317.

Identifying the Risk of Sepsis in Patients With Cancer Using Digital Health Care Records: Machine Learning-Based Approach.

BACKGROUND: Sepsis is diagnosed in millions of people every year, resulting in a high mortality rate. Although patients with sepsis present multimorbid conditions, including cancer, sepsis predictions have mainly focused on patients with severe injuries. OBJECTIVE: In this paper, we present a machine learning-based approach to identify the risk of sepsis in patients with cancer using electronic health records (EHRs). METHODS: We utilized deidentified anonymized EHRs of 8580 patients with cancer from the Samsung Medical Center in Korea in a longitudinal manner between 2014 and 2019. To build a prediction model based on physical status that would differ between sepsis and nonsepsis patients, we analyzed 2462 laboratory test results and 2266 medication prescriptions using graph network and statistical analyses. The medication relationships and lab test results from each analysis were used as additional learning features to train our predictive model. RESULTS: Patients with sepsis showed differential medication trajectories and physical status. For example, in the network-based analysis, narcotic analgesics were prescribed more often in the sepsis group, along with other drugs. Likewise, 35 types of lab tests, including albumin, globulin, and prothrombin time, showed significantly different distributions between sepsis and nonsepsis patients (P<.001). Our model outperformed the model trained using only common EHRs, showing an improved accuracy, area under the receiver operating characteristic (AUROC), and F1 score by 11.9%, 11.3%, and 13.6%, respectively. For the random forest-based model, the accuracy, AUROC, and F1 score were 0.692, 0.753, and 0.602, respectively. CONCLUSIONS: We showed that lab tests and medication relationships can be used as efficient features for predicting sepsis in patients with cancer. Consequently, identifying the risk of sepsis in patients with cancer using EHRs and machine learning is feasible.

Identification of Key Genes in 'Luang Pratahn', Thai Salt-Tolerant Rice, Based on Time-Course Data and Weighted Co-expression Networks.

Salinity is an important environmental factor causing a negative effect on rice production. To prevent salinity effects on rice yields, genetic diversity concerning salt tolerance must be evaluated. In this study, we investigated the salinity responses of rice (Oryza sativa) to determine the critical genes. The transcriptomes of 'Luang Pratahn' rice, a local Thai rice variety with high salt tolerance, were used as a model for analyzing and identifying the key genes responsible for salt-stress tolerance. Based on 3' Tag-Seq data from the time course of salt-stress treatment, weighted gene co-expression network analysis was used to identify key genes in gene modules. We obtained 1,386 significantly differentially expressed genes in eight modules. Among them, six modules indicated a significant correlation within 6, 12, or 48h after salt stress. Functional and pathway enrichment analysis was performed on the co-expressed genes of interesting modules to reveal which genes were mainly enriched within important functions for salt-stress responses. To identify the key genes in salt-stress responses, we considered the two-state co-expression networks, normal growth conditions, and salt stress to investigate which genes were less important in a normal situation but gained more impact under stress. We identified key genes for the response to biotic and abiotic stimuli and tolerance to salt stress. Thus, these novel genes may play important roles in salinity tolerance and serve as potential biomarkers to improve salt tolerance cultivars.

Weighted Gene Coregulation Network Analysis of Promoter DNA Methylation on All-Cause Mortality in Old-Aged Birth Cohorts Finds Modules of High-Risk Associated Biomarkers.

