Identifying Candidate Gene Drivers Associated with Relapse in Pediatric T-Cell Acute Lymphoblastic Leukemia Using a Gene Co-Expression Network Approach

Pediatric T-cell Acute Lymphoblastic Leukemia (T-ALL) relapses are still associated with a dismal outcome, justifying the search for new therapeutic targets and relapse biomarkers. Using single-cell RNA sequencing (scRNAseq) data from three paired samples of pediatric T-ALL at diagnosis and relapse, we first conducted a high-dimensional weighted gene co-expression network analysis (hdWGCNA). This analysis highlighted several gene co-expression networks (GCNs) and identified relapse-associated hub genes, which are considered potential driver genes. Shared relapse-expressed genes were found to be related to antigen presentation (HLA, B2M), cytoskeleton remodeling (TUBB, TUBA1B), translation (ribosomal proteins, EIF1, EEF1B2), immune responses (MIF, EMP3), stress responses (UBC, HSP90AB1/AA1), metabolism (FTH1, NME1/2, ARCL4C), and transcriptional remodeling (NF-κB family genes, FOS-JUN, KLF2, or KLF6). We then utilized sparse partial least squares discriminant analysis to select from a pool of 481 unique leukemic hub genes, which are the genes most discriminant between diagnosis and relapse states (comprising 44, 35, and 31 genes, respectively, for each patient). Applying a Cox regression method to these patient-specific genes, along with transcriptomic and clinical data from the TARGET-ALL AALL0434 cohort, we generated three model gene signatures that efficiently identified relapsed patients within the cohort. Overall, our approach identified new potential relapse-associated genes and proposed three model gene signatures associated with lower survival rates for high-score patients.

Qualitative and quantitative concentration-response modelling of gene co-expression networks to unlock hepatotoxic mechanisms for next generation chemical safety assessment

Next generation risk assessment of chemicals revolves around the use of mechanistic information without animal experimentation. In this regard, toxicogenomics has proven to be a useful tool to elucidate the underlying mechanisms of adverse effects of xenobiotics. In the present study, two widely used human in vitro hepatocyte culture systems, namely primary human hepatocytes (PHH) and human hepatoma HepaRG cells, were exposed to liver toxicants known to induce liver cholestasis, steatosis or necrosis. Benchmark concentration-response modelling was applied to transcriptomics gene co-expression networks (modules) to derive benchmark concentrations (BMCs) and to gain mechanistic insight into the hepatotoxic effects. BMCs derived by concentration-response modelling of gene co-expression modules recapitulated concentration-response modelling of individual genes. Although PHH and HepaRG cells showed overlap in deregulated genes and modules by the liver toxicants, PHH demonstrated a higher responsiveness, based on the lower BMCs of co-regulated gene modules. Such BMCs can be used as transcriptomics point of departure (tPOD) for assessing module-associated cellular (stress) pathways/processes. This approach identified clear tPODs of around maximum systemic concentration (Cmax) levels for the tested drugs, while for cosmetics ingredients the BMCs were 10-100-fold higher than the estimated plasma concentrations. This approach could serve next generation risk assessment practice to identify early responsive modules at low BMCs, that could be linked to key events in liver adverse outcome pathways. In turn, this can assist in delineating potential hazards of new test chemicals using in vitro systems and used in a risk assessment when BMCs are paired with chemical exposure assessment.

The Role of Transcription Factors in the Loss of Inter-Chromosomal Co-Expression for Breast Cancer Subtypes

Breast cancer encompasses a diverse array of subtypes, each exhibiting distinct clinical characteristics and treatment responses. Unraveling the underlying regulatory mechanisms that govern gene expression patterns in these subtypes is essential for advancing our understanding of breast cancer biology. Gene co-expression networks (GCNs) help us identify groups of genes that work in coordination. Previous research has revealed a marked reduction in the interaction of genes located on different chromosomes within GCNs for breast cancer, as well as for lung, kidney, and hematopoietic cancers. However, the reasons behind why genes on the same chromosome often co-express remain unclear. In this study, we investigate the role of transcription factors in shaping gene co-expression networks within the four main breast cancer subtypes: Luminal A, Luminal B, HER2+, and Basal, along with normal breast tissue. We identify communities within each GCN and calculate the transcription factors that may regulate these communities, comparing the results across different phenotypes. Our findings indicate that, in general, regulatory behavior is to a large extent similar among breast cancer molecular subtypes and even in healthy networks. This suggests that transcription factor motif usage does not fully determine long-range co-expression patterns. Specific transcription factor motifs, such as CCGGAAG, appear frequently across all phenotypes, even involving multiple highly connected transcription factors. Additionally, certain transcription factors exhibit unique actions in specific subtypes but with limited influence. Our research demonstrates that the loss of inter-chromosomal co-expression is not solely attributable to transcription factor regulation. Although the exact mechanism responsible for this phenomenon remains elusive, this work contributes to a better understanding of gene expression regulatory programs in breast cancer.

