Estimating genome-wide gene networks using nonparametric Bayesian network models on massively parallel computers

Idiopathic pulmonary fibrosis-specific Bayesian network integrating extracellular vesicle proteome and clinical information

Idiopathic pulmonary fibrosis (IPF) is a progressive disease characterized by severe lung fibrosis and a poor prognosis. Although the biomolecules related to IPF have been extensively studied, molecular mechanisms of the pathogenesis and their association with serum biomarkers and clinical findings have not been fully elucidated. We constructed a Bayesian network using multimodal data consisting of a proteome dataset from serum extracellular vesicles, laboratory examinations, and clinical findings from 206 patients with IPF and 36 controls. Differential protein expression analysis was also performed by edgeR and incorporated into the constructed network. We have successfully visualized the relationship between biomolecules and clinical findings with this approach. The IPF-specific network included modules associated with TGF-β signaling (TGFB1 and LRC32), fibrosis-related (A2MG and PZP), myofibroblast and inflammation (LRP1 and ITIH4), complement-related (SAA1 and SAA2), as well as serum markers, and clinical symptoms (KL-6, SP-D and fine crackles). Notably, it identified SAA2 associated with lymphocyte counts and PSPB connected with the serum markers KL-6 and SP-D, along with fine crackles as clinical manifestations. These results contribute to the elucidation of the pathogenesis of IPF and potential therapeutic targets.

Nicotiana benthamiana XYLEM CYSTEINE PROTEASE genes facilitate tracheary element formation in interfamily grafting

Grafting is a plant propagation technique widely used in agriculture. A recent discovery of the capability of interfamily grafting in Nicotiana has expanded the potential combinations of grafting. In this study, we showed that xylem connection is essential for the achievement of interfamily grafting and investigated the molecular basis of xylem formation at the graft junction. Transcriptome and gene network analyses revealed gene modules for tracheary element (TE) formation during grafting that include genes associated with xylem cell differentiation and immune response. The reliability of the drawn network was validated by examining the role of the Nicotiana benthamiana XYLEM CYSTEINE PROTEASE (NbXCP) genes in TE formation during interfamily grafting. Promoter activities of NbXCP1 and NbXCP2 genes were found in differentiating TE cells in the stem and callus tissues at the graft junction. Analysis of a Nbxcp1;Nbxcp2 loss-of-function mutant indicated that NbXCPs control the timing of de novo TE formation at the graft junction. Moreover, grafts of the NbXCP1 overexpressor increased the scion growth rate as well as the fruit size. Thus, we identified gene modules for TE formation at the graft boundary and demonstrated potential ways to enhance Nicotiana interfamily grafting.

Parameter Learning of Bayesian Network with Multiplicative Synergistic Constraints

Learning the conditional probability table (CPT) parameters of Bayesian networks (BNs) is a key challenge in real-world decision support applications, especially when there are limited data available. The traditional approach to this challenge is introducing domain knowledge/expert judgments that are encoded as qualitative parameter constraints. In this paper, we focus on multiplicative synergistic constraints. The negative multiplicative synergy constraint and positive multiplicative synergy constraint in this paper are symmetric. In order to integrate multiplicative synergistic constraints into the learning process of Bayesian Network parameters, we propose four methods to deal with the multiplicative synergistic constraints based on the idea of classical isotonic regression algorithm. The four methods are simulated by using the lawn moist model and Asia network, and we compared them with the maximum likelihood estimation (MLE) algorithm. Simulation results show that the proposed methods are superior to the MLE algorithm in the accuracy of parameter learning, which can improve the results of the MLE algorithm to obtain more accurate estimators of the parameters. The proposed methods can reduce the dependence of parameter learning on expert experiences. Combining these constraint methods with Bayesian estimation can improve the accuracy of parameter learning under small sample conditions.

Dysregulation of the gene signature of effector regulatory T cells in the early phase of systemic sclerosis

OBJECTIVES

We evaluated flow-cytometric and transcriptome features of peripheral blood immune cells from early-phase (disease duration < 5 years) systemic sclerosis (SSc) in comparison to late-phase SSc.

