An atlas of tissue-specific conserved coexpression for functional annotation and disease gene prediction

A comprehensive transcriptional reference for severity and progression in spinal cord injury reveals novel translational biomarker genes

Spinal cord injury (SCI) is a devastating condition that leads to motor, sensory, and autonomic dysfunction. Current therapeutic options remain limited, emphasizing the need for a comprehensive understanding of the underlying SCI-associated molecular mechanisms. This study characterized distinct SCI phases and severities at the gene and functional levels, focusing on biomarker gene identification. Our approach involved a systematic review, individual transcriptomic analysis, gene meta-analysis, and functional characterization. We compiled a total of fourteen studies with 273 samples, leading to the identification of severity- and phase-specific biomarker genes that allow the precise classification of transcriptomic profiles. We investigated the potential transferability of severity-specific biomarkers and identified a twelve-gene signature that predicted injury prognosis from human blood samples. We also report the development of MetaSCI-app - an interactive web application designed for researchers - that allows the exploration and visualization of all generated results ( https://metasci-cbl.shinyapps.io/metaSCI ). Overall, we present a transcriptomic reference and provide a comprehensive framework for assessing SCI considering severity and time perspectives, all integrated into a user-friendly tool.

Gene communities in co-expression networks across different tissues

With the recent availability of tissue-specific gene expression data, e.g., provided by the GTEx Consortium, there is interest in comparing gene co-expression patterns across tissues. One promising approach to this problem is to use a multilayer network analysis framework and perform multilayer community detection. Communities in gene co-expression networks reveal groups of genes similarly expressed across individuals, potentially involved in related biological processes responding to specific environmental stimuli or sharing common regulatory variations. We construct a multilayer network in which each of the four layers is an exocrine gland tissue-specific gene co-expression network. We develop methods for multilayer community detection with correlation matrix input and an appropriate null model. Our correlation matrix input method identifies five groups of genes that are similarly co-expressed in multiple tissues (a community that spans multiple layers, which we call a generalist community) and two groups of genes that are co-expressed in just one tissue (a community that lies primarily within just one layer, which we call a specialist community). We further found gene co-expression communities where the genes physically cluster across the genome significantly more than expected by chance (on chromosomes 1 and 11). This clustering hints at underlying regulatory elements determining similar expression patterns across individuals and cell types. We suggest that KRTAP3-1, KRTAP3-3, and KRTAP3-5 share regulatory elements in skin and pancreas. Furthermore, we find that CELA3A and CELA3B share associated expression quantitative trait loci in the pancreas. The results indicate that our multilayer community detection method for correlation matrix input extracts biologically interesting communities of genes.

Gene communities in co-expression networks across different tissues

With the recent availability of tissue-specific gene expression data, e.g., provided by the GTEx Consortium, there is interest in comparing gene co-expression patterns across tissues. One promising approach to this problem is to use a multilayer network analysis framework and perform multilayer community detection. Communities in gene co-expression networks reveal groups of genes similarly expressed across individuals, potentially involved in related biological processes responding to specific environmental stimuli or sharing common regulatory variations. We construct a multilayer network in which each of the four layers is an exocrine gland tissue-specific gene co-expression network. We develop methods for multilayer community detection with correlation matrix input and an appropriate null model. Our correlation matrix input method identifies five groups of genes that are similarly co-expressed in multiple tissues (a community that spans multiple layers, which we call a generalist community) and two groups of genes that are co-expressed in just one tissue (a community that lies primarily within just one layer, which we call a specialist community). We further found gene co-expression communities where the genes physically cluster across the genome significantly more than expected by chance (on chromosomes 1 and 11). This clustering hints at underlying regulatory elements determining similar expression patterns across individuals and cell types. We suggest that KRTAP3-1, KRTAP3-3, and KRTAP3-5 share regulatory elements in skin and pancreas. Furthermore, we find that CELA3A and CELA3B share associated expression quantitative trait loci in the pancreas. The results indicate that our multilayer community detection method for correlation matrix input extracts biologically interesting communities of genes.

RNA-RNA competitive interactions: a molecular civil war ruling cell physiology and diseases

The idea that proteins are the main determining factors in the functioning of cells and organisms, and their dysfunctions are the first cause of pathologies, has been predominant in biology and biomedicine until recently. This protein-centered view was too simplistic and failed to explain the physiological and pathological complexity of the cell. About 80% of the human genome is dynamically and pervasively transcribed, mostly as non-protein-coding RNAs (ncRNAs), which competitively interact with each other and with coding RNAs generating a complex RNA network regulating RNA processing, stability, and translation and, accordingly, fine-tuning the gene expression of the cells. Qualitative and quantitative dysregulations of RNA-RNA interaction networks are strongly involved in the onset and progression of many pathologies, including cancers and degenerative diseases. This review will summarize the RNA species involved in the competitive endogenous RNA network, their mechanisms of action, and involvement in pathological phenotypes. Moreover, it will give an overview of the most advanced experimental and computational methods to dissect and rebuild RNA networks.

MOSTWAS: Multi-Omic Strategies for Transcriptome-Wide Association Studies

Traditional predictive models for transcriptome-wide association studies (TWAS) consider only single nucleotide polymorphisms (SNPs) local to genes of interest and perform parameter shrinkage with a regularization process. These approaches ignore the effect of distal-SNPs or other molecular effects underlying the SNP-gene association. Here, we outline multi-omics strategies for transcriptome imputation from germline genetics to allow more powerful testing of gene-trait associations by prioritizing distal-SNPs to the gene of interest. In one extension, we identify mediating biomarkers (CpG sites, microRNAs, and transcription factors) highly associated with gene expression and train predictive models for these mediators using their local SNPs. Imputed values for mediators are then incorporated into the final predictive model of gene expression, along with local SNPs. In the second extension, we assess distal-eQTLs (SNPs associated with genes not in a local window around it) for their mediation effect through mediating biomarkers local to these distal-eSNPs. Distal-eSNPs with large indirect mediation effects are then included in the transcriptomic prediction model with the local SNPs around the gene of interest. Using simulations and real data from ROS/MAP brain tissue and TCGA breast tumors, we show considerable gains of percent variance explained (1-2% additive increase) of gene expression and TWAS power to detect gene-trait associations. This integrative approach to transcriptome-wide imputation and association studies aids in identifying the complex interactions underlying genetic regulation within a tissue and important risk genes for various traits and disorders.

