Gene co-expression network analysis reveals common system-level properties of prognostic genes across cancer types

CCDC167 as a potential therapeutic target and regulator of cell cycle-related networks in breast cancer

According to cancer statistics reported in 2020, breast cancer constitutes 30% of new cancer cases diagnosed in American women. Histological markers of breast cancer are expressions of the estrogen receptor (ER), the progesterone receptor (PR), and human epidermal growth factor receptor (HER)-2. Up to 80% of breast cancers are grouped as ER-positive, which implies a crucial role for estrogen in breast cancer development. Therefore, identifying potential therapeutic targets and investigating their downstream pathways and networks are extremely important for drug development in these patients. Through high-throughput technology and bioinformatics screening, we revealed that coiled-coil domain-containing protein 167 (CCDC167) was upregulated in different types of tumors; however, the role of CCDC167 in the development of breast cancer still remains unclear. Integrating many kinds of databases including ONCOMINE, MetaCore, IPA, and Kaplan-Meier Plotter, we found that high expression levels of CCDC167 predicted poor prognoses of breast cancer patients. Knockdown of CCDC167 attenuated aggressive breast cancer growth and proliferation. We also demonstrated that treatment with fluorouracil, carboplatin, paclitaxel, and doxorubicin resulted in decreased expression of CCDC167 and suppressed growth of MCF-7 cells. Collectively, these findings suggest that CCDC167 has high potential as a therapeutic target for breast cancer.

Exploring chromatin conformation and gene co-expression through graph embedding

MOTIVATION

The relationship between gene co-expression and chromatin conformation is of great biological interest. Thanks to high-throughput chromosome conformation capture technologies (Hi-C), researchers are gaining insights on the tri-dimensional organization of the genome. Given the high complexity of Hi-C data and the difficult definition of gene co-expression networks, the development of proper computational tools to investigate such relationship is rapidly gaining the interest of researchers. One of the most fascinating questions in this context is how chromatin topology correlates with gene co-expression and which physical interaction patterns are most predictive of co-expression relationships.

RESULTS

To address these questions, we developed a computational framework for the prediction of co-expression networks from chromatin conformation data. We first define a gene chromatin interaction network where each gene is associated to its physical interaction profile; then, we apply two graph embedding techniques to extract a low-dimensional vector representation of each gene from the interaction network; finally, we train a classifier on gene embedding pairs to predict if they are co-expressed. Both graph embedding techniques outperform previous methods based on manually designed topological features, highlighting the need for more advanced strategies to encode chromatin information. We also establish that the most recent technique, based on random walks, is superior. Overall, our results demonstrate that chromatin conformation and gene regulation share a non-linear relationship and that gene topological embeddings encode relevant information, which could be used also for downstream analysis.

AVAILABILITY AND IMPLEMENTATION

The source code for the analysis is available at: https://github.com/marcovarrone/gene-expression-chromatin.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

scLM: Automatic Detection of Consensus Gene Clusters Across Multiple Single-cell Datasets

In gene expression profiling studies, including single-cell RNAsequencing (scRNA-seq) analyses, the identification and characterization of co-expressed genes provides critical information on cell identity and function. Gene co-expression clustering in scRNA-seq data presents certain challenges. We show that commonly used methods for single-cell data are not capable of identifying co-expressed genes accurately, and produce results that substantially limit biological expectations of co-expressed genes. Herein, we present single-cell Latent-variable Model (scLM), a gene co-clustering algorithm tailored to single-cell data that performs well at detecting gene clusters with significant biologic context. Importantly, scLM can simultaneously cluster multiple single-cell datasets, i.e., consensus clustering, enabling users to leverage single-cell data from multiple sources for novel comparative analysis. scLM takes raw count data as input and preserves biological variation without being influenced by batch effects from multiple datasets. Results from both simulation data and experimental data demonstrate that scLM outperforms the existing methods with considerably improved accuracy. To illustrate the biological insights of scLM, we apply it to our in-house and public experimental scRNA-seq datasets. scLM identifies novel functional gene modules and refines cell states, which facilitates mechanism discovery and understanding of complex biosystems such as cancers. A user-friendly R package with all the key features of the scLM method is available at https://github.com/QSong-github/scLM.

Identification of Hub Genes Associated With Development and Microenvironment of Hepatocellular Carcinoma by Weighted Gene Co-expression Network Analysis and Differential Gene Expression Analysis

A further understanding of the molecular mechanism of hepatocellular carcinoma (HCC) is necessary to predict a patient's prognosis and develop new targeted gene drugs. This study aims to identify essential genes related to HCC. We used the Weighted Gene Co-expression Network Analysis (WGCNA) and differential gene expression analysis to analyze the gene expression profile of GSE45114 in the Gene Expression Omnibus (GEO) database and The Cancer Genome Atlas database (TCGA). A total of 37 overlapping genes were extracted from four groups of results. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway and Gene Ontology (GO) enrichment analyses were performed on the 37 overlapping genes. Then, we used the STRING database to map the protein interaction (PPI) network of 37 overlapping genes. Ten hub genes were screened according to the Maximal Clique Centrality (MCC) score using the Cytohubba plugin of Cytoscape (including FOS, EGR1, EPHA2, DUSP1, IGFBP3, SOCS2, ID1, DUSP6, MT1G, and MT1H). Most hub genes show a significant association with immune infiltration types and tumor stemness of microenvironment in HCC. According to Univariate Cox regression analysis and Kaplan-Meier survival estimation, SOCS2 was positively correlated with overall survival (OS), and IGFBP3 was negatively correlated with OS. Moreover, the expression of IGFBP3 increased with the increase of the clinical stage, while the expression of SOCS2 decreased with the increase of the clinical stage. In conclusion, our findings suggest that SOCS2 and IGFBP3 may play an essential role in the development of HCC and may serve as a potential biomarker for future diagnosis and treatment.

Tumor microenvironment derived signature predicting relapse-free survival in I-III cancer and preliminary experiment verification

The recurrence in colon cancer contributed to great difficulties in diagnostic and therapeutic treatment. Tumor microenvironment (TME) gains increasing attention recently. After univariate Cox analysis on relapse-free survival (RFS) and ESTIMATE analysis, WGCNA was further conducted to determine the TME and relapse-related genes in I-III colon cancer. Functional enrichment analyses were conducted. Furthermore, seven genes were screened to build a prognostic signature via LASSO and multivariate Cox analysis. Univariate followed multivariate Cox analysis all showed that the risk group calculated by the signature as a significant predictors. The ROC curves showed great prognostic in the internal training group, internal verification group, and independent external verification group. In the training group, the AUC at 1, 3, and 5 years was 0.737, 0.79, and 0.756. In addition, correlation analysis presented that the signature and genes involved in were significantly associated with the TME. Moreover, 3 of 7 genes (FAM78A, SGIP1, and MMP9) were validated to be associated with PDL1 through qRT-PCR.