METHODS

Fifty Japanese patients with SSc (12 early SSc cases and 38 late SSc cases) and 50 age- and sex-matched healthy controls were enrolled. A comparison of flow-cytometric subset proportions and RNA-sequencing of 24 peripheral blood immune cell subsets was performed. We evaluated differentially expressed genes (DEGs), characterized the co-expressed gene modules, and estimated the composition of subpopulations by deconvolution based on single-cell RNA-sequencing data. As a disease control, idiopathic inflammatory myositis (IIM) patients were also evaluated.

RESULTS

Analyzing the data from early and late SSc, Fraction II effector regulatory T cell (Fr. II eTreg) genes showed a remarkable differential gene expression, which was enriched for genes related to oxidative phosphorylation. Although the flow-cytometric proportion of Fr. II eTregs was not changed in early SSc, deconvolution indicated expansion of the activated subpopulation. Co-expressed gene modules of Fr. II eTregs demonstrated enrichment of the DEGs of early SSc and correlation with the proportion of the activated subpopulation. These results suggested that DEGs in Fr. II eTregs from patients with early SSc were closely associated with the increased proportion of the activated subpopulation. Similar dysregulation of Fr. II eTregs was also observed in data from patients with early IIM.

CONCLUSIONS

RNA-seq of immune cells indicated the dysregulation of Fr. II eTregs in early SSc with increased proportion of the activated subpopulation.

Novel cancer subtyping method based on patient-specific gene regulatory network

The identification of cancer subtypes is important for the understanding of tumor heterogeneity. In recent years, numerous computational methods have been proposed for this problem based on the multi-omics data of patients. It is widely accepted that different cancer subtypes are induced by different molecular regulatory networks. However, only a few incorporate the differences between their molecular systems into the identification processes. In this study, we present a novel method to identify cancer subtypes based on patient-specific molecular systems. Our method realizes this by quantifying patient-specific gene networks, which are estimated from their transcriptome data, and by clustering their quantified networks. Comprehensive analyses of The Cancer Genome Atlas (TCGA) datasets applied to our method confirmed that they were able to identify more clinically meaningful cancer subtypes than the existing subtypes and found that the identified subtypes comprised different molecular features. Our findings also show that the proposed method can identify the novel cancer subtypes even with single omics data, which cannot otherwise be captured by existing methods using multi-omics data.

A Fast and Furious Bayesian Network and Its Application of Identifying Colon Cancer to Liver Metastasis Gene Regulatory Networks

Bayesian networks is a powerful method for identifying causal relationships among variables. However, as the network size increases, the time complexity of searching the optimal structure grows exponentially. We proposed a novel search algorithm - Fast and Furious Bayesian Network (FFBN). Compared to the existing greedy search algorithm, FFBN uses significantly fewer model configuration rules to determine the causal direction of edges when constructing the Bayesian network, which leads to greatly improved computational speed. We benchmarked the performance of FFBN by reconstructing gene regulatory networks (GRNs) from two DREAM5 challenge datasets: a synthetic dataset and a larger yeast transcriptome dataset. In both datasets, FFBN shows a much faster speed than the existing greedy search algorithm, while maintaining equally good or better performance in recall and precision. We then constructed three whole transcriptome GRNs for primary liver cancer (PL), primary colon cancer (PC) and colon to liver metastasis (CLM) expression data, which the existing greedy search algorithms failed. Three GRNs contain 12,099 common genes. Unprecedentedly, our newly developed FFBN algorithm is able to build up GRNs at a scale larger than 10,000 genes. Using FFBN, we discovered that CLM has its unique cancer molecular mechanisms and shares a certain degree of similarity with both PL and PC.

Dynamic changes in gene-to-gene regulatory networks in response to SARS-CoV-2 infection

The current pandemic of SARS-CoV-2 has caused extensive damage to society. The characterization of SARS-CoV-2 profiles has been addressed by researchers globally with the aim of resolving this disruptive crisis. This investigation process is indispensable to understand how SARS-CoV-2 behaves in human host cells. However, little is known about the systematic molecular mechanisms involved in the effects of SARS-CoV-2 infection on human host cells. Here, we present gene-to-gene regulatory networks in response to SARS-CoV-2 using a Bayesian network. We examined the dynamic changes in the SARS-CoV-2-purturbated networks established by our proposed framework for gene network analysis, thus revealing that interferon signaling gradually switched to the subsequent inflammatory cytokine signaling cascades. Furthermore, we succeeded in capturing a COVID-19 patient-specific network in which transduction of these signals was concurrently induced. This enabled us to explore the local regulatory systems influenced by SARS-CoV-2 in host cells more precisely at an individual level. Our panel of network analyses has provided new insights into SARS-CoV-2 research from the perspective of cellular systems.