Enhanced Understanding of Molecular Interactions and Function Underlying Pain Processes Through Networks of Transcript Isoforms, Genes, and Gene Families

Introduction

Molecular networks based on the abundance of mRNA at the gene level and pathway networks that relate families or groups of paralog genes have supported the understanding of interactions between molecules. However, multiple molecular mechanisms underlying health and behavior, such as pain signal processing, are modulated by the abundances of the transcript isoforms that originate from alternative splicing, in addition to gene abundances. Alternative splice variants of growth factors, ion channels, and G-protein-coupled receptors can code for proteoforms that can have different effects on pain and nociception. Therefore, networks inferred using abundance from more agglomerative molecular units (eg, gene family, or gene) have limitations in capturing interactions at a more granular level (eg, gene, or transcript isoform, respectively) do not account for changes in the abundance at the transcript isoform level.

Objective

The objective of this study was to evaluate the relative benefits of network inference using abundance patterns at various aggregate levels.

Methods

Sparse networks were inferred using Gaussian Markov random fields and a novel aggregation criterion was used to aggregate network edges. The relative advantages of network aggregation were evaluated on two molecular systems that have different dimensions and connectivity, circadian rhythm and Toll-like receptor pathways, using RNA-sequencing data from mice representing two pain level groups, opioid-induced hyperalgesia and control, and two central nervous system regions, the nucleus accumbens and the trigeminal ganglia.

Results

The inferred networks were benchmarked against the Kyoto Encyclopedia of Genes and Genomes reference pathways using multiple criteria. Networks inferred using more granular information performed better than networks inferred using more aggregate information. The advantage of granular inference varied with the pathway and data set used.

Discussion

The differences in inferred network structure between data sets highlight the differences in OIH effect between central nervous system regions. Our findings suggest that inference of networks using alternative splicing variants can offer complementary insights into the relationship between genes and gene paralog groups.

MyoMiner: explore gene co-expression in normal and pathological muscle

BACKGROUND

High-throughput transcriptomics measures mRNA levels for thousands of genes in a biological sample. Most gene expression studies aim to identify genes that are differentially expressed between different biological conditions, such as between healthy and diseased states. However, these data can also be used to identify genes that are co-expressed within a biological condition. Gene co-expression is used in a guilt-by-association approach to prioritize candidate genes that could be involved in disease, and to gain insights into the functions of genes, protein relations, and signaling pathways. Most existing gene co-expression databases are generic, amalgamating data for a given organism regardless of tissue-type.

METHODS

To study muscle-specific gene co-expression in both normal and pathological states, publicly available gene expression data were acquired for 2376 mouse and 2228 human striated muscle samples, and separated into 142 categories based on species (human or mouse), tissue origin, age, gender, anatomic part, and experimental condition. Co-expression values were calculated for each category to create the MyoMiner database.

RESULTS

Within each category, users can select a gene of interest, and the MyoMiner web interface will return all correlated genes. For each co-expressed gene pair, adjusted p-value and confidence intervals are provided as measures of expression correlation strength. A standardized expression-level scatterplot is available for every gene pair r-value. MyoMiner has two extra functions: (a) a network interface for creating a 2-shell correlation network, based either on the most highly correlated genes or from a list of genes provided by the user with the option to include linked genes from the database and (b) a comparison tool from which the users can test whether any two correlation coefficients from different conditions are significantly different.

CONCLUSIONS

These co-expression analyses will help investigators to delineate the tissue-, cell-, and pathology-specific elements of muscle protein interactions, cell signaling and gene regulation. Changes in co-expression between pathologic and healthy tissue may suggest new disease mechanisms and help define novel therapeutic targets. Thus, MyoMiner is a powerful muscle-specific database for the discovery of genes that are associated with related functions based on their co-expression. MyoMiner is freely available at https://www.sys-myo.com/myominer.

Network-Based Methods and Other Approaches for Predicting lncRNA Functions and Disease Associations

The discovery that a considerable portion of eukaryotic genomes is transcribed and gives rise to long noncoding RNAs (lncRNAs) provides an important new perspective on the transcriptome and raises questions about the centrality of these lncRNAs in gene-regulatory processes and diseases. The rapidly increasing number of mechanistically investigated lncRNAs has provided evidence for distinct functional classes, such as enhancer-like lncRNAs, which modulate gene expression via chromatin looping, and noncoding competing endogenous RNAs (ceRNAs), which act as microRNA decoys. Despite great progress in the last years, the majority of lncRNAs are functionally uncharacterized and their implication for disease biogenesis and progression is unknown. Here, we summarize recent developments in lncRNA function prediction in general and lncRNA-disease associations in particular, with emphasis on in silico methods based on network analysis and on ceRNA function prediction. We believe that such computational techniques provide a valuable aid to prioritize functional lncRNAs or disease-relevant lncRNAs for targeted, experimental follow-up studies.

Conserved DNA Methylation Signatures in Early Maternal Separation and in Twins Discordant for CO2 Sensitivity

Respiratory and emotional responses to blood-acidifying inhalation of CO₂ are markers of some human anxiety disorders, and can be enhanced by repeatedly cross-fostering (RCF) mouse pups from their biological mother to unrelated lactating females. Yet, these dynamics remain poorly understood. We show RCF-associated intergenerational transmission of CO₂ sensitivity in normally-reared mice descending from RCF-exposed females, and describe the accompanying alterations in brain DNA methylation patterns. These epigenetic signatures were compared to DNA methylation profiles of monozygotic twins discordant for emotional reactivity to a CO₂ challenge. Altered methylation was consistently associated with repeated elements and transcriptional regulatory regions among RCF-exposed animals, their normally-reared offspring, and humans with CO₂ hypersensitivity. In both species, regions bearing differential methylation were associated with neurodevelopment, circulation, and response to pH acidification processes, and notably included the ASIC2 gene. Our data show that CO₂ hypersensitivity is associated with specific methylation clusters and genes that subserve chemoreception and anxiety. The methylation status of genes implicated in acid-sensing functions can inform etiological and therapeutic research in this field.

