Integrating siRNA and protein-protein interaction data to identify an expanded insulin signaling network

SIGNAL: A web-based iterative analysis platform integrating pathway and network approaches optimizes hit selection from genome-scale assays

Hit selection from high-throughput assays remains a critical bottleneck in realizing the potential of omic-scale studies in biology. Widely used methods such as setting of cutoffs, prioritizing pathway enrichments, or incorporating predicted network interactions offer divergent solutions yet are associated with critical analytical trade-offs. The specific limitations of these individual approaches and the lack of a systematic way by which to integrate their rankings have contributed to limited overlap in the reported results from comparable genome-wide studies and costly inefficiencies in secondary validation efforts. Using comparative analysis of parallel independent studies as a benchmark, we characterize the specific complementary contributions of each approach and demonstrate an optimal framework to integrate these methods. We describe selection by iterative pathway group and network analysis looping (SIGNAL), an integrated, iterative approach that uses both pathway and network methods to optimize gene prioritization. SIGNAL is accessible as a rapid user-friendly web-based application (https://signal.niaid.nih.gov). A record of this paper's transparent peer review is included in the Supplemental information.

Network models of primary melanoma microenvironments identify key melanoma regulators underlying prognosis

Melanoma is the most lethal skin malignancy, driven by genetic and epigenetic alterations in the complex tumour microenvironment. While large-scale molecular profiling of melanoma has identified molecular signatures associated with melanoma progression, comprehensive systems-level modeling remains elusive. This study builds up predictive gene network models of molecular alterations in primary melanoma by integrating large-scale bulk-based multi-omic and single-cell transcriptomic data. Incorporating clinical, epigenetic, and proteomic data into these networks reveals key subnetworks, cell types, and regulators underlying melanoma progression. Tumors with high immune infiltrates are found to be associated with good prognosis, presumably due to induced CD8+ T-cell cytotoxicity, via MYO1F-mediated M1-polarization of macrophages. Seventeen key drivers of the gene subnetworks associated with poor prognosis, including the transcription factor ZNF180, are tested for their pro-tumorigenic effects in vitro. The anti-tumor effect of silencing ZNF180 is further validated using in vivo xenografts. Experimentally validated targets of ZNF180 are enriched in the ZNF180 centered network and the known pathways such as melanoma cell maintenance and immune cell infiltration. The transcriptional networks and their critical regulators provide insights into the molecular mechanisms of melanomagenesis and pave the way for developing therapeutic strategies for melanoma.

While the molecular profiling of melanoma progression has been extensively characterised by large-scale studies, there is still need for the comprehensive integration of such datasets. Here the authors construct predictive gene network models for prognostic and therapeutic purposes.

An integrative multiomic network model links lipid metabolism to glucose regulation in coronary artery disease

Elevated plasma cholesterol and type 2 diabetes (T2D) are associated with coronary artery disease (CAD). Individuals treated with cholesterol-lowering statins have increased T2D risk, while individuals with hypercholesterolemia have reduced T2D risk. We explore the relationship between lipid and glucose control by constructing network models from the STARNET study with sequencing data from seven cardiometabolic tissues obtained from CAD patients during coronary artery by-pass grafting surgery. By integrating gene expression, genotype, metabolomic, and clinical data, we identify a glucose and lipid determining (GLD) regulatory network showing inverse relationships with lipid and glucose traits. Master regulators of the GLD network also impact lipid and glucose levels in inverse directions. Experimental inhibition of one of the GLD network master regulators, lanosterol synthase (LSS), in mice confirms the inverse relationships to glucose and lipid levels as predicted by our model and provides mechanistic insights.

MasterPATH: network analysis of functional genomics screening data

Background

Functional genomics employs several experimental approaches to investigate gene functions. High-throughput techniques, such as loss-of-function screening and transcriptome profiling, allow to identify lists of genes potentially involved in biological processes of interest (so called hit list). Several computational methods exist to analyze and interpret such lists, the most widespread of which aim either at investigating of significantly enriched biological processes, or at extracting significantly represented subnetworks.