High-throughput toxicogenomic screening of chemicals in the environment using metabolically competent hepatic cell cultures

The ToxCast in vitro screening program has provided concentration-response bioactivity data across more than a thousand assay endpoints for thousands of chemicals found in our environment and commerce. However, most ToxCast screening assays have evaluated individual biological targets in cancer cell lines lacking integrated physiological functionality (such as receptor signaling, metabolism). We evaluated differentiated HepaRGTM cells, a human liver-derived cell model understood to effectively model physiologically relevant hepatic signaling. Expression of 93 gene transcripts was measured by quantitative polymerase chain reaction using Fluidigm 96.96 dynamic arrays in response to 1060 chemicals tested in eight-point concentration-response. A Bayesian framework quantitatively modeled chemical-induced changes in gene expression via six transcription factors including: aryl hydrocarbon receptor, constitutive androstane receptor, pregnane X receptor, farnesoid X receptor, androgen receptor, and peroxisome proliferator-activated receptor alpha. For these chemicals the network model translates transcriptomic data into Bayesian inferences about molecular targets known to activate toxicological adverse outcome pathways. These data also provide new insights into the molecular signaling network of HepaRGTM cell cultures.

Long non-coding RNA HCP5 in cancer

Cancer remains a major threat to human health worldwide. Long non-coding RNA (lncRNA) comprises a group of single-stranded RNA with lengths longer than 200 bp. LncRNAs are aberrantly expressed and play a variety of roles involving multiple cellular processes in cancer. Histocompatibility leukocyte antigen complex P5 (HCP5), initially reported in 1993, is an important lncRNA located between the MICA and MICB genes in MHC I region. HCP5 is involved many autoimmune diseases as well as malignancies. Abnormal HCP5 expression occurs in many types of cancer and its dysregulation appears closely associated with tumor progression. HCP5 is also involved in anti-tumor drug resistance as well. As such, HCP5 represents a promising biomarker and therapeutic target in cancer. In this review, we summarize recent researches and provide an overview of the role and mechanism of HCP5 in human cancer.

System-Based Differential Gene Network Analysis for Characterizing a Sample-Specific Subnetwork

Gene network estimation is a method key to understanding a fundamental cellular system from high throughput omics data. However, the existing gene network analysis relies on having a sufficient number of samples and is required to handle a huge number of nodes and estimated edges, which remain difficult to interpret, especially in discovering the clinically relevant portions of the network. Here, we propose a novel method to extract a biomedically significant subnetwork using a Bayesian network, a type of unsupervised machine learning method that can be used as an explainable and interpretable artificial intelligence algorithm. Our method quantifies sample specific networks using our proposed Edge Contribution value (ECv) based on the estimated system, which realizes condition-specific subnetwork extraction using a limited number of samples. We applied this method to the Epithelial-Mesenchymal Transition (EMT) data set that is related to the process of metastasis and thus prognosis in cancer biology. We established our method-driven EMT network representing putative gene interactions. Furthermore, we found that the sample-specific ECv patterns of this EMT network can characterize the survival of lung cancer patients. These results show that our method unveils the explainable network differences in biological and clinical features through artificial intelligence technology.

Parallel Algorithms for Inferring Gene Regulatory Networks: A Review

System biology problems such as whole-genome network construction from large-scale gene expression data are sophisticated and time-consuming. Therefore, using sequential algorithms are not feasible to obtain a solution in an acceptable amount of time. Today, by using massively parallel computing, it is possible to infer large-scale gene regulatory networks. Recently, establishing gene regulatory networks from large-scale datasets have drawn the noticeable attention of researchers in the field of parallel computing and system biology. In this paper, we attempt to provide a more detailed overview of the recent parallel algorithms for constructing gene regulatory networks. Firstly, fundamentals of gene regulatory networks inference and large-scale datasets challenges are given. Secondly, a detailed description of the four parallel frameworks and libraries including CUDA, OpenMP, MPI, and Hadoop is discussed. Thirdly, parallel algorithms are reviewed. Finally, some conclusions and guidelines for parallel reverse engineering are described.