Co-expression networks reveal the tissue-specific regulation of transcription and splicing

Gene co-expression networks capture biologically important patterns in gene expression data, enabling functional analyses of genes, discovery of biomarkers, and interpretation of genetic variants. Most network analyses to date have been limited to assessing correlation between total gene expression levels in a single tissue or small sets of tissues. Here, we built networks that additionally capture the regulation of relative isoform abundance and splicing, along with tissue-specific connections unique to each of a diverse set of tissues. We used the Genotype-Tissue Expression (GTEx) project v6 RNA sequencing data across 50 tissues and 449 individuals. First, we developed a framework called Transcriptome-Wide Networks (TWNs) for combining total expression and relative isoform levels into a single sparse network, capturing the interplay between the regulation of splicing and transcription. We built TWNs for 16 tissues and found that hubs in these networks were strongly enriched for splicing and RNA binding genes, demonstrating their utility in unraveling regulation of splicing in the human transcriptome. Next, we used a Bayesian biclustering model that identifies network edges unique to a single tissue to reconstruct Tissue-Specific Networks (TSNs) for 26 distinct tissues and 10 groups of related tissues. Finally, we found genetic variants associated with pairs of adjacent nodes in our networks, supporting the estimated network structures and identifying 20 genetic variants with distant regulatory impact on transcription and splicing. Our networks provide an improved understanding of the complex relationships of the human transcriptome across tissues.

Gene Network Construction from Microarray Data Identifies a Key Network Module and Several Candidate Hub Genes in Age-Associated Spatial Learning Impairment

As humans age many suffer from a decrease in normal brain functions including spatial learning impairments. This study aimed to better understand the molecular mechanisms in age-associated spatial learning impairment (ASLI). We used a mathematical modeling approach implemented in Weighted Gene Co-expression Network Analysis (WGCNA) to create and compare gene network models of young (learning unimpaired) and aged (predominantly learning impaired) brains from a set of exploratory datasets in rats in the context of ASLI. The major goal was to overcome some of the limitations previously observed in the traditional meta- and pathway analysis using these data, and identify novel ASLI related genes and their networks based on co-expression relationship of genes. This analysis identified a set of network modules in the young, each of which is highly enriched with genes functioning in broad but distinct GO functional categories or biological pathways. Interestingly, the analysis pointed to a single module that was highly enriched with genes functioning in “learning and memory” related functions and pathways. Subsequent differential network analysis of this “learning and memory” module in the aged (predominantly learning impaired) rats compared to the young learning unimpaired rats allowed us to identify a set of novel ASLI candidate hub genes. Some of these genes show significant repeatability in networks generated from independent young and aged validation datasets. These hub genes are highly co-expressed with other genes in the network, which not only show differential expression but also differential co-expression and differential connectivity across age and learning impairment. The known function of these hub genes indicate that they play key roles in critical pathways, including kinase and phosphatase signaling, in functions related to various ion channels, and in maintaining neuronal integrity relating to synaptic plasticity and memory formation. Taken together, they provide a new insight and generate new hypotheses into the molecular mechanisms responsible for age associated learning impairment, including spatial learning.

Synaptic Interactome Mining Reveals p140Cap as a New Hub for PSD Proteins Involved in Psychiatric and Neurological Disorders

Altered synaptic function has been associated with neurological and psychiatric conditions including intellectual disability, schizophrenia and autism spectrum disorder (ASD). Amongst the recently discovered synaptic proteins is p140Cap, an adaptor that localizes at dendritic spines and regulates their maturation and physiology. We recently showed that p140Cap knockout mice have cognitive deficits, impaired long-term potentiation (LTP) and long-term depression (LTD), and immature, filopodia-like dendritic spines. Only a few p140Cap interacting proteins have been identified in the brain and the molecular complexes and pathways underlying p140Cap synaptic function are largely unknown. Here, we isolated and characterized the p140Cap synaptic interactome by co-immunoprecipitation from crude mouse synaptosomes, followed by mass spectrometry-based proteomics. We identified 351 p140Cap interactors and found that they cluster to sub complexes mostly located in the postsynaptic density (PSD). p140Cap interactors converge on key synaptic processes, including transmission across chemical synapses, actin cytoskeleton remodeling and cell-cell junction organization. Gene co-expression data further support convergent functions: the p140Cap interactors are tightly co-expressed with each other and with p140Cap. Importantly, the p140Cap interactome and its co-expression network show strong enrichment in genes associated with schizophrenia, autism, bipolar disorder, intellectual disability and epilepsy, supporting synaptic dysfunction as a shared biological feature in brain diseases. Overall, our data provide novel insights into the molecular organization of the synapse and indicate that p140Cap acts as a hub for postsynaptic complexes relevant to psychiatric and neurological disorders.

In silico prediction of lncRNA function using tissue specific and evolutionary conserved expression

BACKGROUND

In recent years long non coding RNAs (lncRNAs) have been the subject of increasing interest. Thanks to many recent functional studies, the existence of a large class of lncRNAs with potential regulatory functions is now widely accepted. Although an increasing number of lncRNAs is being characterized and shown to be involved in many biological processes, the functions of the vast majority lncRNA genes is still unknown. Therefore computational methods able to take advantage of the increasing amount of publicly available data to predict lncRNA functions could be very useful.