Differential Co-Expression Analyses Allow the Identification of Critical Signalling Pathways Altered during Tumour Transformation and Progression

Biological systems respond to perturbations through the rewiring of molecular interactions, organised in gene regulatory networks (GRNs). Among these, the increasingly high availability of transcriptomic data makes gene co-expression networks the most exploited ones. Differential co-expression networks are useful tools to identify changes in response to an external perturbation, such as mutations predisposing to cancer development, and leading to changes in the activity of gene expression regulators or signalling. They can help explain the robustness of cancer cells to perturbations and identify promising candidates for targeted therapy, moreover providing higher specificity with respect to standard co-expression methods. Here, we comprehensively review the literature about the methods developed to assess differential co-expression and their applications to cancer biology. Via the comparison of normal and diseased conditions and of different tumour stages, studies based on these methods led to the definition of pathways involved in gene network reorganisation upon oncogenes' mutations and tumour progression, often converging on immune system signalling. A relevant implementation still lagging behind is the integration of different data types, which would greatly improve network interpretability. Most importantly, performance and predictivity evaluation of the large variety of mathematical models proposed would urgently require experimental validations and systematic comparisons. We believe that future work on differential gene co-expression networks, complemented with additional omics data and experimentally tested, will considerably improve our insights into the biology of tumours.

TAF1A and ZBTB41 serve as novel key genes in cervical cancer identified by integrated approaches

Cervical cancer (CC) is the second most common cancer and the leading cause of cancer mortality in women. Numerous studies have found that the development of CC was associated with multiple genes. However, the mechanisms on gene level are enigmatic, hindering the understanding of its functional roles. This study sought to identify prognostic biomarkers of CC, and explore their biological functions. Here we conducted an integrated analysis to screen potential vital genes. Candidate genes were further tested by experiments in clinical specimens and cancer cell line. Then, molecular modeling was used to predict the three-dimensional structure of candidate genes’ proteins, and the interaction pattern was analyzed by docking simulation technique. Among the potential genes identified, we found that TAF1A and ZBTB41 were highly correlated. Furthermore, there was a definite interaction between the proteins of TAF1A and ZBTB41, which was affected by the activity of the p53 signaling pathway. In conclusion, our findings identified TAF1A and ZBTB41 could serve as biomarkers of CC. We confirmed their biological function and deciphered their interaction for the first time, which may be helpful for developing further researches.

Pan-cancer driver copy number alterations identified by joint expression/CNA data analysis

Analysis of large gene expression datasets from biopsies of cancer patients can identify co-expression signatures representing particular biomolecular events in cancer. Some of these signatures involve genomically co-localized genes resulting from the presence of copy number alterations (CNAs), for which analysis of the expression of the underlying genes provides valuable information about their combined role as oncogenes or tumor suppressor genes. Here we focus on the discovery and interpretation of such signatures that are present in multiple cancer types due to driver amplifications and deletions in particular regions of the genome after doing a comprehensive analysis combining both gene expression and CNA data from The Cancer Genome Atlas.

Transcriptional regulation of bark freezing tolerance in apple (Malus domestica Borkh.)

Freezing tolerance is a significant trait in plants that grow in cold environments and survive through the winter. Apple (Malus domestica Borkh.) is a cold-tolerant fruit tree, and the cold tolerance of its bark is important for its survival at low temperatures. However, little is known about the gene activity related to its freezing tolerance. To better understand the gene expression and regulation properties of freezing tolerance in dormant apple trees, we analyzed the transcriptomic divergences in the bark from 1-year-old branches of two apple cultivars, "Golden Delicious" (G) and "Jinhong" (H), which have different levels of cold resistance, under chilling and freezing treatments. "H" can safely overwinter below -30 °C in extremely low-temperature regions, whereas "G" experiences severe freezing damage and death in similar environments. Based on 28 bark transcriptomes (from the epidermis, phloem, and cambium) from 1-year-old branches under seven temperature treatments (from 4 to -29 °C), we identified 4173 and 7734 differentially expressed genes (DEGs) in "G" and "H", respectively, between the chilling and freezing treatments. A gene coexpression network was constructed according to this expression information using weighted gene correlation network analysis (WGCNA), and seven biologically meaningful coexpression modules were identified from the network. The expression profiles of the genes from these modules suggested the gene regulatory pathways that are responsible for the chilling and freezing stress responses of "G" and/or "H." Module 7 was probably related to freezing acclimation and freezing damage in "H" at the lower temperatures. This module contained more interconnected hub transcription factors (TFs) and cold-responsive genes (CORs). Modules 6 and 7 contained C-repeat binding factor (CBF) TFs, and many CBF-dependent homologs were identified as hub genes. We also found that some hub TFs had higher intramodular connectivity (KME) and gene significance (GS) than CBFs. Specifically, most hub TFs in modules 6 and 7 were activated at the beginning of the early freezing stress phase and maintained upregulated expression during the whole freezing stress period in "G" and "H". The upregulation of DEGs related to methionine and carbohydrate biosynthetic processes in "H" under more severe freezing stress supported the maintenance of homeostasis in the cellular membrane. This study improves our understanding of the transcriptional regulation patterns underlying freezing tolerance in the bark of apple branches.

Molecular Correlates of Hemorrhage and Edema Volumes Following Human Intracerebral Hemorrhage Implicate Inflammation, Autophagy, mRNA Splicing, and T Cell Receptor Signaling

Intracerebral hemorrhage (ICH) and perihematomal edema (PHE) volumes are major determinants of ICH outcomes as is the immune system which plays a significant role in damage and repair. Thus, we performed whole-transcriptome analyses of 18 ICH patients to delineate peripheral blood genes and networks associated with ICH volume, absolute perihematomal edema (aPHE) volume, and relative PHE (aPHE/ICH; rPHE). We found 440, 266, and 391 genes correlated with ICH and aPHE volumes and rPHE, respectively (p < 0.005, partial-correlation > |0.6|). These mainly represented inflammatory pathways including NF-κB, TREM1, and Neuroinflammation Signaling-most activated with larger volumes. Weighted Gene Co-Expression Network Analysis identified seven modules significantly correlated with these measures (p < 0.05). Most modules were enriched in neutrophil, monocyte, erythroblast, and/or T cell-specific genes. Autophagy, apoptosis, HIF-1α, inflammatory and neuroinflammatory response (including Toll-like receptors), cell adhesion (including MMP9), platelet activation, T cell receptor signaling, and mRNA splicing were represented in these modules (FDR p < 0.05). Module hub genes, potential master regulators, were enriched in neutrophil-specific genes in three modules. Hub genes included NCF2, NCF4, STX3, and CSF3R, and involved immune response, autophagy, and neutrophil chemotaxis. One module that correlated negatively with ICH volume correlated positively with rPHE. Its genes and hubs were enriched in T cell-specific genes including hubs LCK and ITK, Src family tyrosine kinases whose modulation improved outcomes and reduced BBB dysfunction following experimental ICH. This study uncovers molecular underpinnings associated with ICH and PHE volumes and pathophysiology in human ICH, where knowledge is scarce. The identified pathways and hub genes may represent novel therapeutic targets.

A strategy to incorporate prior knowledge into correlation network cutoff selection

Correlation networks are frequently used to statistically extract biological interactions between omics markers. Network edge selection is typically based on the statistical significance of the correlation coefficients. This procedure, however, is not guaranteed to capture biological mechanisms. We here propose an alternative approach for network reconstruction: a cutoff selection algorithm that maximizes the overlap of the inferred network with available prior knowledge. We first evaluate the approach on IgG glycomics data, for which the biochemical pathway is known and well-characterized. Importantly, even in the case of incomplete or incorrect prior knowledge, the optimal network is close to the true optimum. We then demonstrate the generalizability of the approach with applications to untargeted metabolomics and transcriptomics data. For the transcriptomics case, we demonstrate that the optimized network is superior to statistical networks in systematically retrieving interactions that were not included in the biological reference used for optimization.