The non-specific lethal complex regulates genes and pathways genetically linked to Parkinson's disease

Genetic variants conferring risks for Parkinson's disease have been highlighted through genome-wide association studies, yet exploration of their specific disease mechanisms is lacking. Two Parkinson's disease candidate genes, KAT8 and KANSL1, identified through genome-wide studies and a PINK1-mitophagy screen, encode part of the histone acetylating non-specific lethal complex. This complex localizes to the nucleus, where it plays a role in transcriptional activation, and to mitochondria, where it has been suggested to have a role in mitochondrial transcription. In this study, we sought to identify whether the non-specific lethal complex has potential regulatory relationships with other genes associated with Parkinson's disease in human brain. Correlation in the expression of non-specific lethal genes and Parkinson's disease-associated genes was investigated in primary gene co-expression networks using publicly-available transcriptomic data from multiple brain regions (provided by the Genotype-Tissue Expression Consortium and UK Brain Expression Consortium), whilst secondary networks were used to examine cell type specificity. Reverse engineering of gene regulatory networks generated regulons of the complex, which were tested for heritability using stratified linkage disequilibrium score regression. Prioritized gene targets were then validated in vitro using a QuantiGene multiplex assay and publicly-available chromatin immunoprecipitation-sequencing data. Significant clustering of non-specific lethal genes was revealed alongside Parkinson's disease-associated genes in frontal cortex primary co-expression modules, amongst other brain regions. Both primary and secondary co-expression modules containing these genes were enriched for mainly neuronal cell types. Regulons of the complex contained Parkinson's disease-associated genes and were enriched for biological pathways genetically linked to disease. When examined in a neuroblastoma cell line, 41% of prioritized gene targets showed significant changes in mRNA expression following KANSL1 or KAT8 perturbation. KANSL1 and H4K8 chromatin immunoprecipitation-sequencing data demonstrated non-specific lethal complex activity at many of these genes. In conclusion, genes encoding the non-specific lethal complex are highly correlated with and regulate genes associated with Parkinson's disease. Overall, these findings reveal a potentially wider role for this protein complex in regulating genes and pathways implicated in Parkinson's disease.

Demographic confounders distort inference of gene regulatory and gene co-expression networks in cancer

Gene regulatory networks (GRNs) and gene co-expression networks (GCNs) allow genome-wide exploration of molecular regulation patterns in health and disease. The standard approach for obtaining GRNs and GCNs is to infer them from gene expression data, using computational network inference methods. However, since network inference methods are usually applied on aggregate data, distortion of the networks by demographic confounders might remain undetected, especially because gene expression patterns are known to vary between different demographic groups. In this paper, we present a computational framework to systematically evaluate the influence of demographic confounders on network inference from gene expression data. Our framework compares similarities between networks inferred for different demographic groups with similarity distributions obtained for random splits of the expression data. Moreover, it allows to quantify to which extent demographic groups are represented by networks inferred from the aggregate data in a confounder-agnostic way. We apply our framework to test four widely used GRN and GCN inference methods as to their robustness w. r. t. confounding by age, ethnicity and sex in cancer. Our findings based on more than $ {44000}$ inferred networks indicate that age and sex confounders play an important role in network inference for certain cancer types, emphasizing the importance of incorporating an assessment of the effect of demographic confounders into network inference workflows. Our framework is available as a Python package on GitHub: https://github.com/bionetslab/grn-confounders.

Analysis of High-Risk Neuroblastoma Transcriptome Reveals Gene Co-Expression Signatures and Functional Features

Neuroblastoma represents a neoplastic expansion of neural crest cells in the developing sympathetic nervous system and is childhood's most common extracranial solid tumor. The heterogeneity of gene expression in different types of cancer is well-documented, and genetic features of neuroblastoma have been described by classification, development stage, malignancy, and progression of tumors. Here, we aim to analyze RNA sequencing datasets, publicly available in the GDC data portal, of neuroblastoma tumor samples from various patients and compare them with normal adrenal gland tissue from the GTEx data portal to elucidate the gene expression profile and regulation networks they share. Our results from the differential expression, weighted correlation network, and functional enrichment analyses that we performed with the count data from neuroblastoma and standard normal gland samples indicate that the analysis of transcriptome data from 58 patients diagnosed with high-risk neuroblastoma shares the expression pattern of 104 genes. More importantly, our analyses identify the co-expression relationship and the role of these genes in multiple biological processes and signaling pathways strongly associated with this disease phenotype. Our approach proposes a group of genes and their biological functions to be further investigated as essential molecules and possible therapeutic targets of neuroblastoma regardless of the etiology of individual tumors.

Methylation-related genes involved in renal carcinoma progression

Renal carcinomas are a group of malignant tumors often originating in the cells lining the small tubes in the kidney responsible for filtering waste from the blood and urine production. Kidney tumors arise from the uncontrolled growth of cells in the kidneys and are responsible for a large share of global cancer-related morbidity and mortality. Understanding the molecular mechanisms driving renal carcinoma progression results crucial for the development of targeted therapies leading to an improvement of patient outcomes. Epigenetic mechanisms such as DNA methylation are known factors underlying the development of several cancer types. There is solid experimental evidence of relevant biological functions modulated by methylation-related genes, associated with the progression of different carcinomas. Those mechanisms can often be associated to different epigenetic marks, such as DNA methylation sites or chromatin conformation patterns. Currently, there is no definitive method to establish clear relations between genetic and epigenetic factors that influence the progression of cancer. Here, we developed a data-driven method to find methylation-related genes, so we could find relevant bonds between gene co-expression and methylation-wide-genome regulation patterns able to drive biological processes during the progression of clear cell renal carcinoma (ccRC). With this approach, we found out genes such as ITK oncogene that appear hypomethylated during all four stages of ccRC progression and are strongly involved in immune response functions. Also, we found out relevant tumor suppressor genes such as RAB25 hypermethylated, thus potentially avoiding repressed functions in the AKT signaling pathway during the evolution of ccRC. Our results have relevant implications to further understand some epigenetic-genetic-affected roles underlying the progression of renal cancer.