Overall or all-cause mortality is a key measure of health in a population. Multiple epigenome-wide association studies have been conducted on all-cause mortality with limited significant findings and low replication. To elucidate the coregulated DNA methylation patterns associated with all-cause mortality, we conducted a weighted DNA methylation coregulation network analysis on whole-blood samples of 1,425 older individuals from the Lothian Birth Cohorts of 1921 and 1936. Our network-based analysis defined coregulated DNA methylation patterns in gene promoters into clusters or modules whose correlation with all-cause mortality was assessed by survival analysis. We found two significant modules or gene clusters associated with all-cause mortality in LBC1921 based on their eigengenes; one negatively correlated (p = 8.14E-03, 698 genes) and one positively correlated (p = 4.26E-02, 1,431 genes) with the risk of death. The two modules were replicated in LBC1936 with the same directions of correlation (p = 6.35E-02 and p = 3.64E-02, respectively). Furthermore, the modules revealed 32 genes associated with all-cause mortality (FDR < 0.05) linked to various diseases, including cancer and diabetes. Additionally, we performed pathway analysis and found 22 pathways (FDR < 0.05), including a pathway for taste transduction, which has been shown to be associated with poor prognosis in acutely hospitalized patients, and several pathways were linked to different types of cancer. The results from our network analysis show that DNA methylation of multiple genes could have been coregulated in an association with the overall risk of death. The identified epigenetic markers might help with our understanding of the molecular basis of all-cause mortality and general health.

Gene Expression Network Analysis of Precursor Lesions in Familial Pancreatic Cancer.

Purpose: High-grade pancreatic intraepithelial neoplasia (PanIN) are aggressive premalignant lesions, associated with risk of progression to pancreatic ductal adenocarcinoma (PDAC). A depiction of co-dysregulated gene activity in high-grade familial pancreatic cancer (FPC)-related PanIN lesions may characterize the molecular events during the progression from familial PanIN to PDAC. Materials and Methods: We performed weighted gene coexpression network analysis (WGCNA) to identify clusters of coexpressed genes associated with FPC-related PanIN lesions in 13 samples with PanIN-2/3 from FPC predisposed individuals, 6 samples with PDAC from sporadic pancreatic cancer (SPC) patients, and 4 samples of normal donor pancreatic tissue. Results: WGCNA identified seven differentially expressed gene (DEG) modules and two commonly expressed gene (CEG) modules with significant enrichment for Gene Ontology (GO) terms in FPC and SPC, including three upregulated (p < 5e-05) and four downregulated (p < 6e-04) gene modules in FPC compared to SPC. Among the DEG modules, the upregulated modules include 14 significant genes (p < 1e-06): ALOX12-AS1, BCL2L11, EHD4, C4B, BTN3A3, NDUFA11, RBM4B, MYOC, ZBTB47, TTTY15, NAPRT, LOC102606465, LOC100505711, and PTK2. The downregulated modules include 170 genes (p < 1e-06), among them 13 highly significant genes (p < 1e-10): COL10A1, SAMD9, PLPP4, COMP, POSTN, IGHV4-31, THBS2, MMP9, FNDC1, HOPX, TMEM200A, INHBA, and SULF1. The DEG modules are enriched for GO terms related to mitochondrial structure and adenosine triphosphate metabolic processes, extracellular structure and binding properties, humoral and complement mediated immune response, ligand-gated ion channel activity, and transmembrane receptor activity. Among the CEG modules, IL22RA1, DPEP1, and BCAT1 were found as highly connective hub genes associated with both FPC and SPC. Conclusion: FPC-related PanIN lesions exhibit a common molecular basis with SPC as shown by gene network activities and commonly expressed high-connectivity hub genes. The differential molecular pathology of FPC and SPC involves multiple coexpressed gene clusters enriched for GO terms including extracellular activities and mitochondrion function.

uKIN Combines New and Prior Information with Guided Network Propagation to Accurately Identify Disease Genes.

Protein interaction networks provide a powerful framework for identifying genes causal for complex genetic diseases. Here, we introduce a general framework, uKIN, that uses prior knowledge of disease-associated genes to guide, within known protein-protein interaction networks, random walks that are initiated from newly identified candidate genes. In large-scale testing across 24 cancer types, we demonstrate that our network propagation approach for integrating both prior and new information not only better identifies cancer driver genes than using either source of information alone but also readily outperforms other state-of-the-art network-based approaches. We also apply our approach to genome-wide association data to identify genes functionally relevant for several complex diseases. Overall, our work suggests that guided network propagation approaches that utilize both prior and new data are a powerful means to identify disease genes. uKIN is freely available for download at: https://github.com/Singh-Lab/uKIN.