A system-level approach identifies HIF-2α as a critical regulator of chondrosarcoma progression

Chondrosarcomas, malignant cartilaginous neoplasms, are capable of transitioning to highly aggressive, metastatic, and treatment-refractory states, resulting in significant patient mortality. Here, we aim to uncover the transcriptional program directing such tumor progression in chondrosarcomas. We conduct weighted correlation network analysis to extract a characteristic gene module underlying chondrosarcoma malignancy. Hypoxia-inducible factor-2α (HIF-2α, encoded by EPAS1) is identified as an upstream regulator that governs the malignancy gene module. HIF-2α is upregulated in high-grade chondrosarcoma biopsies and EPAS1 gene amplification is associated with poor prognosis in chondrosarcoma patients. Using tumor xenograft mouse models, we demonstrate that HIF-2α confers chondrosarcomas the capacities required for tumor growth, local invasion, and metastasis. Meanwhile, pharmacological inhibition of HIF-2α, in conjunction with the chemotherapy agents, synergistically enhances chondrosarcoma cell apoptosis and abolishes malignant signatures of chondrosarcoma in mice. We expect that our insights into the pathogenesis of chondrosarcoma will provide guidelines for the development of molecular targeted therapeutics for chondrosarcoma.

Chondrosarcomas are frequently aggressive, understanding the transcriptional changes associated with progression may help in developing new treatments. Here, the authors show that HIF-2α is increased in expression on progression and pharmacological inhibition of the protein together with chemotherapy is a useful strategy for controlling tumour growth in mice.

Identification of Reproducible BCL11A Alterations in Schizophrenia Through Individual-Level Prediction of Coexpression

Previous studies have provided evidence for an alteration of genetic coexpression in schizophrenia (SCZ). However, such analyses have thus far lacked biological specificity for individual genes, which may be critical for identifying illness-relevant effects. Therefore, we applied machine learning to identify gene-specific coexpression differences at the individual subject level and compared these between individuals with SCZ, bipolar disorder, major depressive disorder (MDD), autism spectrum disorder (ASD), and healthy controls. Utilizing transcriptome-wide gene expression data from 21 independent datasets, comprising a total of 9509 participants, we identified a reproducible decrease of BCL11A coexpression across 4 SCZ datasets that showed diagnostic specificity for SCZ when compared with ASD and MDD. We further demonstrate that individual-level coexpression differences can be combined in multivariate coexpression scores that show reproducible illness classification across independent datasets in SCZ and ASD. This study demonstrates that machine learning can capture gene-specific coexpression differences at the individual subject level for SCZ and identify novel biomarker candidates.

Constructing cancer patient-specific and group-specific gene networks with multi-omics data

Background

Cancer is a complex and heterogeneous disease with many possible genetic and environmental causes. The same treatment for patients of the same cancer type often results in different outcomes in terms of efficacy and side effects of the treatment. Thus, the molecular characterization of individual cancer patients is increasingly important to find an effective treatment. Recently a few methods have been developed to construct cancer sample-specific gene networks based on the difference in the mRNA expression levels between the cancer sample and reference samples.

Methods

We constructed a patient-specific network with multi-omics data based on the difference between a reference network and a perturbed reference network by the patient. A network specific to a group of patients was obtained using the average change in correlation coefficients and node degree of patient-specific networks of the group.

Results

In this paper, we present a new method for constructing cancer patient-specific and group-specific gene networks with multi-omics data. The main differences of our method from previous ones are as follows: (1) networks are constructed with multi-omics (mRNA expression, copy number variation, DNA methylation and microRNA expression) data rather than with mRNA expression data alone, (2) background networks are constructed with both normal samples and cancer samples of the specified type to extract cancer-specific gene correlations, and (3) both patient individual-specific networks and patient group-specific networks can be constructed. The results of evaluating our method with several types of cancer show that it constructs more informative and accurate gene networks than previous methods.

Conclusions

The results of evaluating our method with extensive data of seven cancer types show that the difference of gene correlations between the reference samples and a patient sample is a more predictive feature than mRNA expression levels and that gene networks constructed with multi-omics data show a better performance than those with single omics data in predicting cancer for most cancer types. Our approach will be useful for finding genes and gene pairs to tailor treatments to individual characteristics.

Identification of GSN and LAMC2 as Key Prognostic Genes of Bladder Cancer by Integrated Bioinformatics Analysis

Bladder cancer is a common malignancy with mechanisms of pathogenesis and progression. This study aimed to identify the prognostic hub genes, which are the central modulators to regulate the progression and proliferation in the specific subtype of bladder cancer. The identification of the candidate hub gene was performed by weighted gene co-expression network analysis to construct a free-scale gene co-expression network. The gene expression profile of GSE97768 from the Gene Expression Omnibus database was used. The association between prognosis and hub gene was evaluated by The Cancer Genome Atlas database. Four gene-expression modules were significantly related to bladder cancer disease: the red module (human adenocarcinoma lymph node metastasis), the darkturquioise module (grade 2 carcinoma), the lightgreen module (grade 3 carcinoma), and the royalblue module (transitional cell carcinoma lymphatic metastasis). Based on betweenness centrality and survival analysis, we identified laminin subunit gamma-2 (LAMC2) in the grade 2 carcinoma, gelsolin (GSN) in the grade 3 carcinoma, and homeodomain-interacting protein kinase 2 (HIPK2) in the transitional cell carcinoma lymphatic metastasis. Subsequently, the protein levels of LAMC2 and GSN were respectively down-regulated and up-regulated in tumor tissue with the Human Protein Atlas (HPA) database. Our results suggested that LAMC2 and GSN are the central modulators to transfer information in the specific subtype of the disease.

CODC: a Copula-based model to identify differential coexpression

Differential coexpression has recently emerged as a new way to establish a fundamental difference in expression pattern among a group of genes between two populations. Earlier methods used some scoring techniques to detect changes in correlation patterns of a gene pair in two conditions. However, modeling differential coexpression by means of finding differences in the dependence structure of the gene pair has hitherto not been carried out. We exploit a copula-based framework to model differential coexpression between gene pairs in two different conditions. The Copula is used to model the dependency between expression profiles of a gene pair. For a gene pair, the distance between two joint distributions produced by copula is served as differential coexpression. We used five pan-cancer TCGA RNA-Seq data to evaluate the model that outperforms the existing state of the art. Moreover, the proposed model can detect a mild change in the coexpression pattern across two conditions. For noisy expression data, the proposed method performs well because of the popular scale-invariant property of copula. In addition, we have identified differentially coexpressed modules by applying hierarchical clustering on the distance matrix. The identified modules are analyzed through Gene Ontology terms and KEGG pathway enrichment analysis.

Expression of S100A Alarmins in Cord Blood Monocytes Is Highly Associated With Chorioamnionitis and Fetal Inflammation in Preterm Infants

Background: Preterm infants exposed to chorioamnionitis and with a fetal inflammatory response are at risk for neonatal morbidity and adverse outcome. Alarmins S100A8, S100A9, and S100A12 are expressed by myeloid cells and have been associated with inflammatory activation and monocyte modulation.

Aim: To study S100A alarmin expression in cord blood monocytes from term healthy and preterm infants and relate results to clinical findings, inflammatory biomarkers and alarmin protein levels, as well as pathways identified by differentially regulated monocyte genes.

Methods: Cord blood CD14+ monocytes were isolated from healthy term (n = 10) and preterm infants (<30 weeks gestational age, n = 33) by MACS technology. Monocyte RNA was sequenced and gene expression was analyzed by Principal Component Analysis and hierarchical clustering. Pathways were identified by Ingenuity Pathway Analysis. Inflammatory proteins were measured by Multiplex ELISA, and plasma S100A proteins by mass spectrometry. Histological chorioamnionitis (HCA) and fetal inflammatory response syndrome (FIRS) were diagnosed by placenta histological examination.