Results

Here we propose a novel network analysis method and corresponding computational software that employs the shortest path approach and centrality measure to discover members of molecular pathways leading to the studied phenotype, based on functional genomics screening data. The method works on integrated interactomes that consist of both directed and undirected networks – HIPPIE, SIGNOR, SignaLink, TFactS, KEGG, TransmiR, miRTarBase. The method finds nodes and short simple paths with significant high centrality in subnetworks induced by the hit genes and by so-called final implementers – the genes that are involved in molecular events responsible for final phenotypic realization of the biological processes of interest. We present the application of the method to the data from miRNA loss-of-function screen and transcriptome profiling of terminal human muscle differentiation process and to the gene loss-of-function screen exploring the genes that regulates human oxidative DNA damage recognition. The analysis highlighted the possible role of several known myogenesis regulatory miRNAs (miR-1, miR-125b, miR-216a) and their targets (AR, NR3C1, ARRB1, ITSN1, VAV3, TDGF1), as well as linked two major regulatory molecules of skeletal myogenesis, MYOD and SMAD3, to their previously known muscle-related targets (TGFB1, CDC42, CTCF) and also to a number of proteins such as C-KIT that have not been previously studied in the context of muscle differentiation. The analysis also showed the role of the interaction between H3 and SETDB1 proteins for oxidative DNA damage recognition.

Conclusion

The current work provides a systematic methodology to discover members of molecular pathways in integrated networks using functional genomics screening data. It also offers a valuable instrument to explain the appearance of a set of genes, previously not associated with the process of interest, in the hit list of each particular functional genomics screening.

Multiscale causal networks identify VGF as a key regulator of Alzheimer's disease

Though discovered over 100 years ago, the molecular foundation of sporadic Alzheimer's disease (AD) remains elusive. To better characterize the complex nature of AD, we constructed multiscale causal networks on a large human AD multi-omics dataset, integrating clinical features of AD, DNA variation, and gene- and protein-expression. These probabilistic causal models enabled detection, prioritization and replication of high-confidence master regulators of AD-associated networks, including the top predicted regulator, VGF. Overexpression of neuropeptide precursor VGF in 5xFAD mice partially rescued beta-amyloid-mediated memory impairment and neuropathology. Molecular validation of network predictions downstream of VGF was also achieved in this AD model, with significant enrichment for homologous genes identified as differentially expressed in 5xFAD brains overexpressing VGF. Our findings support a causal role for VGF in protecting against AD pathogenesis and progression.

Massive computational identification of somatic variants in exonic splicing enhancers using The Cancer Genome Atlas

Owing to the development of next-generation sequencing (NGS) technologies, a large number of somatic variants have been identified in various types of cancer. However, the functional significance of most somatic variants remains unknown. Somatic variants that occur in exonic splicing enhancer (ESE) regions are thought to prevent serine and arginine-rich (SR) proteins from binding to ESE sequence motifs, which leads to exon skipping. We computationally identified somatic variants in ESEs by compiling numerous open-access datasets from The Cancer Genome Atlas (TCGA). Using somatic variants and RNA-seq data from 9635 patients across 32 TCGA projects, we identified 646 ESE-disrupting variants. The false positive rate of our method, estimated using a permutation test, was approximately 1%. Of these ESE-disrupting variants, approximately 71% were located in the binding motifs of four classical SR proteins. ESE-disrupting variants occurred in proportion to the number of somatic variants, but not necessarily in the specific genes associated with the biological processes of cancer. Existing bioinformatics tools could not predict the pathogenicity of ESE-disrupting variants identified in this study, although these variants could cause exon skipping. We demonstrated that ESE-disrupting nonsense variants tended to escape nonsense-mediated decay surveillance. Using integrated analyses of open access data, we could specifically identify ESE-disrupting variants. We have generated a powerful tool, which can handle datasets without normal samples or raw data, and thus contribute to reducing variants of uncertain significance because our statistical approach only uses the exon-junction read counts from the tumor samples.