RESULTS

Since coding genes are much better annotated than lncRNAs, we attempted to project known functional information regarding proteins onto non coding genes using the guilt by association principle: if a gene shows an expression profile that correlates with those of a set of coding genes involved in a given function, that gene is probably involved in the same function. We computed gene coexpression for 30 human tissues and 9 vertebrates and mined the resulting networks with a methodology inspired by the rank product algorithm used to identify differentially expressed genes. Using different types of reference data we can predict putative new annotations for thousands of lncRNAs and proteins, ranging from cellular localization to relevance for disease and cancer.

CONCLUSIONS

New function of coding genes and lncRNA can be profitably predicted using tissue specific coexpression, as well as expression of orthologous genes in different species. The data are available for download and through a user-friendly web interface at www.funcpred.com .

Characterization of candidate genes in inflammatory bowel disease-associated risk loci

GWAS have linked SNPs to risk of inflammatory bowel disease (IBD), but a systematic characterization of disease-associated genes has been lacking. Prior studies utilized microarrays that did not capture many genes encoded within risk loci or defined expression quantitative trait loci (eQTLs) using peripheral blood, which is not the target tissue in IBD. To address these gaps, we sought to characterize the expression of IBD-associated risk genes in disease-relevant tissues and in the setting of active IBD. Terminal ileal (TI) and colonic mucosal tissues were obtained from patients with Crohn's disease or ulcerative colitis and from healthy controls. We developed a NanoString code set to profile 678 genes within IBD risk loci. A subset of patients and controls were genotyped for IBD-associated risk SNPs. Analyses included differential expression and variance analysis, weighted gene coexpression network analysis, and eQTL analysis. We identified 116 genes that discriminate between healthy TI and colon samples and uncovered patterns in variance of gene expression that highlight heterogeneity of disease. We identified 107 coexpressed gene pairs for which transcriptional regulation is either conserved or reversed in an inflammation-independent or -dependent manner. We demonstrate that on average approximately 60% of disease-associated genes are differentially expressed in inflamed tissue. Last, we identified eQTLs with either genotype-only effects on expression or an interaction effect between genotype and inflammation. Our data reinforce tissue specificity of expression in disease-associated candidate genes, highlight genes and gene pairs that are regulated in disease-relevant tissue and inflammation, and provide a foundation to advance the understanding of IBD pathogenesis.

Genome-Wide Functional Annotation of Human Protein-Coding Splice Variants Using Multiple Instance Learning

The vast majority of human multiexon genes undergo alternative splicing and produce a variety of splice variant transcripts and proteins, which can perform different functions. These protein-coding splice variants (PCSVs) greatly increase the functional diversity of proteins. Most functional annotation algorithms have been developed at the gene level; the lack of isoform-level gold standards is an important intellectual limitation for currently available machine learning algorithms. The accumulation of a large amount of RNA-seq data in the public domain greatly increases our ability to examine the functional annotation of genes at isoform level. In the present study, we used a multiple instance learning (MIL)-based approach for predicting the function of PCSVs. We used transcript-level expression values and gene-level functional associations from the Gene Ontology database. A support vector machine (SVM)-based 5-fold cross-validation technique was applied. Comparatively, genes with multiple PCSVs performed better than single PCSV genes, and performance also improved when more examples were available to train the models. We demonstrated our predictions using literature evidence of ADAM15, LMNA/C, and DMXL2 genes. All predictions have been implemented in a web resource called "IsoFunc", which is freely available for the global scientific community through http://guanlab.ccmb.med.umich.edu/isofunc .

Immuno-Navigator, a batch-corrected coexpression database, reveals cell type-specific gene networks in the immune system

High-throughput gene expression data are one of the primary resources for exploring complex intracellular dynamics in modern biology. The integration of large amounts of public data may allow us to examine general dynamical relationships between regulators and target genes. However, obstacles for such analyses are study-specific biases or batch effects in the original data. Here we present Immuno-Navigator, a batch-corrected gene expression and coexpression database for 24 cell types of the mouse immune system. We systematically removed batch effects from the underlying gene expression data and showed that this removal considerably improved the consistency between inferred correlations and prior knowledge. The data revealed widespread cell type-specific correlation of expression. Integrated analysis tools allow users to use this correlation of expression for the generation of hypotheses about biological networks and candidate regulators in specific cell types. We show several applications of Immuno-Navigator as examples. In one application we successfully predicted known regulators of importance in naturally occurring Treg cells from their expression correlation with a set of Treg-specific genes. For one high-scoring gene, integrin β8 (Itgb8), we confirmed an association between Itgb8 expression in forkhead box P3 (Foxp3)-positive T cells and Treg-specific epigenetic remodeling. Our results also suggest that the regulation of Treg-specific genes within Treg cells is relatively independent of Foxp3 expression, supporting recent results pointing to a Foxp3-independent component in the development of Treg cells.

Posttranscriptional Regulation of PTEN by Competing Endogenous RNAs

PTEN expression can be dysregulated in cancers via multiple mechanisms including genomic loss, epigenetic silencing, transcriptional repression, and posttranscriptional regulation by microRNAs. MicroRNAs are short, noncoding RNAs that regulate gene expression by binding to recognition sites on target transcripts. Recent studies have demonstrated that the competition for shared microRNAs between both protein-coding and noncoding transcripts represents an additional facet of gene regulation. Here, we describe in detail an integrated computational and experimental approach to identify and validate these competing endogenous RNA (ceRNA) interactions.

It is time to take timing seriously in clinical genetics

Observations made by molecular techniques on the genome along the individuals' lifetime indicate that the genome in somatic cells displays changes at molecular, cellular, and organismal levels. Timing of genetic events leading to somatic mosaicism and gene expression dynamism results in a highly important variable for comprehending the role of genetics in health and disease. Consideration of time in clinical genetics should be enthusiastically invested into research strategy, interpretation of the results, diagnostic routine, and particularly in ethical discussions.