A Novel Role of Medicago truncatula KNAT3/4/5-like Class 2 KNOX Transcription Factors in Drought Stress Tolerance

Class 2 KNOX homeobox transcription factors (KNOX2) play a role in promoting cell differentiation in several plant developmental processes. In Arabidopsis, they antagonize the meristematic KNOX1 function during leaf development through the modulation of phytohormones. In Medicago truncatula, three KNOX2 genes belonging to the KNAT3/4/5-like subclass (Mt KNAT3/4/5-like or MtKNOX3-like) redundantly works upstream of a cytokinin-signaling module to control the symbiotic root nodule formation. Their possible role in the response to abiotic stress is as-of-yet unknown. We produced transgenic M. truncatula lines, in which the expression of four MtKNOX3-like genes was knocked down by RNA interference. When tested for response to water withdrawal in the soil, RNAi lines displayed a lower tolerance to drought conditions compared to the control lines, measured as increased leaf water loss, accelerated leaf wilting time, and faster chlorophyll loss. Reanalysis of a transcriptomic M. truncatula drought stress experiment via cluster analysis and gene co-expression networks pointed to a possible role of MtKNOX3-like transcription factors in repressing a proline dehydrogenase gene (MtPDH), specifically at 4 days after water withdrawal. Proline measurement and gene expression analysis of transgenic RNAi plants compared to the controls confirmed the role of KNOX3-like genes in inhibiting proline degradation through the regulation of the MtPDH gene.

CNVs in 8q24.3 do not influence gene co-expression in breast cancer subtypes

Gene co-expression networks are a useful tool in the study of interactions that have allowed the visualization and quantification of diverse phenomena, including the loss of co-expression over long distances in cancerous samples. This characteristic, which could be considered fundamental to cancer, has been widely reported in various types of tumors. Since copy number variations (CNVs) have previously been identified as causing multiple genetic diseases, and gene expression is linked to them, they have often been mentioned as a probable cause of loss of co-expression in cancerous networks. In order to carry out a comparative study of the validity of this statement, we took 477 protein-coding genes from chromosome 8, and the CNVs of 101 genes, also protein-coding, belonging to the 8q24.3 region, a cytoband that is particularly active in the appearance of breast cancer. We created CNVS-conditioned co-expression networks of each of the 101 genes in the 8q24.3 region using conditional mutual information. The study was carried out using the four molecular subtypes of breast cancer (Luminal A, Luminal B, Her2, and Basal), as well as a case corresponding to healthy samples. We observed that in all cancer cases, the measurement of the Kolmogorov-Smirnov statistic shows that there are no significant differences between one and other values of the CNVs for any case. Furthermore, the co-expression interactions are stronger in all cancer subtypes than in the control networks. However, the control network presents a homogeneously distributed set of co-expression interactions, while for cancer networks, the highest interactions are more confined to specific cytobands, in particular 8q24.3 and 8p21.3. With this approach, we demonstrate that despite copy number alterations in the 8q24 region being a common trait in breast cancer, the loss of long-distance co-expression in breast cancer is not determined by CNVs.

Development and validation of a prognostic model and gene co-expression networks for breast carcinoma based on scRNA-seq and bulk-seq data

Background

Breast carcinoma is the most common malignancy among women worldwide. It is characterized by a complex tumor microenvironment (TME), in which there is an intricate combination of different types of cells, which can cause confusion when screening tumor-cell-related signatures or constructing a gene co-expression network. The recent emergence of single-cell RNA sequencing (scRNA-seq) is an effective method for studying the changing omics of cells in complex TMEs.

Methods

The Dysregulated genes of malignant epithelial cells was screened by performing a comprehensive analysis of the public scRNA-seq data of 58 samples. Co-expression and Gene Set Enrichment Analysis (GSEA) analysis were performed based on scRNA-seq data of malignant cells to illustrate the potential function of these dysregulated genes. Iterative LASSO-Cox was used to perform a second-round screening among these dysregulated genes for constructing risk group. Finally, a breast cancer prognosis prediction model was constructed based on risk grouping and other clinical characteristics.

Results

Our results indicated a transcriptional signature of 1,262 genes for malignant breast cancer epithelial cells. To estimate the function of these genes in breast cancer, we also constructed a co-expression network of these dysregulated genes at single-cell resolution, and further validated the results using more than 300 published transcriptomics datasets and 31 Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) screening datasets. Moreover, we developed a reliable predictive model based on the scRNA-seq and bulk-seq datasets.

Conclusions

Our findings provide insights into the transcriptomics and gene co-expression networks during breast carcinoma progression and suggest potential candidate biomarkers and therapeutic targets for the treatment of breast carcinoma. Our results are available via a web app (https://prognosticpredictor.shinyapps.io/GCNBC/).

Integrated proteomics and metabolomics analysis reveals hubs protein and network alterations in myasthenia gravis

Background: Thymoma-associated myasthenia gravis (TAMG) is a well-described subtype of Myasthenia gravis (MG). Nevertheless, the detailed proteins and bioprocess differentiating TAMG from TAMG (−) thymoma have remained unclear.