Identification of Key Signaling Pathways Orchestrating Substrate Topography Directed Osteogenic Differentiation Through High-Throughput siRNA Screening

Fibrous scaffolds are used for bone tissue engineering purposes with great success across a variety of polymers with different physical and chemical properties. It is now evident that the correct degree of curvature promotes increased cytoskeletal tension on osteoprogenitors leading to osteogenic differentiation. However, the mechanotransductive pathways involved in this phenomenon are not fully understood. To achieve a reproducible and specific cellular response, an increased mechanistic understanding of the molecular mechanisms driving the fibrous scaffold mediated bone regeneration must be understood. High throughput siRNA mediated screening technology has been utilized for dissecting molecular targets that are important in certain cellular phenotypes. In this study, we used siRNA mediated gene silencing to understand the osteogenic differentiation observed on fibrous scaffolds. A high-throughput siRNA screen was conducted using a library collection of 863 genes including important human kinase and phosphatase targets on pre-osteoblast SaOS-2 cells. The cells were grown on electrospun poly(methyl methacrylate) (PMMA) scaffolds with a diameter of 0.938 ± 0.304 µm and a flat surface control. The osteogenic transcription factor RUNX2 was quantified with an in-cell western (ICW) assay for the primary screen and significant targets were selected via two sample t-test. After selecting the significant targets, a secondary screen was performed to identify osteoinductive markers that also effect cell shape on fibrous topography. Finally, we report the most physiologically relevant molecular signaling mechanisms that are involved in growth factor free, fibrous topography mediated osteoinduction. We identified GTPases, membrane channel proteins, and microtubule associated targets that promote an osteoinductive cell shape on fibrous scaffolds.

Construction of Signaling Pathways with RNAi Data and Multiple Reference Networks

Signaling networks are involved in almost all major diseases such as cancer. As a result of this, understanding how signaling networks function is vital for finding new treatments for many diseases. Using gene knockdown assays such as RNA interference (RNAi) technology, many genes involved in these networks can be identified. However, determining the interactions between these genes in the signaling networks using only experimental techniques is very challenging, as performing extensive experiments is very expensive and sometimes, even impractical. Construction of signaling networks from RNAi data using computational techniques have been proposed as an alternative way to solve this challenging problem. However, the earlier approaches are either not scalable to large scale networks, or their accuracy levels are not satisfactory. In this study, we integrate RNAi data given on a target network with multiple reference signaling networks and phylogenetic trees to construct the topology of the target signaling network. In our work, the network construction is considered as finding the minimum number of edit operations on given multiple reference networks, in which their contributions are weighted by their phylogenetic distances to the target network. The edit operations on the reference networks lead to a target network that satisfies the RNAi knockdown observations. Here, we propose two new reference-based signaling network construction methods that provide optimal results and scale well to large-scale signaling networks of hundreds of components. We compare the performance of these approaches to the state-of-the-art reference-based network construction method SiNeC on synthetic, semi-synthetic, and real datasets. Our analyses show that the proposed methods outperform SiNeC method in terms of accuracy. Furthermore, we show that our methods function well even if evolutionarily distant reference networks are used. Application of our methods to the Apoptosis and Wnt signaling pathways recovers the known protein-protein interactions and suggests additional relevant interactions that can be tested experimentally.

Functional genomics of lipid metabolism in the oleaginous yeast Rhodosporidium toruloides

The basidiomycete yeast Rhodosporidium toruloides (also known as Rhodotorula toruloides) accumulates high concentrations of lipids and carotenoids from diverse carbon sources. It has great potential as a model for the cellular biology of lipid droplets and for sustainable chemical production. We developed a method for high-throughput genetics (RB-TDNAseq), using sequence-barcoded Agrobacterium tumefaciens T-DNA insertions. We identified 1,337 putative essential genes with low T-DNA insertion rates. We functionally profiled genes required for fatty acid catabolism and lipid accumulation, validating results with 35 targeted deletion strains. We identified a high-confidence set of 150 genes affecting lipid accumulation, including genes with predicted function in signaling cascades, gene expression, protein modification and vesicular trafficking, autophagy, amino acid synthesis and tRNA modification, and genes of unknown function. These results greatly advance our understanding of lipid metabolism in this oleaginous species and demonstrate a general approach for barcoded mutagenesis that should enable functional genomics in diverse fungi.