Results: S100A8, S100A9, and S100A12 gene expression was significantly increased and with a wider range in preterm vs. term infants. High S100A8 and S100A9 gene expression (n = 17) within the preterm group was strongly associated with spontaneous onset of delivery, HCA, FIRS and elevated inflammatory proteins in cord blood, while low expression (n = 16) was associated with impaired fetal growth and physician-initiated delivery. S100A8 and S100A9 protein levels were significantly lower in preterm vs. term infants, but within the preterm group high S100A gene expression, spontaneous onset of labor, HCA and FIRS were associated with elevated protein levels. One thousand nine hundred genes were differentially expressed in preterm infants with high vs. low S100A alarmin expression. Analysis of 124 genes differentially expressed in S100A high as well as FIRS and HCA groups identified 18 common pathways and S100A alarmins represented major hubs in network analyses.

Conclusion: High expression of S100A alarmins in cord blood monocytes identifies a distinct clinical risk group of preterm infants exposed to chorioamnionitis and with a fetal inflammatory response. Gene and pathway analyses suggest that high S100A alarmin expression also affects monocyte function. The connection with monocyte phenotype and inflammation-stimulated S100A expression in other cell types (e.g., neutrophils) warrants further investigation.

Identification of Hub Genes Associated With Development of Head and Neck Squamous Cell Carcinoma by Integrated Bioinformatics Analysis

Improved insight into the molecular mechanisms of head and neck squamous cell carcinoma (HNSCC) is required to predict prognosis and develop a new therapeutic strategy for targeted genes. The aim of this study is to identify significant genes associated with HNSCC and to further analyze its prognostic significance. In our study, the cancer genome atlas (TCGA) HNSCC database and the gene expression profiles of GSE6631 from the Gene Expression Omnibus (GEO) were used to explore the differential co-expression genes in HNSCC compared with normal tissues. A total of 29 differential co-expression genes were screened out by Weighted Gene Co-expression Network Analysis (WGCNA) and differential gene expression analysis methods. As suggested in functional annotation analysis using the R clusterProfiler package, these genes were mainly enriched in epidermis development and differentiation (biological process), apical plasma membrane and cell-cell junction (cellular component), and enzyme inhibitor activity (molecular function). Furthermore, in a protein-protein interaction (PPI) network containing 21 nodes and 25 edges, the ten hub genes (S100A8, S100A9, IL1RN, CSTA, ANXA1, KRT4, TGM3, SCEL, PPL, and PSCA) were identified using the CytoHubba plugin of Cytoscape. The expression of the ten hub genes were all downregulated in HNSCC tissues compared with normal tissues. Based on survival analysis, the lower expression of CSTA was associated with worse overall survival (OS) in patients with HNSCC. Finally, the protein level of CSTA, which was validated by the Human Protein Atlas (HPA) database, was down-regulated consistently with mRNA levels in head and neck cancer samples. In summary, our study demonstrated that a survival-related gene is highly correlated with head and neck cancer development. Thus, CSTA may play important roles in the progression of head and neck cancer and serve as a potential biomarker for future diagnosis and treatment.

Identification of gene co-expression modules and hub genes associated with the invasiveness of pituitary adenoma

In pituitary adenoma (PA), invasiveness is the main cause of recurrence and poor prognosis. Thus, identifying specific biomarkers for diagnosis and effective treatment of invasive PAs is of great clinical significance. In this study, from the Gene Expression Omnibus database, we obtained and combined several microarrays of PA by the "sva" R package. Weighted gene co-expression network analysis was performed to construct a scale-free topology model and analyze the relationships between the modules and clinical traits. Our analysis results indicated that three key modules (dark turquoise, saddle brown, and steel blue) were associated with the invasiveness of PA. Kyoto Encyclopedia of Genes and Genomes pathway enrichment analysis and Gene Ontology analysis were performed for the functional annotation of the key modules. In addition, the hub genes in the three modules were identified and screened by differential expression analysis between normal samples and PA samples. Three upregulated differentially expressed genes (DGAT2, PIGZ, and DHRS2) were identified. The Fisher's exact test and receiver operating characteristic curve were used to validate the capability of these genes to distinguish invasive traits, and transcription factor interaction networks were used to further explore the underlying mechanisms of the three genes. Moreover, a lower expression level of DGAT2 in invasive PA tissue than in noninvasive PA tissue was validated by quantitative reverse transcription-polymerase chain reaction. In general, this study contributes to potential molecular biomarkers of invasive PAs and provides a broader perspective for diagnosis and new therapeutic targets for the invasive PAs.

Construction, comparison and evolution of networks in life sciences and other disciplines

Network approaches have become pervasive in many research fields. They allow for a more comprehensive understanding of complex relationships between entities as well as their group-level properties and dynamics. Many networks change over time, be it within seconds or millions of years, depending on the nature of the network. Our focus will be on comparative network analyses in life sciences, where deciphering temporal network changes is a core interest of molecular, ecological, neuropsychological and evolutionary biologists. Further, we will take a journey through different disciplines, such as social sciences, finance and computational gastronomy, to present commonalities and differences in how networks change and can be analysed. Finally, we envision how borrowing ideas from these disciplines could enrich the future of life science research.

Integrative analyses of noncoding RNAs reveal the potential mechanisms augmenting tumor malignancy in lung adenocarcinoma

Precise noncoding RNA (ncRNA)-based network prediction is necessary to reveal ncRNA functions and pathological mechanisms. Here, we established a systemic pipeline to identify prognostic ncRNAs, predict their functions and explore their pathological mechanisms in lung adenocarcinoma (LUAD). After in silico and experimental validation based on evaluations of prognostic value in multiple LUAD cohorts, we selected the PTTG3P pseudogene from among other prognostic ncRNAs (MIR497HG, HSP078, TBX5-AS1, LOC100506990 and C14orf64) for mechanistic studies. PTTG3P upregulation in LUAD cells shortens the metaphase to anaphase transition in mitosis, increases cell viability after cisplatin or paclitaxel treatment, facilitates tumor growth that leads to poor survival in orthotopic lung models, and is associated with a poor survival rate in LUAD patients in the TCGA cohort who received chemotherapy. Mechanistically, PTTG3P acts as an ncRNA that interacts with the transcription factor FOXM1 to regulate the transcriptional activation of the mitotic checkpoint kinase BUB1B, which augments tumor growth and chemoresistance and leads to poor outcomes for LUAD patients. Overall, we established a systematic strategy to uncover prognostic ncRNAs with functional prediction methods suitable for pan-cancer studies. Moreover, we revealed that PTTG3P, due to its upregulation of the PTTG3P/FOXM1/BUB1B axis, could be a therapeutic target for LUAD patients.

Network Properties of Cancer Prognostic Gene Signatures in the Human Protein Interactome

Prognostic gene signatures are critical in cancer prognosis assessments and their pinpoint treatments. However, their network properties remain unclear. Here, we obtained nine prognostic gene sets including 1439 prognostic genes of different cancers from related publications. Four network centralities were used to examine the network properties of prognostic genes (PG) compared with other gene sets based on the Human Protein Reference Database (HPRD) and String networks. We also proposed three novel network measures for further investigating the network properties of prognostic gene sets (PGS) besides clustering coefficient. The results showed that PG did not occupy key positions in the human protein interaction network and were more similar to essential genes rather than cancer genes. However, PGS had significantly smaller intra-set distance (IAD) and inter-set distance (IED) in comparison with random sets (p-value < 0.001). Moreover, we also found that PGS tended to be distributed within network modules rather than between modules (p-value < 0.01), and the functional intersection of the modules enriched with PGS was closely related to cancer development and progression. Our research reveals the common network properties of cancer prognostic gene signatures in the human protein interactome. We argue that these are biologically meaningful and useful for understanding their molecular mechanism.