Methods: The proteomics and metabolomics were carried out on serum samples from thymoma group (n = 60, TNMG), TAMG (+) thymoma group (n = 70, TAMG (+)), and TAMG (−) thymomas group (n = 62, TAMG (−)), and controls (n = 159). groups. Proteomics and metabolomics analyses, including weighted gene co-expression network analysis (WGCNA), was conducted to detect the hub proteins and metabolomics processes that could differentiate TAMG (+) from TAMG (−) thymomas. MetaboAnalyst was used to examine the integration of proteomic and metabolomic analysis to differentiate TAMG (+) from TAMG (−) thymomas.

Results: The of module–trait correlation of WGCNA analysis identified KRT1, GSN, COL6A1, KRT10, FOLR2, KRT9, KRT2, TPI1, ARF3, LYZ, ADIPOQ, SEMA4B, IGKV1-27, MASP2, IGF2R was associated with TAMG (+) thymomas. In addition, organismal systems-immune system and metabolism-biosynthesis of other secondary metabolites were closely related to the mechanism of TAMG (+) pathogenesis.

Conclusion: Our integrated proteomics and metabolomics analysis supply a systems-level view of proteome changes in TAMG (+), TAMG (−) thymomas and exposes disease-associated protein network alterations involved in.

Comparative Analyses of Gene Co-expression Networks: Implementations and Applications in the Study of Evolution

Similarities and differences in the associations of biological entities among species can provide us with a better understanding of evolutionary relationships. Often the evolution of new phenotypes results from changes to interactions in pre-existing biological networks and comparing networks across species can identify evidence of conservation or adaptation. Gene co-expression networks (GCNs), constructed from high-throughput gene expression data, can be used to understand evolution and the rise of new phenotypes. The increasing abundance of gene expression data makes GCNs a valuable tool for the study of evolution in non-model organisms. In this paper, we cover motivations for why comparing these networks across species can be valuable for the study of evolution. We also review techniques for comparing GCNs in the context of evolution, including local and global methods of graph alignment. While some protein-protein interaction (PPI) bioinformatic methods can be used to compare co-expression networks, they often disregard highly relevant properties, including the existence of continuous and negative values for edge weights. Also, the lack of comparative datasets in non-model organisms has hindered the study of evolution using PPI networks. We also discuss limitations and challenges associated with cross-species comparison using GCNs, and provide suggestions for utilizing co-expression network alignments as an indispensable tool for evolutionary studies going forward.

Comparing Statistical Tests for Differential Network Analysis of Gene Modules

Genes often work together to perform complex biological processes, and "networks" provide a versatile framework for representing the interactions between multiple genes. Differential network analysis (DiNA) quantifies how this network structure differs between two or more groups/phenotypes (e.g., disease subjects and healthy controls), with the goal of determining whether differences in network structure can help explain differences between phenotypes. In this paper, we focus on gene co-expression networks, although in principle, the methods studied can be used for DiNA for other types of features (e.g., metabolome, epigenome, microbiome, proteome, etc.). Three common applications of DiNA involve (1) testing whether the connections to a single gene differ between groups, (2) testing whether the connection between a pair of genes differs between groups, or (3) testing whether the connections within a "module" (a subset of 3 or more genes) differs between groups. This article focuses on the latter, as there is a lack of studies comparing statistical methods for identifying differentially co-expressed modules (DCMs). Through extensive simulations, we compare several previously proposed test statistics and a new p-norm difference test (PND). We demonstrate that the true positive rate of the proposed PND test is competitive with and often higher than the other methods, while controlling the false positive rate. The R package discoMod (differentially co-expressed modules) implements the proposed method and provides a full pipeline for identifying DCMs: clustering tools to derive gene modules, tests to identify DCMs, and methods for visualizing the results.

Luminal A Breast Cancer Co-expression Network: Structural and Functional Alterations

Luminal A is the most common breast cancer molecular subtype in women worldwide. These tumors have characteristic yet heterogeneous alterations at the genomic and transcriptomic level. Gene co-expression networks (GCNs) have contributed to better characterize the cancerous phenotype. We have previously shown an imbalance in the proportion of intra-chromosomal (cis-) over inter-chromosomal (trans-) interactions when comparing cancer and healthy tissue GCNs. In particular, for breast cancer molecular subtypes (Luminal A included), the majority of high co-expression interactions connect gene-pairs in the same chromosome, a phenomenon that we have called loss of trans- co-expression. Despite this phenomenon has been described, the functional implication of this specific network topology has not been studied yet. To understand the biological role that communities of co-expressed genes may have, we constructed GCNs for healthy and Luminal A phenotypes. Network modules were obtained based on their connectivity patterns and they were classified according to their chromosomal homophily (proportion of cis-/trans- interactions). A functional overrepresentation analysis was performed on communities in both networks to observe the significantly enriched processes for each community. We also investigated possible mechanisms for which the loss of trans- co-expression emerges in cancer GCN. To this end we evaluated transcription factor binding sites, CTCF binding sites, differential gene expression and copy number alterations (CNAs) in the cancer GCN. We found that trans- communities in Luminal A present more significantly enriched categories than cis- ones. Processes, such as angiogenesis, cell proliferation, or cell adhesion were found in trans- modules. The differential expression analysis showed that FOXM1, CENPA, and CIITA transcription factors, exert a major regulatory role on their communities by regulating expression of their target genes in other chromosomes. Finally, identification of CNAs, displayed a high enrichment of deletion peaks in cis- communities. With this approach, we demonstrate that network topology determine, to at certain extent, the function in Luminal A breast cancer network. Furthermore, several mechanisms seem to be acting together to avoid trans- co-expression. Since this phenomenon has been observed in other cancer tissues, a remaining question is whether the loss of long distance co-expression is a novel hallmark of cancer.