NGSEA: Network-Based Gene Set Enrichment Analysis for Interpreting Gene Expression Phenotypes with Functional Gene Sets.

Gene set enrichment analysis (GSEA) is a popular tool to identify underlying biological processes in clinical samples using their gene expression phenotypes. GSEA measures the enrichment of annotated gene sets that represent biological processes for differentially expressed genes (DEGs) in clinical samples. GSEA may be suboptimal for functional gene sets; however, because DEGs from the expression dataset may not be functional genes per se but dysregulated genes perturbed by bona fide functional genes. To overcome this shortcoming, we developed network-based GSEA (NGSEA), which measures the enrichment score of functional gene sets using the expression difference of not only individual genes but also their neighbors in the functional network. We found that NGSEA outperformed GSEA in identifying pathway gene sets for matched gene expression phenotypes. We also observed that NGSEA substantially improved the ability to retrieve known anti-cancer drugs from patient-derived gene expression data using drug-target gene sets compared with another method, Connectivity Map. We also repurposed FDA-approved drugs using NGSEA and experimentally validated budesonide as a chemical with anti-cancer effects for colorectal cancer. We, therefore, expect that NGSEA will facilitate both pathway interpretation of gene expression phenotypes and anti-cancer drug repositioning. NGSEA is freely available at www.inetbio.org/ngsea.

Identification of Novel Components of Target-of-Rapamycin Signaling Pathway by Network-Based Multi-Omics Integrative Analysis.

Target of rapamycin (TOR) is a major signaling pathway and regulator of cell growth. TOR serves as a hub of many signaling routes, and is implicated in the pathophysiology of numerous human diseases, including cancer, diabetes, and neurodegeneration. Therefore, elucidation of unknown components of TOR signaling that could serve as potential biomarkers and drug targets has a great clinical importance. In this study, our aim is to integrate transcriptomics, interactomics, and regulomics data in Saccharomyces cerevisiae using a network-based multiomics approach to enlighten previously unidentified, potential components of TOR signaling. We constructed the TOR-signaling protein interaction network, which was used as a template to search for TOR-mediated rapamycin and caffeine signaling paths. We scored the paths passing from at least one component of TOR Complex 1 or 2 (TORC1/TORC2) using the co-expression levels of the genes in the transcriptome data of the cells grown in the presence of rapamycin or caffeine. The resultant network revealed seven hitherto unannotated proteins, namely, Atg14p, Rim20p, Ret2p, Spt21p, Ylr257wp, Ymr295cp, and Ygr017wp, as potential components of TOR-mediated rapamycin and caffeine signaling in yeast. Among these proteins, we suggest further deciphering of the role of Ylr257wp will be particularly informative in the future because it was the only protein whose removal from the constructed network hindered the signal transduction to the TORC1 effector kinase Npr1p. In conclusion, this study underlines the value of network-based multiomics integrative data analysis in discovering previously unidentified components of the signaling networks by revealing potential components of TOR signaling for future experimental validation.

A novel joint HCPMMP method for automatically classifying Alzheimer's and different stage MCI patients.