Co-expression networks uncover regulation of splicing and transcription markers of disease

Gene co-expression networks based on gene expression data are usually used to capture biologically significant patterns, enabling the discovery of biomarkers and interpretation of regulatory relationships. However, the coordination of numerous splicing changes within and across genes can exert a substantial impact on the function of these genes. This is particularly impactful in studies of the properties of the nervous system, which can be masked in the networks that only assess the correlation between gene expression levels. A bioinformatics approach was developed to uncover the role of alternative splicing and associated transcriptional networks using RNA-seq profiles. Data from 40 samples, including control and two treatments associated with sensitivity to stimuli across two central nervous system regions that can present differential splicing, were explored. The gene expression and relative isoform levels were integrated into a transcriptome-wide matrix, and then Graphical Lasso was applied to capture the interactions between genes and isoforms. Next, functional enrichment analysis enabled the discovery of pathways dysregulated at the isoform or gene levels and the interpretation of these interactions within a central nervous region. In addition, a Bayesian biclustering strategy was used to reconstruct treatment-specific networks from gene expression profile, allowing the identification of hub molecules and visualization of highly connected modules of isoforms and genes in specific conditions. Our bioinformatics approach can offer comparable insights into the discovery of biomarkers and therapeutic targets for a wide range of diseases and conditions.

ManiNetCluster: a novel manifold learning approach to reveal the functional links between gene networks

BACKGROUND

The coordination of genomic functions is a critical and complex process across biological systems such as phenotypes or states (e.g., time, disease, organism, environmental perturbation). Understanding how the complexity of genomic function relates to these states remains a challenge. To address this, we have developed a novel computational method, ManiNetCluster, which simultaneously aligns and clusters gene networks (e.g., co-expression) to systematically reveal the links of genomic function between different conditions. Specifically, ManiNetCluster employs manifold learning to uncover and match local and non-linear structures among networks, and identifies cross-network functional links.

RESULTS

We demonstrated that ManiNetCluster better aligns the orthologous genes from their developmental expression profiles across model organisms than state-of-the-art methods (p-value <2.2×10^-16). This indicates the potential non-linear interactions of evolutionarily conserved genes across species in development. Furthermore, we applied ManiNetCluster to time series transcriptome data measured in the green alga Chlamydomonas reinhardtii to discover the genomic functions linking various metabolic processes between the light and dark periods of a diurnally cycling culture. We identified a number of genes putatively regulating processes across each lighting regime.

CONCLUSIONS

ManiNetCluster provides a novel computational tool to uncover the genes linking various functions from different networks, providing new insight on how gene functions coordinate across different conditions. ManiNetCluster is publicly available as an R package at https://github.com/daifengwanglab/ManiNetCluster.

Finding prognostic gene pairs for cancer from patient-specific gene networks

BACKGROUND

Molecular characterization of individual cancer patients is important because cancer is a complex and heterogeneous disease with many possible genetic and environmental causes. Many studies have been conducted to identify diagnostic or prognostic gene signatures for cancer from gene expression profiles. However, some gene signatures may fail to serve as diagnostic or prognostic biomarkers and gene signatures may not be found in gene expression profiles.

METHODS

In this study, we developed a general method for constructing patient-specific gene correlation networks and for identifying prognostic gene pairs from the networks. A patient-specific gene correlation network was constructed by comparing a reference gene correlation network from normal samples to a network perturbed by a single patient sample. The main difference of our method from previous ones includes (1) it is focused on finding prognostic gene pairs rather than prognostic genes and (2) it can identify prognostic gene pairs from gene expression profiles even when no significant prognostic genes exist.

RESULTS

Evaluation of our method with extensive data sets of three cancer types (breast invasive carcinoma, colon adenocarcinoma, and lung adenocarcinoma) showed that our approach is general and that gene pairs can serve as more reliable prognostic signatures for cancer than genes.

CONCLUSIONS

Our study revealed that prognosis of individual cancer patients is associated with the existence of prognostic gene pairs in the patient-specific network and the size of a subnetwork of the prognostic gene pairs in the patient-specific network. Although preliminary, our approach will be useful for finding gene pairs to predict survival time of patients and to tailor treatments to individual characteristics. The program for dynamically constructing patient-specific gene networks and for finding prognostic gene pairs is available at http://bclab.inha.ac.kr/pancancer.

Time-resolved dual RNA-seq of tissue uncovers Pseudomonas plecoglossicida key virulence genes in host-pathogen interaction with Epinephelus coioides

Bacterial pathogen-host interactions are highly dynamic, regulated processes that have been primarily investigated using in vitro assays. The dynamics of bacterial pathogen-host interplay in vivo are poorly understood. Using time-resolved dual RNA-seq in a Pseudomonas plecoglossicida-Epinephelus coioides infection model, we observed that bacterial genes encoding classical virulence factors and host genes involved in immune regulation were dynamically expressed during infection. Using network inferencing, we were able to predict interspecies regulatory networks linking bacterial virulence genes to host immune genes. Together with gene co-expression network analysis of the pathogen, secY was predicted to be a key virulence gene for P. plecoglossicida pathogenicity in the host, fliN was predicted to be a less important virulence gene. The results of bioinformatics prediction were confirmed by animal infection experiments. Our work provides the first paradigm to study dynamic alterations of bacterial pathogen and host interactions based on the elucidation of time-resolved interactive transcriptomes in vivo, and may be developed into a novel and universal method for revealing the true complexity of the bacterial infection process.

Identification of Modules With Similar Gene Regulation and Metabolic Functions Based on Co-expression Data

Biological systems respond to environmental perturbations and to a large diversity of compounds through gene interactions, and these genetic factors comprise complex networks. In particular, a wide variety of gene co-expression networks have been constructed in recent years thanks to the dramatic increase of experimental information obtained with techniques, such as microarrays and RNA sequencing. These networks allow the identification of groups of co-expressed genes that can function in the same process and, in turn, these networks may be related to biological functions of industrial, medical and academic interest. In this study, gene co-expression networks for 17 bacterial organisms from the COLOMBOS database were analyzed via weighted gene co-expression network analysis and clustered into modules of genes with similar expression patterns for each species. These networks were analyzed to determine relevant modules through a hypergeometric approach based on a set of transcription factors and enzymes for each genome. The richest modules were characterized using PFAM families and KEGG metabolic maps. Additionally, we conducted a Gene Ontology analysis for enrichment of biological functions. Finally, we identified modules that shared similarity through all the studied organisms by using comparative genomics.