Loss of Long Distance Co-Expression in Lung Cancer

Lung cancer is one of the deadliest, most aggressive cancers. Abrupt changes in gene expression represent an important challenge to understand and fight the disease. Gene co-expression networks (GCNs) have been widely used to study the genomic regulatory landscape of human cancer. Here, based on 1,143 RNA-Seq experiments from the TCGA collaboration, we constructed GCN for the most common types of lung tumors: adenocarcinoma (TAD) and squamous cells (TSCs) as well as their respective control networks (NAD and NSC). We compared the number of intra-chromosome (cis-) and inter-chromosome (trans-) co-expression interactions in normal and cancer GCNs. We compared the number of shared interactions between TAD and TSC, as well as in NAD and NSC, to observe which phenotypes were more alike. By means of an over-representation analysis, we associated network topology features with biological functions. We found that TAD and TSC present mostly cis- small disconnected components, whereas in control GCNs, both types have a giant trans- component. In both cancer networks, we observed cis- components in which genes not only belong to the same chromosome but to the same cytoband or to neighboring cytobands. This supports the hypothesis that in lung cancer, gene co-expression is constrained to small neighboring regions. Despite this loss of distant co-expression observed in TAD and TSC, there are some remaining trans- clusters. These clusters seem to play relevant roles in the carcinogenic processes. For instance, some clusters in TAD and TSC are associated with the immune system, response to virus, or control of gene expression. Additionally, other non-enriched trans- clusters are composed of one gene and several associated pseudo-genes, as in the case of the FTH1 gene. The appearance of those common trans- clusters reflects that the gene co-expression program in lung cancer conserves some aspects for cell maintenance. Unexpectedly, 0.48% of the edges are shared between control networks; conversely, 35% is shared between lung cancer GCNs, a 73-fold larger intersection. This suggests that in lung cancer a process of de-differentiation may be occurring. To further investigate the implications of the loss of distant co-expression, it will become necessary to broaden the investigation with other omic-based approaches. However, the present approach provides a basis for future work toward an integrative perspective of abnormal transcriptional regulatory programs in lung cancer.

K-Module Algorithm: An Additional Step to Improve the Clustering Results of WGCNA Co-Expression Networks

Among biological networks, co-expression networks have been widely studied. One of the most commonly used pipelines for the construction of co-expression networks is weighted gene co-expression network analysis (WGCNA), which can identify highly co-expressed clusters of genes (modules). WGCNA identifies gene modules using hierarchical clustering. The major drawback of hierarchical clustering is that once two objects are clustered together, it cannot be reversed; thus, re-adjustment of the unbefitting decision is impossible. In this paper, we calculate the similarity matrix with the distance correlation for WGCNA to construct a gene co-expression network, and present a new approach called the k-module algorithm to improve the WGCNA clustering results. This method can assign all genes to the module with the highest mean connectivity with these genes. This algorithm re-adjusts the results of hierarchical clustering while retaining the advantages of the dynamic tree cut method. The validity of the algorithm is verified using six datasets from microarray and RNA-seq data. The k-module algorithm has fewer iterations, which leads to lower complexity. We verify that the gene modules obtained by the k-module algorithm have high enrichment scores and strong stability. Our method improves upon hierarchical clustering, and can be applied to general clustering algorithms based on the similarity matrix, not limited to gene co-expression network analysis.

Gene Expression and Co-expression Networks Are Strongly Altered Through Stages in Clear Cell Renal Carcinoma

Clear cell renal carcinoma (ccRC) is a highly heterogeneous and progressively malignant disease. Analyzing ccRC progression in terms of modifications at the molecular and genetic level may help us to develop a broader understanding of its patho-physiology and may give us a glimpse toward improved therapeutics. In this work, by using TCGA data, we studied the molecular progression of the four main ccRC stages (i, ii, iii, iv) in two different yet complementary approaches: (a) gene expression and (b) gene co-expression. For (a) we analyzed the differential gene expression between each stage and the control non-cancer group. We compared the progression molecular signature between stages, and observed those genes that change their expression patterns through progression stages. For (b) we constructed and analyzed co-expression networks for the four ccRC progression stages, as well as for the control phenotype, to observe whether and how the co-expression landscape changes with progression. We separated genomic interactions into intra-chromosome (cis-) and inter-chromosome (trans-). Finally, we intersected those networks and performed functional enrichment analysis. All calculations were made over different network sizes, from the top 100 edges to top 1,000,000. We show that differential expression is quite similar between ccRC progression stages. However, interestingly, two genes, namely SLC6A19 and PLG show a significant progressive decrease in their expression according to ccRC stage, meanwhile two other genes, SAA2-SAA4 and CXCL13 show progressive increase. Despite the high similarity between gene expression profiles, all networks are substantially different between them in terms of their topological features. Control network has a larger proportion of trans- interactions, meanwhile for any stage, the amount of cis- interactions is higher, independent of the network cut-off. The majority of interactions in any network are phenotype-specific. Only 189 interactions are shared between the five networks, and 533 edges are ccRC-specific, independent of the stage. The small resulting connected components in both cases are formed by genes with the same differential expression trend, and are associated with important biological processes, such as cell cycle or immune system, suggesting that activity of these categories follows the differential expression trend. With this approach we have shown that, even if the expression program is similar during ccRC progression, the co-expression programs strongly differ. More research is needed to understand the delicate interplay between expression and co-expression, but this is a first approach to enclose both approaches in an integrative view aimed at a deeper understanding in gene regulation in tumor evolution.