Transcriptomic analysis of mononuclear phagocyte differentiation and activation

Monocytes and macrophages differentiate from progenitor cells under the influence of colony-stimulating factors. Genome-scale data have enabled the identification of the sets of genes that are associated with specific functions and the mechanisms by which thousands of genes are regulated in response to pathogen challenge. In large datasets, it is possible to identify large sets of genes that are coregulated with the transcription factors that regulate them. They include macrophage-specific genes, interferon-responsive genes, early inflammatory genes, and those associated with endocytosis. Such analyses can also extract macrophage-associated signatures from large cancer tissue datasets. However, cluster analysis provides no support for a signature that distinguishes macrophages from antigen-presenting dendritic cells, nor the classification of macrophage activation states as classical versus alternative, or M1 versus M2. Although there has been a focus on a small subset of lineage-enriched transcription factors, such as PU.1, more than half of the transcription factors in the genome can be expressed in macrophage lineage cells under some state of activation, and they interact in a complex network. The network architecture is conserved across species, but many of the target genes evolve rapidly and differ between mouse and human. The data and publication deluge related to macrophage biology require the development of new analytical tools and ways of presenting information in an accessible form.

UNCLES: method for the identification of genes differentially consistently co-expressed in a specific subset of datasets

Background

Collective analysis of the increasingly emerging gene expression datasets are required. The recently proposed binarisation of consensus partition matrices (Bi-CoPaM) method can combine clustering results from multiple datasets to identify the subsets of genes which are consistently co-expressed in all of the provided datasets in a tuneable manner. However, results validation and parameter setting are issues that complicate the design of such methods. Moreover, although it is a common practice to test methods by application to synthetic datasets, the mathematical models used to synthesise such datasets are usually based on approximations which may not always be sufficiently representative of real datasets.

Results

Here, we propose an unsupervised method for the unification of clustering results from multiple datasets using external specifications (UNCLES). This method has the ability to identify the subsets of genes consistently co-expressed in a subset of datasets while being poorly co-expressed in another subset of datasets, and to identify the subsets of genes consistently co-expressed in all given datasets. We also propose the M-N scatter plots validation technique and adopt it to set the parameters of UNCLES, such as the number of clusters, automatically. Additionally, we propose an approach for the synthesis of gene expression datasets using real data profiles in a way which combines the ground-truth-knowledge of synthetic data and the realistic expression values of real data, and therefore overcomes the problem of faithfulness of synthetic expression data modelling. By application to those datasets, we validate UNCLES while comparing it with other conventional clustering methods, and of particular relevance, biclustering methods. We further validate UNCLES by application to a set of 14 real genome-wide yeast datasets as it produces focused clusters that conform well to known biological facts. Furthermore, in-silico-based hypotheses regarding the function of a few previously unknown genes in those focused clusters are drawn.

Conclusions

The UNCLES method, the M-N scatter plots technique, and the expression data synthesis approach will have wide application for the comprehensive analysis of genomic and other sources of multiple complex biological datasets. Moreover, the derived in-silico-based biological hypotheses represent subjects for future functional studies.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-015-0614-0) contains supplementary material, which is available to authorized users.

Sharing and Specificity of Co-expression Networks across 35 Human Tissues

To understand the regulation of tissue-specific gene expression, the GTEx Consortium generated RNA-seq expression data for more than thirty distinct human tissues. This data provides an opportunity for deriving shared and tissue specific gene regulatory networks on the basis of co-expression between genes. However, a small number of samples are available for a majority of the tissues, and therefore statistical inference of networks in this setting is highly underpowered. To address this problem, we infer tissue-specific gene co-expression networks for 35 tissues in the GTEx dataset using a novel algorithm, GNAT, that uses a hierarchy of tissues to share data between related tissues. We show that this transfer learning approach increases the accuracy with which networks are learned. Analysis of these networks reveals that tissue-specific transcription factors are hubs that preferentially connect to genes with tissue specific functions. Additionally, we observe that genes with tissue-specific functions lie at the peripheries of our networks. We identify numerous modules enriched for Gene Ontology functions, and show that modules conserved across tissues are especially likely to have functions common to all tissues, while modules that are upregulated in a particular tissue are often instrumental to tissue-specific function. Finally, we provide a web tool, available at mostafavilab.stat.ubc.ca/GNAT, which allows exploration of gene function and regulation in a tissue-specific manner.

Systems genetics identifies Sestrin 3 as a regulator of a proconvulsant gene network in human epileptic hippocampus

Gene-regulatory network analysis is a powerful approach to elucidate the molecular processes and pathways underlying complex disease. Here we employ systems genetics approaches to characterize the genetic regulation of pathophysiological pathways in human temporal lobe epilepsy (TLE). Using surgically acquired hippocampi from 129 TLE patients, we identify a gene-regulatory network genetically associated with epilepsy that contains a specialized, highly expressed transcriptional module encoding proconvulsive cytokines and Toll-like receptor signalling genes. RNA sequencing analysis in a mouse model of TLE using 100 epileptic and 100 control hippocampi shows the proconvulsive module is preserved across-species, specific to the epileptic hippocampus and upregulated in chronic epilepsy. In the TLE patients, we map the trans-acting genetic control of this proconvulsive module to Sestrin 3 (SESN3), and demonstrate that SESN3 positively regulates the module in macrophages, microglia and neurons. Morpholino-mediated Sesn3 knockdown in zebrafish confirms the regulation of the transcriptional module, and attenuates chemically induced behavioural seizures in vivo.

An extensive analysis of disease-gene associations using network integration and fast kernel-based gene prioritization methods

OBJECTIVE

In the context of "network medicine", gene prioritization methods represent one of the main tools to discover candidate disease genes by exploiting the large amount of data covering different types of functional relationships between genes. Several works proposed to integrate multiple sources of data to improve disease gene prioritization, but to our knowledge no systematic studies focused on the quantitative evaluation of the impact of network integration on gene prioritization. In this paper, we aim at providing an extensive analysis of gene-disease associations not limited to genetic disorders, and a systematic comparison of different network integration methods for gene prioritization.