The fungus Rhodosporidium toruloides can grow on substances extracted from plant matter that is inedible to humans such as corn stalks, wood pulp, and grasses. Under some growth conditions, the fungus can accumulate massive stores of hydrocarbon-rich fats and pigments. A community of scientists and engineers has begun genetically modifying R. toruloides to convert these naturally produced fats and pigments into fuels, chemicals and medicines. These could form sustainable replacements for products made from petroleum or harvested from threatened animal and plant species.

Fungi, plants, animals and other eukaryotes store fat in specialized compartments called lipid droplets. The genes that control the metabolism – the production, use and storage – of fat in lipid bodies have been studied in certain eukaryotes, including species of yeast. However, R. toruloides is only distantly related to the most well-studied of these species. This means that we cannot be certain that a gene will play the same role in R. toruloides as in those species.

To assemble the most comprehensive list possible of the genes in R. toruloides that affect the production, use, or storage of fat in lipid bodies, Coradetti, Pinel et al. constructed a population of hundreds of thousands of mutant fungal strains, each with its own unique DNA ‘barcode’. The effects that mutations in over 6,000 genes had on growth and fat accumulation in these fungi were measured simultaneously in several experiments. This general approach is not new, but technical limitations had, until now, restricted its use in fungi to a few species.

Coradetti, Pinel et al. identified hundreds of genes that affected the ability of R. toruloides to metabolise fat. Many of these genes were related to genes with known roles in fat metabolism in other eukaryotes. Other genes are involved in different cell processes, such as the recycling of waste products in the cell. Their identification adds weight to the view that the links between these cellular processes and fat metabolism are deep and widespread amongst eukaryotes. Finally, some of the genes identified by Coradetti, Pinel et al. are not closely related to any well-studied genes. Further study of these genes could help us to understand why R. toruloides can accumulate much larger amounts of fat than most other fungi.

The methods developed by Coradetti, Pinel et al. should be possible to implement in many species of fungi. As a result these techniques may eventually contribute to the development of new treatments for human fungal diseases, the protection of important food crops, and a deeper understanding of the roles various fungi play in the broader ecosystem.

Genome-Scale Networks Link Neurodegenerative Disease Genes to α-Synuclein through Specific Molecular Pathways

Numerous genes and molecular pathways are implicated in neurodegenerative proteinopathies, but their inter-relationships are poorly understood. We systematically mapped molecular pathways underlying the toxicity of alpha-synuclein (α-syn), a protein central to Parkinson's disease. Genome-wide screens in yeast identified 332 genes that impact α-syn toxicity. To "humanize" this molecular network, we developed a computational method, TransposeNet. This integrates a Steiner prize-collecting approach with homology assignment through sequence, structure, and interaction topology. TransposeNet linked α-syn to multiple parkinsonism genes and druggable targets through perturbed protein trafficking and ER quality control as well as mRNA metabolism and translation. A calcium signaling hub linked these processes to perturbed mitochondrial quality control and function, metal ion transport, transcriptional regulation, and signal transduction. Parkinsonism gene interaction profiles spatially opposed in the network (ATP13A2/PARK9 and VPS35/PARK17) were highly distinct, and network relationships for specific genes (LRRK2/PARK8, ATXN2, and EIF4G1/PARK18) were confirmed in patient induced pluripotent stem cell (iPSC)-derived neurons. This cross-species platform connected diverse neurodegenerative genes to proteinopathy through specific mechanisms and may facilitate patient stratification for targeted therapy.