Gender Differential Transcriptome in Gastric and Thyroid Cancers

Cancer has an important and considerable gender differential susceptibility confirmed by several epidemiological studies. Gastric (GC) and thyroid cancer (TC) are examples of malignancies with a higher incidence in males and females, respectively. Beyond environmental predisposing factors, it is expected that gender-specific gene deregulation contributes to this differential incidence. We performed a detailed characterization of the transcriptomic differences between genders in normal and tumor tissues from stomach and thyroid using Genotype-Tissue Expression (GTEx) and The Cancer Genome Atlas (TCGA) data. We found hundreds of sex-biased genes (SBGs). Most of the SBGs shared by normal and tumor belong to sexual chromosomes, while the normal and tumor-specific tend to be found in the autosomes. Expression of several cancer-associated genes is also found to differ between sexes in both types of tissue. Thousands of differentially expressed genes (DEGs) between paired tumor-normal tissues were identified in GC and TC. For both cancers, in the most susceptible gender, the DEGs were mostly under-expressed in the tumor tissue, with an enrichment for tumor-suppressor genes (TSGs). Moreover, we found gene networks preferentially associated to males in GC and to females in TC and correlated with cancer histological subtypes. Our results shed light on the molecular differences and commonalities between genders and provide novel insights in the differential risk underlying these cancers.

Gene Co-expression Is Distance-Dependent in Breast Cancer

Breast carcinomas are characterized by anomalous gene regulatory programs. As is well-known, gene expression programs are able to shape phenotypes. Hence, the understanding of gene co-expression may shed light on the underlying mechanisms behind the transcriptional regulatory programs affecting tumor development and evolution. For instance, in breast cancer, there is a clear loss of inter-chromosomal (trans-) co-expression, compared with healthy tissue. At the same time cis- (intra-chromosomal) interactions are favored in breast tumors. In order to have a deeper understanding of regulatory phenomena in cancer, here, we constructed Gene Co-expression Networks by using TCGA-derived RNA-seq whole-genome samples corresponding to the four breast cancer molecular subtypes, as well as healthy tissue. We quantify the cis-/trans- co-expression imbalance in all phenotypes. Additionally, we measured the association between co-expression and physical distance between genes, and characterized the ratio of intra/inter-cytoband interactions per phenotype. We confirmed loss of trans- co-expression in all molecular subtypes. We also observed that gene cis- co-expression decays abruptly with distance in all tumors in contrast with healthy tissue. We observed co-expressed gene hotspots, that tend to be connected at cytoband regions, and coincide accurately with already known copy number altered regions, such as Chr17q12, or Chr8q24.3 for all subtypes. Our methodology recovered different alterations already reported for specific breast cancer subtypes, showing how co-expression network approaches might help to capture distinct events that modify the cell regulatory program.

Identification of Prognostic Biomarkers for Multiple Solid Tumors Using a Human Villi Development Model

The processes of embryonic development that rely on epithelial-mesenchymal transition (EMT) for the implantation of trophoblast cells are co-opted by tumors, reflecting their inherent uncontrolled characteristics and leading to invasion and metastasis. Although tumorigenesis and embryogenesis have similar EMT characteristics, trophoblasts have been shown to exhibit "physiological metastasis" or be "pseudo-malignant," resulting in different outcomes. The gene co-expression network is the basis of embryonic development and tumorigenesis. We hypothesize that if the gene co-expression network in tumors is "off-track" from that in villi, it is more likely to develop into malignant tumors and have a worse prognosis, and we proposed the "off-track theory" for the first time. In this study, we examined gene co-expression networks in villi and multiple solid tumors. Through network functional enrichment analyses, we found that most tumors and villi exhibited a significantly enriched EMT, but the genes that performed this function were not identical. Then, we identified the "off-track genes" in the EMT-related gene interaction network using the "off-track theory," and through survival analysis, we discovered that the risk score of "off-track genes" was associated with poor survival of cancer patients. Our study indicated that villi development is a reliable and strictly regulated model that can illuminate the trajectory of human cancer development and that the gene co-expression networks in tumor development are "off-track" from those in villi. These "off-track genes" may have a substantial impact on tumor development and could reveal novel prognostic biomarkers.

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.