MATERIALS AND METHODS

We collected nine different functional networks representing different functional relationships between genes, and we combined them through both unweighted and weighted network integration methods. We then prioritized genes with respect to each of the considered 708 medical subject headings (MeSH) diseases by applying classical guilt-by-association, random walk and random walk with restart algorithms, and the recently proposed kernelized score functions.

RESULTS

The results obtained with classical random walk algorithms and the best single network achieved an average area under the curve (AUC) across the 708 MeSH diseases of about 0.82, while kernelized score functions and network integration boosted the average AUC to about 0.89. Weighted integration, by exploiting the different "informativeness" embedded in different functional networks, outperforms unweighted integration at 0.01 significance level, according to the Wilcoxon signed rank sum test. For each MeSH disease we provide the top-ranked unannotated candidate genes, available for further bio-medical investigation.

CONCLUSIONS

Network integration is necessary to boost the performances of gene prioritization methods. Moreover the methods based on kernelized score functions can further enhance disease gene ranking results, by adopting both local and global learning strategies, able to exploit the overall topology of the network.

Drug repositioning for orphan genetic diseases through Conserved Anticoexpressed Gene Clusters (CAGCs)

Background

The development of new therapies for orphan genetic diseases represents an extremely important medical and social challenge. Drug repositioning, i.e. finding new indications for approved drugs, could be one of the most cost- and time-effective strategies to cope with this problem, at least in a subset of cases. Therefore, many computational approaches based on the analysis of high throughput gene expression data have so far been proposed to reposition available drugs. However, most of these methods require gene expression profiles directly relevant to the pathologic conditions under study, such as those obtained from patient cells and/or from suitable experimental models. In this work we have developed a new approach for drug repositioning, based on identifying known drug targets showing conserved anti-correlated expression profiles with human disease genes, which is completely independent from the availability of ‘ad hoc’ gene expression data-sets.

Results

By analyzing available data, we provide evidence that the genes displaying conserved anti-correlation with drug targets are antagonistically modulated in their expression by treatment with the relevant drugs. We then identified clusters of genes associated to similar phenotypes and showing conserved anticorrelation with drug targets. On this basis, we generated a list of potential candidate drug-disease associations. Importantly, we show that some of the proposed associations are already supported by independent experimental evidence.

Conclusions

Our results support the hypothesis that the identification of gene clusters showing conserved anticorrelation with drug targets can be an effective method for drug repositioning and provide a wide list of new potential drug-disease associations for experimental validation.

An expression atlas of human primary cells: inference of gene function from coexpression networks

BACKGROUND

The specialisation of mammalian cells in time and space requires genes associated with specific pathways and functions to be co-ordinately expressed. Here we have combined a large number of publically available microarray datasets derived from human primary cells and analysed large correlation graphs of these data.

RESULTS

Using the network analysis tool BioLayout Express3D we identify robust co-associations of genes expressed in a wide variety of cell lineages. We discuss the biological significance of a number of these associations, in particular the coexpression of key transcription factors with the genes that they are likely to control.

CONCLUSIONS

We consider the regulation of genes in human primary cells and specifically in the human mononuclear phagocyte system. Of particular note is the fact that these data do not support the identity of putative markers of antigen-presenting dendritic cells, nor classification of M1 and M2 activation states, a current subject of debate within immunological field. We have provided this data resource on the BioGPS web site (http://biogps.org/dataset/2429/primary-cell-atlas/) and on macrophages.com (http://www.macrophages.com/hu-cell-atlas).

The RNA binding protein ESRP1 fine-tunes the expression of pluripotency-related factors in mouse embryonic stem cells

In pluripotent stem cells, there is increasing evidence for crosstalk between post-transcriptional and transcriptional networks, offering multifold steps at which pluripotency can be controlled. In addition to well-studied transcription factors, chromatin modifiers and miRNAs, RNA-binding proteins are emerging as fundamental players in pluripotency regulation. Here, we report a new role for the RNA-binding protein ESRP1 in the control of pluripotency. Knockdown of Esrp1 in mouse embryonic stem cells induces, other than the well-documented epithelial to mesenchymal-like state, also an increase in expression of the core transcription factors Oct4, Nanog and Sox2, thereby enhancing self-renewal of these cells. Esrp1-depleted embryonic stem cells displayed impaired early differentiation in vitro and formed larger teratomas in vivo when compared to control embryonic stem cells. We also show that ESRP1 binds to Oct4 and Sox2 mRNAs and decreases their polysomal loading. ESRP1 thus acts as a physiological regulator of the finely-tuned balance between self-renewal and commitment to a restricted developmental fate. Importantly, both mouse and human epithelial stem cells highly express ESRP1, pinpointing the importance of this RNA-binding protein in stem cell biology.

Concordance of gene expression in human protein complexes reveals tissue specificity and pathology

Disease-causing variants in human genes usually lead to phenotypes specific to only a few tissues. Here, we present a method for predicting tissue specificity based on quantitative deregulation of protein complexes. The underlying assumption is that the degree of coordinated expression among proteins in a complex within a given tissue may pinpoint tissues that will be affected by a mutation in the complex and coordinated expression may reveal the complex to be active in the tissue. We identified known disease genes and their protein complex partners in a high-quality human interactome. Each susceptibility gene's tissue involvement was ranked based on coordinated expression with its interaction partners in a non-disease global map of human tissue-specific expression. The approach demonstrated high overall area under the curve (0.78) and was very successfully benchmarked against a random model and an approach not using protein complexes. This was illustrated by correct tissue predictions for three case studies on leptin, insulin-like-growth-factor 2 and the inhibitor of NF-κB kinase subunit gamma that show high concordant expression in biologically relevant tissues. Our method identifies novel gene-phenotype associations in human diseases and predicts the tissues where associated phenotypic effects may arise.