Random Matrix Analysis for Gene Interaction Networks in Cancer Cells

Investigations of topological uniqueness of gene interaction networks in cancer cells are essential for understanding the disease. Although cancer is considered to originate from the topological alteration of a huge molecular interaction network in cellular systems, the theoretical study to investigate such complex networks is still insufficient. It is necessary to predict the behavior of a huge complex interaction network from the behavior of a finite size network. Based on the random matrix theory, we study the distribution of the nearest neighbor level spacings P(s) of interaction matrices of gene networks in human cancer cells. The interaction matrices are computed using the Cancer Network Galaxy (TCNG) database which is a repository of gene interactions inferred by a Bayesian network model. 256 NCBI GEO entries regarding gene expressions in human cancer cells have been used for the inference. We observe the Wigner distribution of P(s) when the gene networks are dense networks that have more than ~38,000 edges. In the opposite case, when the networks have smaller numbers of edges, the distribution P(s) becomes the Poisson distribution. We investigate relevance of P(s) both to the sparseness of the networks and to edge frequency factor which is the reliance (likelihood) of the inferred gene interactions.

Oncogenic roles of TOPK and MELK, and effective growth suppression by small molecular inhibitors in kidney cancer cells

T–lymphokine-activated killer cell–originated protein kinase (TOPK) and maternal embryonic leucine zipper kinase (MELK) have been reported to play critical roles in cancer cell proliferation and maintenance of stemness. In this study, we investigated possible roles of TOPK and MELK in kidney cancer cells and found their growth promotive effect as well as some feedback mechanism between these two molecules. Interestingly, the blockade of either of these two kinases effectively caused downregulation of forkhead box protein M1 (FOXM1) activity which is known as an oncogenic transcriptional factor in various types of cancer cells. Small molecular compound inhibitors against TOPK (OTS514) and MELK (OTS167) effectively suppressed the kidney cancer cell growth, and the combination of these two compounds additively worked and showed the very strong growth suppressive effect on kidney cancer cells. Collectively, our results suggest that both TOPK and MELK are promising molecular targets for kidney cancer treatment and that dual blockade of OTS514 and OTS167 may bring additive anti-tumor effects with low risk of side effects.

Microarray Expression Data Identify DCC as a Candidate Gene for Early Meningioma Progression

Meningiomas are the most common primary brain tumors bearing in a minority of cases an aggressive phenotype. Although meningiomas are stratified according to their histology and clinical behavior, the underlying molecular genetics predicting aggressiveness are not thoroughly understood. We performed whole transcript expression profiling in 10 grade I and four grade II meningiomas, three of which invaded the brain. Microarray expression analysis identified deleted in colorectal cancer (DCC) as a differentially expressed gene (DEG) enabling us to cluster meningiomas into DCC low expression (3 grade I and 3 grade II tumors), DCC medium expression (2 grade I and 1 grade II tumors), and DCC high expression (5 grade I tumors) groups. Comparison between the DCC low expression and DCC high expression groups resulted in 416 DEGs (p-value < 0.05; fold change > 2). The most significantly downregulated genes in the DCC low expression group comprised DCC, phosphodiesterase 1C (PDE1C), calmodulin-dependent 70kDa olfactomedin 2 (OLFM2), glutathione S-transferase mu 5 (GSTM5), phosphotyrosine interaction domain containing 1 (PID1), sema domain, transmembrane domain (TM) and cytoplasmic domain, (semaphorin) 6D (SEMA6D), and indolethylamine N-methyltransferase (INMT). The most significantly upregulated genes comprised chromosome 5 open reading frame 63 (C5orf63), homeodomain interacting protein kinase 2 (HIPK2), and basic helix-loop-helix family, member e40 (BHLHE40). Biofunctional analysis identified as predicted top upstream regulators beta-estradiol, TGFB1, Tgf beta complex, LY294002, and dexamethasone and as predicted top regulator effectors NFkB, PIK3R1, and CREBBP. The microarray expression data served also for a comparison between meningiomas from female and male patients and for a comparison between brain invasive and non-invasive meningiomas resulting in a number of significant DEGs and related biofunctions. In conclusion, based on its expression levels, DCC may constitute a valid biomarker to identify those benign meningiomas at risk for progression.