Transcription Factor Networks in Leaves of Cichorium endivia: New Insights into the Relationship Between Photosynthesis and Leaf Development

Cichorium endivia is a leafy crop closely related to Lactuca sativa that comprises two major botanical varieties characterized by a high degree of intraspecific morphological variation: var. latifolium with broad leaves (escarole) and var. crispum with narrow crisp curly leaves (endive). To investigate the relationship between leaf morphology and photosynthetic activity, escaroles and endives were used as a crop model due to the striking morphological diversity of their leaves. We constructed a leaf database for transcription factors (TFs) and photosynthesis-related genes from a refined C. endivia transcriptome and used RNA-seq transcriptomic data from leaves of four commercial endive and escarole cultivars to explore transcription factor regulatory networks. Cluster and gene co-expression network (GCN) analyses identified two main anticorrelated modules that control photosynthesis. Analysis of the GCN network topological properties identified known and novel hub genes controlling photosynthesis, and candidate developmental genes at the boundaries between shape and function. Differential expression analysis between broad and curly leaves suggested three novel TFs putatively involved in leaf shape diversity. Physiological analysis of the photosynthesis properties and gene expression studies on broad and curly leaves provided new insights into the relationship between leaf shape and function.

Dehydration Stress Memory: Gene Networks Linked to Physiological Responses During Repeated Stresses of Zea mays

Stress memory refers to the observation that an initial, sub-lethal stress alters plants' responses to subsequent stresses. Previous transcriptome analyses of maize seedlings exposed to a repeated dehydration stress has revealed the existence of transcriptional stress memory in Zea mays. Whether drought-related physiological responses also display memory and how transcriptional memory translates into physiological memory are fundamental questions that are still unanswered. Using a systems-biology approach we investigate whether/how transcription memory responses established in the genome-wide analysis of Z. mays correlate with 14 physiological parameters measured during a repeated exposure of maize seedlings to dehydration stress. Co-expression network analysis revealed ten gene modules correlating strongly with particular physiological processes, and one module displaying strong, yet divergent, correlations with several processes suggesting involvement of these genes in coordinated responses across networks. Two processes key to the drought response, stomatal conductance and non-photochemical quenching, displayed contrasting memory patterns that may reflect trade-offs related to metabolic costs versus benefits of cellular protection. The main contribution of this study is the demonstration of coordinated changes in transcription memory responses at the genome level and integrated physiological responses at the cellular level upon repetitive stress exposures. The results obtained by the network-based systems analysis challenge the commonly held view that short-term physiological responses to stress are primarily mediated biochemically.

Discovery and validation of a glioblastoma co-expressed gene module

Tumors exhibit complex patterns of aberrant gene expression. Using a knowledge-independent, noise-reducing gene co-expression network construction software called KINC, we created multiple RNAseq-based gene co-expression networks relevant to brain and glioblastoma biology. In this report, we describe the discovery and validation of a glioblastoma-specific gene module that contains 22 co-expressed genes. The genes are upregulated in glioblastoma relative to normal brain and lower grade glioma samples; they are also hypo-methylated in glioblastoma relative to lower grade glioma tumors. Among the proneural, neural, mesenchymal, and classical glioblastoma subtypes, these genes are most-highly expressed in the mesenchymal subtype. Furthermore, high expression of these genes is associated with decreased survival across each glioblastoma subtype. These genes are of interest to glioblastoma biology and our gene interaction discovery and validation workflow can be used to discover and validate co-expressed gene modules derived from any co-expression network.

Evolutionary Analysis of DELLA-Associated Transcriptional Networks

DELLA proteins are transcriptional regulators present in all land plants which have been shown to modulate the activity of over 100 transcription factors in Arabidopsis, involved in multiple physiological and developmental processes. It has been proposed that DELLAs transduce environmental information to pre-wired transcriptional circuits because their stability is regulated by gibberellins (GAs), whose homeostasis largely depends on environmental signals. The ability of GAs to promote DELLA degradation coincides with the origin of vascular plants, but the presence of DELLAs in other land plants poses at least two questions: what regulatory properties have DELLAs provided to the behavior of transcriptional networks in land plants, and how has the recruitment of DELLAs by GA signaling affected this regulation. To address these issues, we have constructed gene co-expression networks of four different organisms within the green lineage with different properties regarding DELLAs: Arabidopsis thaliana and Solanum lycopersicum (both with GA-regulated DELLA proteins), Physcomitrella patens (with GA-independent DELLA proteins) and Chlamydomonas reinhardtii (a green alga without DELLA), and we have examined the relative evolution of the subnetworks containing the potential DELLA-dependent transcriptomes. Network analysis indicates a relative increase in parameters associated with the degree of interconnectivity in the DELLA-associated subnetworks of land plants, with a stronger effect in species with GA-regulated DELLA proteins. These results suggest that DELLAs may have played a role in the coordination of multiple transcriptional programs along evolution, and the function of DELLAs as regulatory 'hubs' became further consolidated after their recruitment by GA signaling in higher plants.

ChlamyNET: a Chlamydomonas gene co-expression network reveals global properties of the transcriptome and the early setup of key co-expression patterns in the green lineage

BACKGROUND

Chlamydomonas reinhardtii is the model organism that serves as a reference for studies in algal genomics and physiology. It is of special interest in the study of the evolution of regulatory pathways from algae to higher plants. Additionally, it has recently gained attention as a potential source for bio-fuel and bio-hydrogen production. The genome of Chlamydomonas is available, facilitating the analysis of its transcriptome by RNA-seq data. This has produced a massive amount of data that remains fragmented making necessary the application of integrative approaches based on molecular systems biology.