Integrated transcriptional and competitive endogenous RNA networks are cross-regulated in permissive molecular environments

Competitive endogenous (ce)RNAs cross-regulate each other through sequestration of shared microRNAs and form complex regulatory networks based on their microRNA signature. However, the molecular requirements for ceRNA cross-regulation and the extent of ceRNA networks remain unknown. Here, we present a mathematical mass-action model to determine the optimal conditions for ceRNA activity in silico. This model was validated using phosphatase and tensin homolog (PTEN) and its ceRNA VAMP (vesicle-associated membrane protein)-associated protein A (VAPA) as paradigmatic examples. A computational assessment of the complexity of ceRNA networks revealed that transcription factor and ceRNA networks are intimately intertwined. Notably, we found that ceRNA networks are responsive to transcription factor up-regulation or their aberrant expression in cancer. Thus, given optimal molecular conditions, alterations of one ceRNA can have striking effects on integrated ceRNA and transcriptional networks.

Protein sets define disease states and predict in vivo effects of drug treatment

Gaining understanding of common complex diseases and their treatments are the main drivers for life sciences. As we show here, comprehensive protein set analyses offer new opportunities to decipher functional molecular networks of diseases and assess the efficacy and side-effects of treatments in vivo. Using mass spectrometry, we quantitatively detected several thousands of proteins and observed significant changes in protein pathways that were (dys-) regulated in diet-induced obesity mice. Analysis of the expression and post-translational modifications of proteins in various peripheral metabolic target tissues including adipose, heart, and liver tissue generated functional insights in the regulation of cell and tissue homeostasis during high-fat diet feeding and medication with two antidiabetic compounds. Protein set analyses singled out pathways for functional characterization, and indicated, for example, early-on potential cardiovascular complication of the diabetes drug rosiglitazone. In vivo protein set detection can provide new avenues for monitoring complex disease processes, and for evaluating preclinical drug candidates.

BMP-mediated functional cooperation between Dlx5;Dlx6 and Msx1;Msx2 during mammalian limb development

The Dlx and Msx homeodomain transcription factors play important roles in the control of limb development. The combined disruption of Msx1 and Msx2, as well as that of Dlx5 and Dlx6, lead to limb patterning defects with anomalies in digit number and shape. Msx1;Msx2 double mutants are characterized by the loss of derivatives of the anterior limb mesoderm which is not observed in either of the simple mutants. Dlx5;Dlx6 double mutants exhibit hindlimb ectrodactyly. While the morphogenetic action of Msx genes seems to involve the BMP molecules, the mode of action of Dlx genes still remains elusive. Here, examining the limb phenotypes of combined Dlx and Msx mutants we reveal a new Dlx-Msx regulatory loop directly involving BMPs. In Msx1;Dlx5;Dlx6 triple mutant mice (TKO), beside the expected ectrodactyly, we also observe the hallmark morphological anomalies of Msx1;Msx2 double mutants suggesting an epistatic role of Dlx5 and Dlx6 over Msx2. In Msx2;Dlx5;Dlx6 TKO mice we only observe an aggravation of the ectrodactyly defect without changes in the number of the individual components of the limb. Using a combination of qPCR, ChIP and bioinformatic analyses, we identify two Dlx/Msx regulatory pathways: 1) in the anterior limb mesoderm a non-cell autonomous Msx-Dlx regulatory loop involves BMP molecules through the AER and 2) in AER cells and, at later stages, in the limb mesoderm the regulation of Msx2 by Dlx5 and Dlx6 occurs also cell autonomously. These data bring new elements to decipher the complex AER-mesoderm dialogue that takes place during limb development and provide clues to understanding the etiology of congenital limb malformations.

Profiling, Bioinformatic, and Functional Data on the Developing Olfactory/GnRH System Reveal Cellular and Molecular Pathways Essential for This Process and Potentially Relevant for the Kallmann Syndrome

During embryonic development, immature neurons in the olfactory epithelium (OE) extend axons through the nasal mesenchyme, to contact projection neurons in the olfactory bulb. Axon navigation is accompanied by migration of the GnRH+ neurons, which enter the anterior forebrain and home in the septo-hypothalamic area. This process can be interrupted at various points and lead to the onset of the Kallmann syndrome (KS), a disorder characterized by anosmia and central hypogonadotropic hypogonadism. Several genes has been identified in human and mice that cause KS or a KS-like phenotype. In mice a set of transcription factors appears to be required for olfactory connectivity and GnRH neuron migration; thus we explored the transcriptional network underlying this developmental process by profiling the OE and the adjacent mesenchyme at three embryonic ages. We also profiled the OE from embryos null for Dlx5, a homeogene that causes a KS-like phenotype when deleted. We identified 20 interesting genes belonging to the following categories: (1) transmembrane adhesion/receptor, (2) axon-glia interaction, (3) scaffold/adapter for signaling, (4) synaptic proteins. We tested some of them in zebrafish embryos: the depletion of five (of six) Dlx5 targets affected axonal extension and targeting, while three (of three) affected GnRH neuron position and neurite organization. Thus, we confirmed the importance of cell-cell and cell-matrix interactions and identified new molecules needed for olfactory connection and GnRH neuron migration. Using available and newly generated data, we predicted/prioritized putative KS-disease genes, by building conserved co-expression networks with all known disease genes in human and mouse. The results show the overall validity of approaches based on high-throughput data and predictive bioinformatics to identify genes potentially relevant for the molecular pathogenesis of KS. A number of candidate will be discussed, that should be tested in future mutation screens.

Disease-gene discovery by integration of 3D gene expression and transcription factor binding affinities

MOTIVATION

The computational evaluation of candidate genes for hereditary disorders is a non-trivial task. Several excellent methods for disease-gene prediction have been developed in the past 2 decades, exploiting widely differing data sources to infer disease-relevant functional relationships between candidate genes and disorders. We have shown recently that spatially mapped, i.e. 3D, gene expression data from the mouse brain can be successfully used to prioritize candidate genes for human Mendelian disorders of the central nervous system.