Parallel Mutual Information Based Construction of Genome-Scale Networks on the Intel® Xeon Phi™ Coprocessor

Construction of whole-genome networks from large-scale gene expression data is an important problem in systems biology. While several techniques have been developed, most cannot handle network reconstruction at the whole-genome scale, and the few that can, require large clusters. In this paper, we present a solution on the Intel Xeon Phi coprocessor, taking advantage of its multi-level parallelism including many x86-based cores, multiple threads per core, and vector processing units. We also present a solution on the Intel® Xeon® processor. Our solution is based on TINGe, a fast parallel network reconstruction technique that uses mutual information and permutation testing for assessing statistical significance. We demonstrate the first ever inference of a plant whole genome regulatory network on a single chip by constructing a 15,575 gene network of the plant Arabidopsis thaliana from 3,137 microarray experiments in only 22 minutes. In addition, our optimization for parallelizing mutual information computation on the Intel Xeon Phi coprocessor holds out lessons that are applicable to other domains.

Inference of TFRNs (2)

This chapter introduces to a study aiming at comprehensively understanding the transcription factor regulatory networks (TFRNs) that govern the process of a cell differentiation. Here we focus on the adipocyte differentiation. For the cell differentiation, we inferred its TFRN using the Bayesian network (BN) method. BNs have been widely used to estimate TFRNs. Many BN methods have been developed to estimate networks from TF expression data. However, BN-based methods require huge computational time to estimate large-scale networks. This chapter introduces to a BN-based deterministic method with reduced computational time. This approach generates all the combinational subnetworks of three TFs, estimates networks of the subnetworks by BN, and unites the networks into a single large network. This method decreases the search space of predicting TFRNs without degrading the solution accuracy compared with the greedy hill climbing (GHC) method. This chapter also presents a massively parallel implementation for the BN-based inference of TFRNs. The system enables us to estimate large-scale (>10,000 transcripts) multiple TFRNs from expression profiles of various tissues and conditions. The comparison among estimated TFRNs of adipose tissues with stimulus induction is conducted. The various regulations to Ucp1 (uncoupled protein 1) in those networks may reflect different responses of the tissues under the stimulus induction.

Reconstruction of large-scale gene regulatory networks using Bayesian model averaging

Gene regulatory networks provide the systematic view of molecular interactions in a complex living system. However, constructing large-scale gene regulatory networks is one of the most challenging problems in systems biology. Also large burst sets of biological data require a proper integration technique for reliable gene regulatory network construction. Here we present a new reverse engineering approach based on Bayesian model averaging which attempts to combine all the appropriate models describing interactions among genes. This Bayesian approach with a prior based on the Gibbs distribution provides an efficient means to integrate multiple sources of biological data. In a simulation study with maximum of 2000 genes, our method shows better sensitivity than previous elastic-net and Gaussian graphical models, with a fixed specificity of 0.99. The study also shows that the proposed method outperforms the other standard methods for a DREAM dataset generated by nonlinear stochastic models. In brain tumor data analysis, three large-scale networks consisting of 4422 genes were built using the gene expression of non-tumor, low and high grade tumor mRNA expression samples, along with DNA-protein binding affinity information. We found that genes having a large variation of degree distribution among the three tumor networks are the ones that see most involved in regulatory and developmental processes, which possibly gives a novel insight concerning conventional differentially expressed gene analysis.

Reverse engineering and analysis of large genome-scale gene networks

Reverse engineering the whole-genome networks of complex multicellular organisms continues to remain a challenge. While simpler models easily scale to large number of genes and gene expression datasets, more accurate models are compute intensive limiting their scale of applicability. To enable fast and accurate reconstruction of large networks, we developed Tool for Inferring Network of Genes (TINGe), a parallel mutual information (MI)-based program. The novel features of our approach include: (i) B-spline-based formulation for linear-time computation of MI, (ii) a novel algorithm for direct permutation testing and (iii) development of parallel algorithms to reduce run-time and facilitate construction of large networks. We assess the quality of our method by comparison with ARACNe (Algorithm for the Reconstruction of Accurate Cellular Networks) and GeneNet and demonstrate its unique capability by reverse engineering the whole-genome network of Arabidopsis thaliana from 3137 Affymetrix ATH1 GeneChips in just 9 min on a 1024-core cluster. We further report on the development of a new software Gene Network Analyzer (GeNA) for extracting context-specific subnetworks from a given set of seed genes. Using TINGe and GeNA, we performed analysis of 241 Arabidopsis AraCyc 8.0 pathways, and the results are made available through the web.