RESULTS

We constructed a gene co-expression network based on RNA-seq data and developed a web-based tool, ChlamyNET, for the exploration of the Chlamydomonas transcriptome. ChlamyNET exhibits a scale-free and small world topology. Applying clustering techniques, we identified nine gene clusters that capture the structure of the transcriptome under the analyzed conditions. One of the most central clusters was shown to be involved in carbon/nitrogen metabolism and signalling, whereas one of the most peripheral clusters was involved in DNA replication and cell cycle regulation. The transcription factors and regulators in the Chlamydomonas genome have been identified in ChlamyNET. The biological processes potentially regulated by them as well as their putative transcription factor binding sites were determined. The putative light regulated transcription factors and regulators in the Chlamydomonas genome were analyzed in order to provide a case study on the use of ChlamyNET. Finally, we used an independent data set to cross-validate the predictive power of ChlamyNET.

CONCLUSIONS

The topological properties of ChlamyNET suggest that the Chlamydomonas transcriptome posseses important characteristics related to error tolerance, vulnerability and information propagation. The central part of ChlamyNET constitutes the core of the transcriptome where most authoritative hub genes are located interconnecting key biological processes such as light response with carbon and nitrogen metabolism. Our study reveals that key elements in the regulation of carbon and nitrogen metabolism, light response and cell cycle identified in higher plants were already established in Chlamydomonas. These conserved elements are not only limited to transcription factors, regulators and their targets, but also include the cis-regulatory elements recognized by them.

Epigenetic modulation of brain gene networks for cocaine and alcohol abuse

Cocaine and alcohol are two substances of abuse that prominently affect the central nervous system (CNS). Repeated exposure to cocaine and alcohol leads to longstanding changes in gene expression, and subsequent functional CNS plasticity, throughout multiple brain regions. Epigenetic modifications of histones are one proposed mechanism guiding these enduring changes to the transcriptome. Characterizing the large number of available biological relationships as network models can reveal unexpected biochemical relationships. Clustering analysis of variation from whole-genome sequencing of gene expression (RNA-Seq) and histone H3 lysine 4 trimethylation (H3K4me3) events (ChIP-Seq) revealed the underlying structure of the transcriptional and epigenomic landscape within hippocampal postmortem brain tissue of drug abusers and control cases. Distinct sets of interrelated networks for cocaine and alcohol abuse were determined for each abusive substance. The network approach identified subsets of functionally related genes that are regulated in agreement with H3K4me3 changes, suggesting cause and effect relationships between this epigenetic mark and gene expression. Gene expression networks consisted of recognized substrates for addiction, such as the dopamine- and cAMP-regulated neuronal phosphoprotein PPP1R1B/DARPP-32 and the vesicular glutamate transporter SLC17A7/VGLUT1 as well as potentially novel molecular targets for substance abuse. Through a systems biology based approach our results illustrate the utility of integrating epigenetic and transcript expression to establish relevant biological networks in the human brain for addiction. Future work with laboratory models may clarify the functional relevance of these gene networks for cocaine and alcohol, and provide a framework for the development of medications for the treatment of addiction.

Biostatistical approaches for the reconstruction of gene co-expression networks based on transcriptomic data

Techniques in molecular biology have permitted the gathering of an extremely large amount of information relating organisms and their genes. The current challenge is assigning a putative function to thousands of genes that have been detected in different organisms. One of the most informative types of genomic data to achieve a better knowledge of protein function is gene expression data. Based on gene expression data and assuming that genes involved in the same function should have a similar or correlated expression pattern, a function can be attributed to those genes with unknown functions when they appear to be linked in a gene co-expression network (GCN). Several tools for the construction of GCNs have been proposed and applied to plant gene expression data. Here, we review recent methodologies used for plant gene expression data and compare the results, advantages and disadvantages in order to help researchers in their choice of a method for the construction of GCNs.

A contribution to the study of plant development evolution based on gene co-expression networks

Phototrophic eukaryotes are among the most successful organisms on Earth due to their unparalleled efficiency at capturing light energy and fixing carbon dioxide to produce organic molecules. A conserved and efficient network of light-dependent regulatory modules could be at the bases of this success. This regulatory system conferred early advantages to phototrophic eukaryotes that allowed for specialization, complex developmental processes and modern plant characteristics. We have studied light-dependent gene regulatory modules from algae to plants employing integrative-omics approaches based on gene co-expression networks. Our study reveals some remarkably conserved ways in which eukaryotic phototrophs deal with day length and light signaling. Here we describe how a family of Arabidopsis transcription factors involved in photoperiod response has evolved from a single algal gene according to the innovation, amplification and divergence theory of gene evolution by duplication. These modifications of the gene co-expression networks from the ancient unicellular green algae Chlamydomonas reinhardtii to the modern brassica Arabidopsis thaliana may hint on the evolution and specialization of plants and other organisms.

A Novel Network Model for Molecular Prognosis

Network-based genome-wide association studies (NWAS) utilize the molecular interactions between genes and functional pathways in biomarker identification. This study presents a novel network-based methodology for identifying prognostic gene signatures to predict cancer recurrence. The methodology contains the following steps: 1) Constructing genome-wide coexpression networks for different disease states (metastatic vs. non-metastatic). Prediction logic is used to induct valid implication relations between each pair of gene expression profiles in terms of formal logic rules. 2) Identifying differential components associated with specific disease states from the genome-wide coexpression networks. 3) Dissecting network modules that are tightly connected with major disease signal hallmarks from the disease specific differential components. 4) Identifying most significant genes/probes associated with clinical outcome from the pathway connected network modules. Using this methodology, a 14-gene prognostic signature was identified for accurate patient stratification in early stage lung cancer.