Regulostat Inferelator: a novel network biology platform to uncover molecular devices that predetermine cellular response phenotypes

With the emergence of genome editing technologies and synthetic biology, it is now possible to engineer genetic circuits driving a cell's phenotypic response to a stressor. However, capturing a continuous response, rather than simply a binary ‘on’ or ‘off’ response, remains a bioengineering challenge. No tools currently exist to identify gene candidates responsible for predetermining and fine-tuning cell response phenotypes. To address this gap, we devised a novel Regulostat Inferelator (RSI) algorithm to decipher intrinsic molecular devices or networks that predetermine cellular phenotypic responses. The RSI algorithm is designed to extract gene expression patterns from basal transcriptomic data in order to identify ‘regulostat’ constituent gene pairs, which exhibit rheostat-like mode-of-cooperation capable of fine-tuning cellular response. Our proof-of-concept study provides computational evidence for the existence of regulostats and that these networks predetermine cellular response prior to exposure to a stressor or drug. In addition, our work, for the first time, provides evidence of context-specific, drug–regulostat interactions in predetermining drug response phenotypes in cancer cells. Given RSI-inferred regulostat networks offer insights for prioritizing gene candidates capable of rendering a resistant phenotype sensitive to a given drug, we envision that this tool will be of great value in bioengineering and medicine.

Computational Inference of Gene Co-Expression Networks for the identification of Lung Carcinoma Biomarkers: An Ensemble Approach

Gene Networks (GN), have emerged as an useful tool in recent years for the analysis of different diseases in the field of biomedicine. In particular, GNs have been widely applied for the study and analysis of different types of cancer. In this context, Lung carcinoma is among the most common cancer types and its short life expectancy is partly due to late diagnosis. For this reason, lung cancer biomarkers that can be easily measured are highly demanded in biomedical research. In this work, we present an application of gene co-expression networks in the modelling of lung cancer gene regulatory networks, which ultimately served to the discovery of new biomarkers. For this, a robust GN inference was performed from microarray data concomitantly using three different co-expression measures. Results identified a major cluster of genes involved in SRP-dependent co-translational protein target to membrane, as well as a set of 28 genes that were exclusively found in networks generated from cancer samples. Amongst potential biomarkers, genes N C K A P 1 L and D M D are highlighted due to their implications in a considerable portion of lung and bronchus primary carcinomas. These findings demonstrate the potential of GN reconstruction in the rational prediction of biomarkers.

Lower connectivity of tumor coexpression networks is not specific to cancer

Global level network analysis of molecular links is necessary for systems level view of complex diseases like cancer. Using genome-wide expression datasets, we constructed and compared gene co-expression based specific networks of pre-cancerous tumors (adenoma) and cancerous tumors (carcinoma) with paired normal networks to assess for any possible changes in network connectivity. Previously, loss of connectivity was reported as a characteristic of cancer samples. Here, we observed that pre-cancerous conditions also had significantly less connections than paired normal samples. We observed a loss of connectivity trend for colorectal adenoma, aldosterone producing adenoma and uterine leiomyoma. We also showed that the loss of connectivity trend is not specific to positive or negative correlation based networks. Differential hub genes, which were the most highly differentially less connected genes in tumor, were mostly different between different datasets. No common gene list could be defined which underlies the lower connectivity of tumor specific networks. Connectivity of colorectal cancer methylation targets was different from other genes. Extracellular space related terms were enriched in negative correlation based differential hubs and common methylation targets of colorectal carcinoma. Our results indicate a systems level change of lower connectivity as cells transform to not only cancer but also pre-cancerous conditions. This systems level behavior could not be attributed to a group of genes.

Identification of a Tumor Microenvironment-Related Eight-Gene Signature for Predicting Prognosis in Lower-Grade Gliomas

Lower-grade gliomas (LrGG), characterized by invasiveness and heterogeneity, remain incurable with current therapies. Clinicopathological features provide insufficient information to guide optimal individual treatment and cannot predict prognosis completely. Recently, an increasing amount of studies indicate that the tumor microenvironment plays a pivotal role in tumor malignancy and treatment responses. However, studies focusing on the tumor microenvironment (TME) of LrGG are still limited. In this study, taking advantage of the currently popular computational methods for estimating the infiltration of tumor-associated normal cells in tumor samples and using weighted gene co-expression network analysis, we screened the co-expressed gene modules associated with the TME and further identified the prognostic hub genes in these modules. Furthermore, eight prognostic hub genes (ARHGDIB, CLIC1, OAS3, PDIA4, PARP9, STAT1, TAP2, and TAGLN2) were selected to construct a prognostic risk score model using the least absolute shrinkage and selection operator method. Univariate and multivariate Cox regression analysis demonstrated that the risk score was an independent prognostic factor for LrGG. Moreover, time-dependent ROC curves indicated that our model had favorable efficiency in predicting both short- and long-term prognosis in LrGG patients, and the stratified survival analysis demonstrated that our model had prognostic value for different subgroups of LrGG patients. Additionally, our model had potential value for predicting the sensitivity of LrGG patients to radio- and chemotherapy. Besides, differential expression analysis showed that the eight genes were aberrantly expressed in LrGG compared to normal brain tissue. Correlation analysis revealed that the expression of the eight genes was significantly associated with the infiltration levels of six types of immune cells in LrGG. In summary, the TME-related eight-gene signature was significantly associated with the prognosis of LrGG patients. They might act as potential prognostic biomarkers for LrGG patients, offer better stratification for future clinical trials, and be candidate targets for immunotherapy, thus deserving further investigation.

Identifying novel biomarkers of the pediatric influenza infection by weighted co-expression network analysis

Background

Despite the high yearly prevalence of Influenza, the pathogenesis mechanism and involved genes have not been fully known. Finding the patterns and mapping the complex interactions between different genes help us to find the possible biomarkers and treatment targets.

Methods

Herein, weighted gene co-expression network analysis (WGCNA) was employed to construct a co-expression network among genes identified by microarray analysis of the pediatric influenza-infected samples.

Results

Three of the 38 modules were found as the most related modules to influenza infection. At a functional level, we found that the genes in these modules regulate the immune responses, protein targeting, and defense to virus. Moreover, the analysis of differentially expressed genes disclosed 719 DEGs between the normal and infected subjects. The comprehensive investigation of genes in the module involved in immune system and viral defense (yellow module) revealed that SP110, HERC5, SAMD9L, RTP4, C19orf66, HELZ2, EPSTI1, and PHF11 which were also identified as DEGs (except C19orf66) have the potential to be as the biomarkers and also drug targeting for the treatment of pediatric influenza.

Conclusions

The WGCN analysis revealed co-expressed genes which were involved in the innate immune system and defense to virus. The differentially expressed genes in the identified modules can be considered for designing drug targets. Moreover, modules can help to find pathogenesis routes in the future.

Gene networks in cancer are biased by aneuploidies and sample impurities

Gene regulatory network inference is a standard technique for obtaining structured regulatory information from, for instance, gene expression measurements. Methods performing this task have been extensively evaluated on synthetic, and to a lesser extent real data sets. In contrast to these test evaluations, applications to gene expression data of human cancers are often limited by fewer samples and more potential regulatory links, and are biased by copy number aberrations as well as cell mixtures and sample impurities. Here, we take networks inferred from TCGA cohorts as an example to show that (1) transcription factor annotations are essential to obtain reliable networks, and (2) even for state of the art methods, we expect that between 20 and 80% of edges are caused by copy number changes and cell mixtures rather than transcription factor regulation.