MEDICI: Mining Essentiality Data to Identify Critical Interactions for Cancer Drug Target Discovery and Development

Protein-protein interactions (PPIs) mediate the transmission and regulation of oncogenic signals that are essential to cellular proliferation and survival, and thus represent potential targets for anti-cancer therapeutic discovery. Despite their significance, there is no method to experimentally disrupt and interrogate the essentiality of individual endogenous PPIs. The ability to computationally predict or infer PPI essentiality would help prioritize PPIs for drug discovery and help advance understanding of cancer biology. Here we introduce a computational method (MEDICI) to predict PPI essentiality by combining gene knockdown studies with network models of protein interaction pathways in an analytic framework. Our method uses network topology to model how gene silencing can disrupt PPIs, relating the unknown essentialities of individual PPIs to experimentally observed protein essentialities. This model is then deconvolved to recover the unknown essentialities of individual PPIs. We demonstrate the validity of our approach via prediction of sensitivities to compounds based on PPI essentiality and differences in essentiality based on genetic mutations. We further show that lung cancer patients have improved overall survival when specific PPIs are no longer present, suggesting that these PPIs may be potentially new targets for therapeutic development. Software is freely available at https://github.com/cooperlab/MEDICI. Datasets are available at https://ctd2.nci.nih.gov/dataPortal.

Identifying the topology of signaling networks from partial RNAi data

Background

Methods for inferring signaling networks using single gene knockdown RNAi experiments and reference networks have been proposed in recent years. These methods assume that RNAi information is available for all the genes in the signal transduction pathway, i.e., complete. This assumption does not always hold up since RNAi experiments are often incomplete and information for some genes is missing.

Results

In this article, we develop two methods to construct signaling networks from incomplete RNAi data with the help of a reference network. These methods infer the RNAi constraints for the missing genes such that the inferred network is closest to the reference network. We perform extensive experiments with both real and synthetic networks and demonstrate that these methods produce accurate results efficiently.

Conclusions

Application of our methods to Wnt signal transduction pathway has shown that our methods can be used to construct highly accurate signaling networks from experimental data in less than 100 ms. The two methods that produce accurate results efficiently show great promise of constructing real signaling networks.

Sphingosine-1-Phosphate Receptor 2 Regulates Proinflammatory Cytokine Production and Osteoclastogenesis

Sphingosine-1-phosphate receptor 2 (S1PR2) couples with the Gi, Gq, and G_12/13 group of proteins, which modulate an array of cellular signaling pathways and affect immune responses to multiple stimuli. In this study, we demonstrated that knockdown of S1PR2 by a specific S1PR2 shRNA lentiviral vector significantly inhibited IL-1β, IL-6, and TNF-α protein levels induced by oral pathogen Aggregatibacter actinomycetemcomitans (A. actinomycetemcomitans) in murine bone marrow-derived monocytes and macrophages (BMMs) compared with controls. In addition, knockdown of S1PR2 by the S1PR2 shRNA lentiviral vector suppressed p-PI3K, p-ERK, p-JNK, p-p38, and p-NF-κBp65 protein expressions induced by A. actinomycetemcomitans. Furthermore, bone marrow cells treated with the S1PR2 shRNA lentiviral vector inhibited osteoclastogenesis induced by RANKL compared with controls. The S1PR2 shRNA suppressed the mRNA levels of six osteoclastogenic factors including nuclear factor of activated T-cells cytoplasmic calcineurin-dependent 1 (NFATc1), cathepsin K (Ctsk), acid phosphatase 5 (Acp5), osteoclast-associated receptor (Oscar), dendritic cells specific transmembrane protein (Dcstamp), and osteoclast stimulatory transmembrane protein (Ocstamp) in bone marrow cells. We conclude that S1PR2 plays an essential role in modulating proinflammatory cytokine production and osteoclastogenesis. Blocking S1PR2 signaling might be a novel therapeutic strategy to treat inflammatory bone loss diseases.