RESULTS

We improved our previous work 2-fold: (i) we demonstrate that condition-independent transcription factor binding affinities of the candidate genes' promoters are relevant for disease-gene prediction and can be integrated with our previous approach to significantly enhance its predictive power; and (ii) we define a novel similarity measure-termed Relative Intensity Overlap-for both 3D gene expression patterns and binding affinity profiles that better exploits their disease-relevant information content. Finally, we present novel disease-gene predictions for eight loci associated with different syndromes of unknown molecular basis that are characterized by mental retardation.

Causal co-expression method with module analysis to screen drugs with specific target

The considerable increase of investment in research and development by the pharmaceutical industry over the past three decades has not added the number of approved new drugs. An important issue ignored by drug discovery practice is the multi-dimensional interaction network between drugs and their targets. Thus, it is essential to view drug actions through the lens of network biology. In the current study, based on the co-expression network of transcription factors and their downstream genes, we proposed a novel approach, called causal co-expression method with module analysis, to screen drugs with specific target and fewer side effects. We presented a causal co-expression method with module analysis and it could be used in analyzing the microarray data of different drug candidates. At first, the differential wiring value (DW) was calculated to find some causal transcription factors (TFs) by combining with differential expression genes in the regulated networks. After the discovery of the causal TFs, co-expression module analysis method was applied to mine molecular pharmacology pathways around these causal TFs at molecular level. We applied our methods to two drug candidates, Argyrin A and Bortezomib, both with anti-cancer activities. We first obtained some differentially expressed transcription factors of cells treated with Argyrin A or Bortezomib. Nearly all these transcription factors are associated with the tumor suppressor protein p27kip1. Furthermore, module analysis showed that Bortezomib inhibited tumor growth not specifically by cell cycle and cell proliferation pathway, but through many basic metabolic processes which result in cell toxicity. In contrast, Argyrin A had influence on cell cycle, and was involved in DNA damage repair at the same time, showing that Argyrin A was a more suitable drug for anti-cancer treatment. Our study revealed that the causal co-expression method with module analysis was effective and can be used as a tool to evaluate drug candidates.

Tissue-enriched expression profiles in Aedes aegypti identify hemocyte-specific transcriptome responses to infection

Hemocytes are integral components of mosquito immune mechanisms such as phagocytosis, melanization, and production of antimicrobial peptides. However, our understanding of hemocyte-specific molecular processes and their contribution to shaping the host immune response remains limited. To better understand the immunophysiological features distinctive of hemocytes, we conducted genome-wide analysis of hemocyte-enriched transcripts, and examined how tissue-enriched expression patterns change with the immune status of the host. Our microarray data indicate that the hemocyte-enriched trascriptome is dynamic and context-dependent. Analysis of transcripts enriched after bacterial challenge in circulating hemocytes with respect to carcass added a dimension to evaluating infection-responsive genes and immune-related gene families. We resolved patterns of transcriptional change unique to hemocytes from those that are likely shared by other immune responsive tissues, and identified clusters of genes preferentially induced in hemocytes, likely reflecting their involvement in cell type specific functions. In addition, the study revealed conserved hemocyte-enriched molecular repertoires, which might be implicated in core hemocyte function by cross-species meta-analysis of microarray expression data from Anopheles gambiae and Drosophila melanogaster.

Computational approaches to disease-gene prediction: rationale, classification and successes

The identification of genes involved in human hereditary diseases often requires the time-consuming and expensive examination of a great number of possible candidate genes, since genome-wide techniques such as linkage analysis and association studies frequently select many hundreds of 'positional' candidates. Even considering the positive impact of next-generation sequencing technologies, the prioritization of candidate genes may be an important step for disease-gene identification. In this paper we develop a basic classification scheme for computational approaches to disease-gene prediction and apply it to exhaustively review bioinformatics tools that have been developed for this purpose, focusing on conceptual aspects rather than technical detail and performance. Finally, we discuss some past successes obtained by computational approaches to illustrate their beneficial contribution to medical research.

Coding-independent regulation of the tumor suppressor PTEN by competing endogenous mRNAs

Here, we demonstrate that protein-coding RNA transcripts can crosstalk by competing for common microRNAs, with microRNA response elements as the foundation of this interaction. We have termed such RNA transcripts as competing endogenous RNAs (ceRNAs). We tested this hypothesis in the context of PTEN, a key tumor suppressor whose abundance determines critical outcomes in tumorigenesis. By a combined computational and experimental approach, we identified and validated endogenous protein-coding transcripts that regulate PTEN, antagonize PI3K/AKT signaling, and possess growth- and tumor-suppressive properties. Notably, we also show that these genes display concordant expression patterns with PTEN and copy number loss in cancers. Our study presents a road map for the prediction and validation of ceRNA activity and networks and thus imparts a trans-regulatory function to protein-coding mRNAs.

In vivo identification of tumor- suppressive PTEN ceRNAs in an oncogenic BRAF-induced mouse model of melanoma

We recently proposed that competitive endogenous RNAs (ceRNAs) sequester microRNAs to regulate mRNA transcripts containing common microRNA recognition elements (MREs). However, the functional role of ceRNAs in cancer remains unknown. Loss of PTEN, a tumor suppressor regulated by ceRNA activity, frequently occurs in melanoma. Here, we report the discovery of significant enrichment of putative PTEN ceRNAs among genes whose loss accelerates tumorigenesis following Sleeping Beauty insertional mutagenesis in a mouse model of melanoma. We validated several putative PTEN ceRNAs and further characterized one, the ZEB2 transcript. We show that ZEB2 modulates PTEN protein levels in a microRNA-dependent, protein coding-independent manner. Attenuation of ZEB2 expression activates the PI3K/AKT pathway, enhances cell transformation, and commonly occurs in human melanomas and other cancers expressing low PTEN levels. Our study genetically identifies multiple putative microRNA decoys for PTEN, validates ZEB2 mRNA as a bona fide PTEN ceRNA, and demonstrates that abrogated ZEB2 expression cooperates with BRAF(V600E) to promote melanomagenesis.