A 360-area surface-based cortical parcellation was recently generated using multimodal data in a group average of 210 healthy young adults from the Human Connectome Project (HCP). In order to automatically and accurately identify mild cognitive impairment (MCI) at its two levels (early MCI and late MCI), Alzheimer's disease (AD) and healthy control (HC), a novel joint HCP MMP method was first proposed to delineate the cortical architecture and function connectivity in a group of non healthy adults. The proposed method was applied to register a dataset of 96 resting-state functional connectomes from the Alzheimer's Disease Neuroimaging Initiative (ADNI) to Connectivity Informatics Technology Initiative (CIFTI) space and parcellated brain into human connectome project multi-modal parcellation (HCPMMP) with 360 areas. Various network features in each node of the connectivity network were considered as the candidate features for classification.The fine-grained multi-modal based on HCP-MMP combined with machine learning in identification for EMCI, LMCI, AD and HC. Applying various network features, including strength, betweenness centrality, clustering coefficient, local efficiency, eigenvector centrality, etc, we trained and tested several machine learning models. Thousands of features were processed by filter and wrapper feature selection procedures, and finally there were thirty features to be selected to achieve classification accuracies of 93.8% for EMCI vs. HC, 95.8% for LMCI vs. HC, 95.8% for AD vs. HC, and 91.7% for LMCI vs. AD, respectively by using support vector machine (SVM) algorithm. Most of the selected features locate in the region of temporal or cingulate cortex. Compared with previous studies, our results demonstrate the superiority of the proposed method over existing techniques.

Re-exploring the core genes and modules in the human frontal cortex during chronological aging: insights from network-based analysis of transcriptomic studies.

Frontal cortical dysfunction is a fundamental pathology contributing to age-associated behavioral and cognitive deficits that predispose older adults to neurodegenerative diseases. It is established that aging increases the risk of frontal cortical dysfunction; however, the underlying molecular mechanism remains elusive. Here, we used an integrative meta-analysis to combine five frontal cortex microarray studies with a combined sample population of 161 younger and 155 older individuals. A network-based analysis was used to describe an outline of human frontal cortical aging to identify core genes whose expression changes with age and to reveal the interrelationships among these genes. We found that histone deacetylase 1 (HDAC1) and YES proto-oncogene 1 (YES1) are the two most upregulated genes, while cell division cycle 42 (CDC42) is the central regulatory gene decreased in the aged human frontal cortex. Quantitative PCR assays revealed corresponding changes in frontal cortical Hdac1, Yes1 and Cdc42 mRNA levels in an established aging mouse model. Moreover, analysis of the GSE48350 dataset confirmed similar changes in HDAC1, CDC42 and YES1 expression in Alzheimer's disease, thereby providing a molecular connection between aging and Alzheimer's disease (AD). This framework of network-based analysis could provide novel strategies for detecting and monitoring aging in the brain.

A Network-Based Integrative Workflow to Unravel Mechanisms Underlying Disease Progression.

Unraveling mechanisms underlying diseases has motivated the development of systems biology approaches. The key challenges for the development of mathematical models and computational tool are (1) the size of molecular networks, (2) the nonlinear nature of spatio-temporal interactions, and (3) feedback loops in the structure of interaction networks. We here propose an integrative workflow that combines structural analyses of networks, high-throughput data, and mechanistic modeling. As an illustration of the workflow, we use prostate cancer as a case study with the aim of identifying key functional components associated with primary to metastasis transitions. Analysis carried out by the workflow revealed that HOXD10, BCL2, and PGR are the most important factors affected in primary prostate samples, whereas, in the metastatic state, STAT3, JUN, and JUNB are playing a central role. The identified key elements of each network are validated using patient survival analysis. The workflow presented here allows experimentalists to use heterogeneous data sources for the identification of diagnostic and prognostic signatures.

Network-Based Coverage of Mutational Profiles Reveals Cancer Genes.

A central goal in cancer genomics is to identify the somatic alterations that underpin tumor initiation and progression. While commonly mutated cancer genes are readily identifiable, those that are rarely mutated across samples are difficult to distinguish from the large numbers of other infrequently mutated genes. We introduce a method, nCOP, that considers per-individual mutational profiles within the context of protein-protein interaction networks in order to identify small connected subnetworks of genes that, while not individually frequently mutated, comprise pathways that are altered across (i.e., "cover") a large fraction of individuals. By analyzing 6,038 samples across 24 different cancer types, we demonstrate that nCOP is highly effective in identifying cancer genes, including those with low mutation frequencies. Overall, our work demonstrates that combining per-individual mutational information with interaction networks is a powerful approach for tackling the mutational heterogeneity observed across cancers.