A combined binary interaction and phenotypic map of C. elegans cell polarity proteins

The establishment of cell polarity is an essential process for the development of multicellular organisms and the functioning of cells and tissues. Here, we combine large-scale protein interaction mapping with systematic phenotypic profiling to study the network of physical interactions that underlies polarity establishment and maintenance in the nematode Caenorhabditis elegans. Using a fragment-based yeast two-hybrid strategy, we identified 439 interactions between 296 proteins, as well as the protein regions that mediate these interactions. Phenotypic profiling of the network resulted in the identification of 100 physically interacting protein pairs for which RNAi-mediated depletion caused a defect in the same polarity-related process. We demonstrate the predictive capabilities of the network by showing that the physical interaction between the RhoGAP PAC-1 and PAR-6 is required for radial polarization of the C. elegans embryo. Our network represents a valuable resource of candidate interactions that can be used to further our insight into cell polarization.

In Vitro-Pooled shRNA Screening to Identify Determinants of Radiosensitivity

Short hairpin RNA (shRNA)-pooled screening is a valuable and cost-effective tool for assaying the contribution of individual genes to cell viability and proliferation on a genomic scale. Here we describe the key considerations for the design and execution of a pooled shRNA screen to identify determinants of radiosensitivity.

Network analysis of gene essentiality in functional genomics experiments

Many genomic techniques have been developed to study gene essentiality genome-wide, such as CRISPR and shRNA screens. Our analyses of public CRISPR screens suggest protein interaction networks, when integrated with gene expression or histone marks, are highly predictive of gene essentiality. Meanwhile, the quality of CRISPR and shRNA screen results can be significantly enhanced through network neighbor information. We also found network neighbor information to be very informative on prioritizing ChIP-seq target genes and survival indicator genes from tumor profiling. Thus, our study provides a general method for gene essentiality analysis in functional genomic experiments ( http://nest.dfci.harvard.edu ).

A disease module in the interactome explains disease heterogeneity, drug response and captures novel pathways and genes in asthma

Recent advances in genetics have spurred rapid progress towards the systematic identification of genes involved in complex diseases. Still, the detailed understanding of the molecular and physiological mechanisms through which these genes affect disease phenotypes remains a major challenge. Here, we identify the asthma disease module, i.e. the local neighborhood of the interactome whose perturbation is associated with asthma, and validate it for functional and pathophysiological relevance, using both computational and experimental approaches. We find that the asthma disease module is enriched with modest GWAS P-values against the background of random variation, and with differentially expressed genes from normal and asthmatic fibroblast cells treated with an asthma-specific drug. The asthma module also contains immune response mechanisms that are shared with other immune-related disease modules. Further, using diverse omics (genomics, gene-expression, drug response) data, we identify the GAB1 signaling pathway as an important novel modulator in asthma. The wiring diagram of the uncovered asthma module suggests a relatively close link between GAB1 and glucocorticoids (GCs), which we experimentally validate, observing an increase in the level of GAB1 after GC treatment in BEAS-2B bronchial epithelial cells. The siRNA knockdown of GAB1 in the BEAS-2B cell line resulted in a decrease in the NFkB level, suggesting a novel regulatory path of the pro-inflammatory factor NFkB by GAB1 in asthma.

Revealing molecular mechanisms by integrating high-dimensional functional screens with protein interaction data