Measurement of Conditional Relatedness Between Genes Using Fully Convolutional Neural Network

Measuring conditional relatedness, the degree of relation between a pair of genes in a certain condition, is a basic but difficult task in bioinformatics, as traditional co-expression analysis methods rely on co-expression similarities, well known with high false positive rate. Complement with prior-knowledge similarities is a feasible way to tackle the problem. However, classical combination machine learning algorithms fail in detection and application of the complex mapping relations between similarities and conditional relatedness, so a powerful predictive model will have enormous benefit for measuring this kind of complex mapping relations. To this need, we propose a novel deep learning model of convolutional neural network with a fully connected first layer, named fully convolutional neural network (FCNN), to measure conditional relatedness between genes using both co-expression and prior-knowledge similarities. The results on validation and test datasets show FCNN model yields an average 3.0% and 2.7% higher accuracy values for identifying gene–gene interactions collected from the COXPRESdb, KEGG, and TRRUST databases, and a benchmark dataset of Xiao-Yong et al. research, by grid-search 10-fold cross validation, respectively. In order to estimate the FCNN model, we conduct a further verification on the GeneFriends and DIP datasets, and the FCNN model obtains an average of 1.8% and 7.6% higher accuracy, respectively. Then the FCNN model is applied to construct cancer gene networks, and also calls more practical results than other compared models and methods. A website of the FCNN model and relevant datasets can be accessed from https://bmbl.bmi.osumc.edu/FCNN.

miR-92b-3p Functions As A Key Gene In Esophageal Squamous Cell Cancer As Determined By Co-Expression Analysis

Background

Esophageal squamous cell carcinoma (ESCC) is a highly aggressive malignancy. The aims of the present study were to screen the critical miRNA and corresponding target genes that related to development of ESCC by weighted gene correlation network analysis (WGCNA) and investigate the functions by experimental validation.

Methods

Datasets of mRNA and miRNA expression data were downloaded from GEO. The R software was used for data preprocessing and differential expression gene analysis. The differentially expressed protein-coding genes (DEGs) and miRNAs (DEMs) were selected (FDR <0.05 or |Fold Change (FC)| >1.5). Meanwhile, 81 expression data of ESCC patients in TCGA combined with clinic information were applied by WGCNA to create networks. The correlational analyses between each module and clinical parameters were conducted, and enrichment analyses of GO and KEGG were subsequently performed. Then, a series of experiments were conducted in ESCC cells by use of miRNA mimics.

Results

In total, 4,023 DEGs and 328 DEMs were screened. After checking good genes and samples, 3,841 genes (3,696 DEGs and 145 DEMs) were used for WGCNA. As a consequence, altogether 11 gene modules were found. Among them, the brown modules were found to be strongly inversely associated with pathological grade. Meanwhile, has-mir-92b, the only miRNA in brown module, had a positive correlation with grade and negatively correlated with potential target gene (KFL4 and DCS2) in the same module. Furthermore, an increased expression of miR-92b-3p and down-regulated KLF4 and DSC2 protein was detected in the ESCC clinical samples. Up-regulated miR-92b-3p shortened G0/G1 phase and promote ESCC cells invasion and migration. Furthermore, we verified that DSC2 and KFL4 was target genes of miR-92b-3p by luciferase report assay.

Conclusion

WGCNA is an efficient approach to system biology. By this procedure, miR-92b-3p was identified as an ESCC-promoting gene by target KLF4 and DCS2.

Challenges in microRNAs' targetome prediction and validation

MicroRNAs (miRNAs) are small RNA molecules with important roles in post-transcriptional regulation of gene expression. In recent years, the predicted number of miRNAs has skyrocketed, largely as a consequence of high-throughput sequencing technologies becoming ubiquitous. This dramatic increase in miRNA candidates poses multiple challenges in terms of data deposition, curation, and validation. Although multiple databases containing miRNA annotations and targets have been developed, ensuring data quality by validating miRNA-target interactions requires the efforts of the research community. In order to generate databases containing biologically active miRNAs, it is imperative to overcome a multitude of hurdles, including restricted miRNA expression patterns, distinct miRNA biogenesis machineries, and divergent miRNA-mRNA interaction dynamics. In the present review, we discuss recent advances and limitations in miRNA prediction, identification, and validation. Lastly, we focus on the most enriched neuronal miRNA, miR-124, and its gene regulatory network in human neurons, which has been revealed using a combined computational and experimental approach.

Transcriptome and Gene Coexpression Network Analyses of Two Wild Populations Provides Insight into the High-Salinity Adaptation Mechanisms of Crassostrea ariakensis

Crassostrea ariakensis naturally distributes in the intertidal and estuary region with relative low salinity ranging from 10 to 25‰. To understand the adaptive capacity of oysters to salinity stress, we conducted transcriptome analysis to investigate the metabolic pathways of salinity stress effectors in oysters from two different geographical sites, namely at salinities of 16, 23, and 30‰. We completed transcriptome sequencing of 18 samples and a total of 52,392 unigenes were obtained after assembly. Differentially expressed gene (DEG) analysis and weighted gene correlation network analysis (WGCNA) were performed using RNA-Seq transcriptomic data from eye-spot larvae at different salinities and from different populations. The results showed that at moderately high salinities (23 and 30‰), genes related to osmotic agents, oxidation-reduction processes, and related regulatory networks of complex transcriptional regulation and signal transduction pathways dominated to counteract the salinity stress. Moreover, there were adaptive differences in salinity response mechanisms, especially at high salinity, in oyster larvae from different populations. These results provide a framework for understanding the interactions of multiple pathways at the system level and for elucidating the complex cellular processes involved in responding to osmotic stress and maintaining growth. Furthermore, the results facilitate further research into the biological processes underlying physiological adaptations to hypertonic stress in marine invertebrates and provide a molecular basis for our subsequent search for high salinity-tolerant populations.

Coordinated Activity of Transcriptional Networks Responding to the Pattern of Action Potential Firing in Neurons

Transcriptional responses to the appropriate temporal pattern of action potential firing are essential for long-term adaption of neuronal properties to the functional activity of neural circuits and environmental experience. However, standard transcriptome analysis methods can be too limited in identifying critical aspects that coordinate temporal coding of action potential firing with transcriptome response. A Pearson correlation analysis was applied to determine how pairs of genes in the mouse dorsal root ganglion (DRG) neurons are coordinately expressed in response to stimulation producing the same number of action potentials by two different temporal patterns. Analysis of 4728 distinct gene-pairs related to calcium signaling, 435,711 pairs of transcription factors, 820 pairs of voltage-gated ion channels, and 86,862 pairs of calcium signaling genes with transcription factors indicated that genes become coordinately activated by distinct action potential firing patterns and this depends on the duration of stimulation. Moreover, a measure of expression variance revealed that the control of transcripts abundances is sensitive to the pattern of stimulation. Thus, action potentials impact intracellular signaling and the transcriptome in dynamic manner that not only alter gene expression levels significantly (as previously reported) but also affects the control of their expression fluctuations and profoundly remodel the transcriptional networks.

TGCnA: temporal gene coexpression network analysis using a low-rank plus sparse framework

Various gene network models with distinct physical nature have been widely used in biological studies. For temporal transcriptomic studies, the current dynamic models either ignore the temporal variation in the network structure or fail to scale up to a large number of genes due to severe computational bottlenecks and sample size limitation. Although the correlation-based gene networks are computationally affordable, they have limitations after being applied to gene expression time-course data. We proposed Temporal Gene Coexpression Network Analysis (TGCnA) framework for the transcriptomic time-course data. The mathematical nature of TGCnA is the joint modeling of multiple covariance matrices across time points using a 'low-rank plus sparse' framework, in which the network similarity across time points is explicitly modeled in the low-rank component. We demonstrated the advantage of TGCnA in covariance matrix estimation and gene module discovery using both simulation data and real transcriptomic data. The code is available at https://github.com/QiZhangStat/TGCnA.