Cell cycle gene networks are associated with melanoma prognosis

Background

Our understanding of the molecular pathways that underlie melanoma remains incomplete. Although several published microarray studies of clinical melanomas have provided valuable information, we found only limited concordance between these studies. Therefore, we took an in vitro functional genomics approach to understand melanoma molecular pathways.

Methodology/Principal Findings

Affymetrix microarray data were generated from A375 melanoma cells treated in vitro with siRNAs against 45 transcription factors and signaling molecules. Analysis of this data using unsupervised hierarchical clustering and Bayesian gene networks identified proliferation-association RNA clusters, which were co-ordinately expressed across the A375 cells and also across melanomas from patients. The abundance in metastatic melanomas of these cellular proliferation clusters and their putative upstream regulators was significantly associated with patient prognosis. An 8-gene classifier derived from gene network hub genes correctly classified the prognosis of 23/26 metastatic melanoma patients in a cross-validation study. Unlike the RNA clusters associated with cellular proliferation described above, co-ordinately expressed RNA clusters associated with immune response were clearly identified across melanoma tumours from patients but not across the siRNA-treated A375 cells, in which immune responses are not active. Three uncharacterised genes, which the gene networks predicted to be upstream of apoptosis- or cellular proliferation-associated RNAs, were found to significantly alter apoptosis and cell number when over-expressed in vitro.

Conclusions/Significance

This analysis identified co-expression of RNAs that encode functionally-related proteins, in particular, proliferation-associated RNA clusters that are linked to melanoma patient prognosis. Our analysis suggests that A375 cells in vitro may be valid models in which to study the gene expression modules that underlie some melanoma biological processes (e.g., proliferation) but not others (e.g., immune response). The gene expression modules identified here, and the RNAs predicted by Bayesian network inference to be upstream of these modules, are potential prognostic biomarkers and drug targets.

An estimation method for inference of gene regulatory net-work using Bayesian network with uniting of partial problems

Background

Bayesian networks (BNs) have been widely used to estimate gene regulatory networks. Many BN methods have been developed to estimate networks from microarray data. However, two serious problems reduce the effectiveness of current BN methods. The first problem is that BN-based methods require huge computational time to estimate large-scale networks. The second is that the estimated network cannot have cyclic structures, even if the actual network has such structures.

Results

In this paper, we present a novel BN-based deterministic method with reduced computational time that allows cyclic structures. Our approach generates all the combinational triplets of genes, estimates networks of the triplets by BN, and unites the networks into a single network containing all genes. This method decreases the search space of predicting gene regulatory networks without degrading the solution accuracy compared with the greedy hill climbing (GHC) method. The order of computational time is the cube of number of genes. In addition, the network estimated by our method can include cyclic structures.

Conclusions

We verified the effectiveness of the proposed method for all known gene regulatory networks and their expression profiles. The results demonstrate that this approach can predict regulatory networks with reduced computational time without degrading the solution accuracy compared with the GHC method.

Systems biology data analysis methodology in pharmacogenomics

Pharmacogenetics aims to elucidate the genetic factors underlying the individual's response to pharmacotherapy. Coupled with the recent (and ongoing) progress in high-throughput genotyping, sequencing and other genomic technologies, pharmacogenetics is rapidly transforming into pharmacogenomics, while pursuing the primary goals of identifying and studying the genetic contribution to drug therapy response and adverse effects, and existing drug characterization and new drug discovery. Accomplishment of both of these goals hinges on gaining a better understanding of the underlying biological systems; however, reverse-engineering biological system models from the massive datasets generated by the large-scale genetic epidemiology studies presents a formidable data analysis challenge. In this article, we review the recent progress made in developing such data analysis methodology within the paradigm of systems biology research that broadly aims to gain a 'holistic', or 'mechanistic' understanding of biological systems by attempting to capture the entirety of interactions between the components (genetic and otherwise) of the system.

Gene network inference and visualization tools for biologists: application to new human transcriptome datasets

Gene regulatory networks inferred from RNA abundance data have generated significant interest, but despite this, gene network approaches are used infrequently and often require input from bioinformaticians. We have assembled a suite of tools for analysing regulatory networks, and we illustrate their use with microarray datasets generated in human endothelial cells. We infer a range of regulatory networks, and based on this analysis discuss the strengths and limitations of network inference from RNA abundance data. We welcome contact from researchers interested in using our inference and visualization tools to answer biological questions.