Functional genomics screens using multi-parametric assays are powerful approaches for identifying genes involved in particular cellular processes. However, they suffer from problems like noise, and often provide little insight into molecular mechanisms. A bottleneck for addressing these issues is the lack of computational methods for the systematic integration of multi-parametric phenotypic datasets with molecular interactions. Here, we present Integrative Multi Profile Analysis of Cellular Traits (IMPACT). The main goal of IMPACT is to identify the most consistent phenotypic profile among interacting genes. This approach utilizes two types of external information: sets of related genes (IMPACT-sets) and network information (IMPACT-modules). Based on the notion that interacting genes are more likely to be involved in similar functions than non-interacting genes, this data is used as a prior to inform the filtering of phenotypic profiles that are similar among interacting genes. IMPACT-sets selects the most frequent profile among a set of related genes. IMPACT-modules identifies sub-networks containing genes with similar phenotype profiles. The statistical significance of these selections is subsequently quantified via permutations of the data. IMPACT (1) handles multiple profiles per gene, (2) rescues genes with weak phenotypes and (3) accounts for multiple biases e.g. caused by the network topology. Application to a genome-wide RNAi screen on endocytosis showed that IMPACT improved the recovery of known endocytosis-related genes, decreased off-target effects, and detected consistent phenotypes. Those findings were confirmed by rescreening 468 genes. Additionally we validated an unexpected influence of the IGF-receptor on EGF-endocytosis. IMPACT facilitates the selection of high-quality phenotypic profiles using different types of independent information, thereby supporting the molecular interpretation of functional screens.

Genome-scale functional genomics screens are important tools for investigating the function of genes. Technological progress allows for the simultaneous measurement of multiple parameters quantifying the response of cells to gene perturbations such as RNA interference. Such multi-dimensional screens provide rich data, but there is a lack of computational methods for interpreting these complex measurements. We have developed two computational methods that combine the data from multi-dimensional functional genomics screens with protein interaction information. These methods search for phenotype patterns that are consistent among interacting genes. Thereby, we could reduce the noise in the data and facilitate the mechanistic interpretation of the findings. The performance of the methods was demonstrated through application to a genome-wide screen studying endocytosis. Subsequent experimental validation demonstrated the improved detection of phenotypic profiles through the use of protein interaction data. Our analysis revealed unexpected roles of specific network modules and protein complexes with respect to endocytosis. Detailed follow-up experiments investigating the dynamics of endocytosis uncovered crosstalk between the cancer-related EGF and IGF pathways with so far unknown effects on endocytosis and cargo trafficking.

Integrative approaches for finding modular structure in biological networks

A central goal of systems biology is to elucidate the structural and functional architecture of the cell. To this end, large and complex networks of molecular interactions are being rapidly generated for humans and model organisms. A recent focus of bioinformatics research has been to integrate these networks with each other and with diverse molecular profiles to identify sets of molecules and interactions that participate in a common biological function - that is, 'modules'. Here, we classify such integrative approaches into four broad categories, describe their bioinformatic principles and review their applications.

Using random walks to identify cancer-associated modules in expression data

BACKGROUND

The etiology of cancer involves a complex series of genetic and environmental conditions. To better represent and study the intricate genetics of cancer onset and progression, we construct a network of biological interactions to search for groups of genes that compose cancer-related modules. Three cancer expression datasets are investigated to prioritize genes and interactions associated with cancer outcomes. Using a graph-based approach to search for communities of phenotype-related genes in microarray data, we find modules of genes associated with cancer phenotypes in a weighted interaction network.

RESULTS

We implement Walktrap, a random-walk-based community detection algorithm, to identify biological modules predisposing to tumor growth in 22 hepatocellular carcinoma samples (GSE14520), adenoma development in 32 colorectal cancer samples (GSE8671), and prognosis in 198 breast cancer patients (GSE7390). For each study, we find the best scoring partitions under a maximum cluster size of 200 nodes. Significant modules highlight groups of genes that are functionally related to cancer and show promise as therapeutic targets; these include interactions among transcription factors (SPIB, RPS6KA2 and RPS6KA6), cell-cycle regulatory genes (BRSK1, WEE1 and CDC25C), modulators of the cell-cycle and proliferation (CBLC and IRS2) and genes that regulate and participate in the map-kinase pathway (MAPK9, DUSP1, DUSP9, RIPK2). To assess the performance of Walktrap to find genomic modules (Walktrap-GM), we evaluate our results against other tools recently developed to discover disease modules in biological networks. Compared with other highly cited module-finding tools, jActiveModules and Matisse, Walktrap-GM shows strong performance in the discovery of modules enriched with known cancer genes.