Distinct signatures of lung cancer types: aberrant mucin O-glycosylation and compromised immune response

BACKGROUND

Genomic initiatives such as The Cancer Genome Atlas (TCGA) contain data from -omics profiling of thousands of tumor samples, which may be used to decipher cancer signaling, and related alterations. Managing and analyzing data from large-scale projects, such as TCGA, is a demanding task. It is difficult to dissect the high complexity hidden in genomic data and to account for inter-tumor heterogeneity adequately.

METHODS

In this study, we used a robust statistical framework along with the integration of diverse bioinformatic tools to analyze next-generation sequencing data from more than 1000 patients from two different lung cancer subtypes, i.e., the lung adenocarcinoma (LUAD) and the squamous cell carcinoma (LUSC).

RESULTS

We used the gene expression data to identify co-expression modules and differentially expressed genes to discriminate between LUAD and LUSC. We identified a group of genes which could act as specific oncogenes or tumor suppressor genes in one of the two lung cancer types, along with two dual role genes. Our results have been validated against other transcriptomics data of lung cancer patients.

CONCLUSIONS

Our integrative approach allowed us to identify two key features: a substantial up-regulation of genes involved in O-glycosylation of mucins in LUAD, and a compromised immune response in LUSC. The immune-profile associated with LUSC might be linked to the activation of three oncogenic pathways, which promote the evasion of the antitumor immune response. Collectively, our results provide new future directions for the design of target therapies in lung cancer.

Serial gene co-expression network approach to mine biological meanings from integrated transcriptomes of the porcine endometrium during estrous cycle

The estrous cycle is a complex process regulated by several hormones. To understand the dynamic changes in gene expression that takes place in the swine endometrium during the estrous cycle relative to the day of estrus onset, we performed RNA-sequencing analysis on days 0, 3, 6, 9, 12, 15, and 18, resulting in the identification of 4495 differentially expressed genes (DEGs; Q ≤ 0.05 and |log₂FC| ≥ 1) at various phases in the estrous cycle. These DEGs were integrated into multiple gene co-expression networks based on different fold changes and correlation coefficient (R²) thresholds and a suitable network, which included 899 genes (|log2FC| ≥ 2 and R² ≥ 0.99), was identified for downstream analyses based on the biological relevance of the Gene Ontology (GO) terms enriched. The genes in this network were partitioned into 6 clusters based on the expression pattern. Several GO terms including cell cycle, apoptosis, hormone signaling, and lipid biosynthetic process were found to be enriched. Furthermore, we found 15 significant KEGG pathways, including cell adhesion molecules, cytokine-cytokine receptor signaling, steroid biosynthesis, and estrogen signaling pathways. We identified several genes and GO terms to be stage-specific. Moreover, the identified genes and pathways extend our understanding of porcine endometrial regulation during estrous cycle and will serve as a good resource for future studies.

EdgeScaping: Mapping the spatial distribution of pairwise gene expression intensities

Gene co-expression networks (GCNs) are constructed from Gene Expression Matrices (GEMs) in a bottom up approach where all gene pairs are tested for correlation within the context of the input sample set. This approach is computationally intensive for many current GEMs and may not be scalable to millions of samples. Further, traditional GCNs do not detect non-linear relationships missed by correlation tests and do not place genetic relationships in a gene expression intensity context. In this report, we propose EdgeScaping, which constructs and analyzes the pairwise gene intensity network in a holistic, top down approach where no edges are filtered. EdgeScaping uses a novel technique to convert traditional pairwise gene expression data to an image based format. This conversion not only performs feature compression, making our algorithm highly scalable, but it also allows for exploring non-linear relationships between genes by leveraging deep learning image analysis algorithms. Using the learned embedded feature space we implement a fast, efficient algorithm to cluster the entire space of gene expression relationships while retaining gene expression intensity. Since EdgeScaping does not eliminate conventionally noisy edges, it extends the identification of co-expression relationships beyond classically correlated edges to facilitate the discovery of novel or unusual expression patterns within the network. We applied EdgeScaping to a human tumor GEM to identify sets of genes that exhibit conventional and non-conventional interdependent non-linear behavior associated with brain specific tumor sub-types that would be eliminated in conventional bottom-up construction of GCNs. Edgescaping source code is available at https://github.com/bhusain/EdgeScaping under the MIT license.

Network-based feature selection reveals substructures of gene modules responding to salt stress in rice

Rice, an important food resource, is highly sensitive to salt stress, which is directly related to food security. Although many studies have identified physiological mechanisms that confer tolerance to the osmotic effects of salinity, the link between rice genotype and salt tolerance is not very clear yet. Association of gene co-expression network and rice phenotypic data under stress has penitential to identify stress-responsive genes, but there is no standard method to associate stress phenotype with gene co-expression network. A novel method for integration of gene co-expression network and stress phenotype data was developed to conduct a system analysis to link genotype to phenotype. We applied a LASSO-based method to the gene co-expression network of rice with salt stress to discover key genes and their interactions for salt tolerance-related phenotypes. Submodules in gene modules identified from the co-expression network were selected by the LASSO regression, which establishes a linear relationship between gene expression profiles and physiological responses, that is, sodium/potassium condenses under salt stress. Genes in these submodules have functions related to ion transport, osmotic adjustment, and oxidative tolerance. We argued that these genes in submodules are biologically meaningful and useful for studies on rice salt tolerance. This method can be applied to other studies to efficiently and reliably integrate co-expression network and phenotypic data.

A comprehensive study on genome-wide coexpression network of KHDRBS1/Sam68 reveals its cancer and patient-specific association

Human KHDRBS1/Sam68 is an oncogenic splicing factor involved in signal transduction and pre-mRNA splicing. We explored the molecular mechanism of KHDRBS1 to be a prognostic marker in four different cancers. Within specific cancer, including kidney renal papillary cell carcinoma (KIRP), lung adenocarcinoma (LUAD), acute myeloid leukemia (LAML), and ovarian cancer (OV), KHDRBS1 expression is heterogeneous and patient specific. In KIRP and LUAD, higher expression of KHDRBS1 affects the patient survival, but not in LAML and OV. Genome-wide coexpression analysis reveals genes and transcripts which are coexpressed with KHDRBS1 in KIRP and LUAD, form the functional modules which are majorly involved in cancer-specific events. However, in case of LAML and OV, such modules are absent. Irrespective of the higher expression of KHDRBS1, the significant divergence of its biological roles and prognostic value is due to its cancer-specific interaction partners and correlation networks. We conclude that rewiring of KHDRBS1 interactions in cancer is directly associated with patient prognosis.

clpV is a key virulence gene during in vivo Pseudomonas plecoglossicida infection

Interaction between bacterial pathogen and aquatic animal host is exceedingly complex, which involves large dynamic changes in gene expression during different stages of the disease. However, research on identifying key virulence genes based on the dynamics of gene expression changes of a one-sided bacterial pathogen in tissue has not been reported so far across different stages of infectious disease. The clpV for the T6SS of Pseudomonas plecoglossicida was identified for a candidate for key virulence gene based on dynamic changes of gene expression. For the Epinephelus coioides infected using clpV-RNAi strain, no deaths were observed up to 20 dpi. The spleens, kidneys and livers of all the E. coioides that received clpV-RNAi strain failed to develop visible nodules at 5-8 dpi, with the swelling gradually disappearing. The burdens of clpV-RNAi strain in the spleen and blood were greatly reduced at most of the time points after injection, and the burdens of clpV-RNAi strain in the head kidneys and trunk kidneys also had a sharp reduction from 72 to 120 hpi. This paper provides a new insight into the discovery of key virulence genes of pathogens in infected tissue systems.