CONCLUSIONS

These results demonstrate that the Walktrap-GM algorithm identifies modules significantly enriched with cancer genes, their joint effects and promising candidate genes. The approach performs well when evaluated against similar tools and smaller overall module size allows for more specific functional annotation and facilitates the interpretation of these modules.

Integrative approaches for predicting protein function and prioritizing genes for complex phenotypes using protein interaction networks

With the rapid development of biotechnologies, many types of biological data including molecular networks are now available. However, to obtain a more complete understanding of a biological system, the integration of molecular networks with other data, such as molecular sequences, protein domains and gene expression profiles, is needed. A key to the use of networks in biological studies is the definition of similarity among proteins over the networks. Here, we review applications of similarity measures over networks with a special focus on the following four problems: (i) predicting protein functions, (ii) prioritizing genes related to a phenotype given a set of seed genes that have been shown to be related to the phenotype, (iii) prioritizing genes related to a phenotype by integrating gene expression profiles and networks and (iv) identification of false positives and false negatives from RNAi experiments. Diffusion kernels are demonstrated to give superior performance in all these tasks, leading to the suggestion that diffusion kernels should be the primary choice for a network similarity metric over other similarity measures such as direct neighbors and shortest path distance.

Structure and dynamics of molecular networks: a novel paradigm of drug discovery: a comprehensive review

Despite considerable progress in genome- and proteome-based high-throughput screening methods and in rational drug design, the increase in approved drugs in the past decade did not match the increase of drug development costs. Network description and analysis not only give a systems-level understanding of drug action and disease complexity, but can also help to improve the efficiency of drug design. We give a comprehensive assessment of the analytical tools of network topology and dynamics. The state-of-the-art use of chemical similarity, protein structure, protein-protein interaction, signaling, genetic interaction and metabolic networks in the discovery of drug targets is summarized. We propose that network targeting follows two basic strategies. The "central hit strategy" selectively targets central nodes/edges of the flexible networks of infectious agents or cancer cells to kill them. The "network influence strategy" works against other diseases, where an efficient reconfiguration of rigid networks needs to be achieved by targeting the neighbors of central nodes/edges. It is shown how network techniques can help in the identification of single-target, edgetic, multi-target and allo-network drug target candidates. We review the recent boom in network methods helping hit identification, lead selection optimizing drug efficacy, as well as minimizing side-effects and drug toxicity. Successful network-based drug development strategies are shown through the examples of infections, cancer, metabolic diseases, neurodegenerative diseases and aging. Summarizing >1200 references we suggest an optimized protocol of network-aided drug development, and provide a list of systems-level hallmarks of drug quality. Finally, we highlight network-related drug development trends helping to achieve these hallmarks by a cohesive, global approach.

An analysis and validation pipeline for large-scale RNAi-based screens

Large-scale RNAi-based screens are a major technology, but require adequate prioritization and validation of candidate genes from the primary screen. In this work, we performed a large-scale pooled shRNA screen in mouse embryonic stem cells (ESCs) to discover genes associated with oxidative stress resistance and found several candidates. We then developed a bioinformatics pipeline to prioritize these candidates incorporating effect sizes, functional enrichment analysis, interaction networks and gene expression information. To validate candidates, we mixed normal cells with cells expressing the shRNA coupled to a fluorescent protein, which allows control cells to be used as an internal standard, and thus we could detect shRNAs with subtle effects. Although we did not identify genes associated with oxidative stress resistance, as a proof-of-concept of our pipeline we demonstrate a detrimental role of Edd1 silencing in ESC growth. Our methods may be useful for candidate gene prioritization of large-scale RNAi-based screens.

Integrative analysis of a cross-loci regulation network identifies App as a gene regulating insulin secretion from pancreatic islets

Complex diseases result from molecular changes induced by multiple genetic factors and the environment. To derive a systems view of how genetic loci interact in the context of tissue-specific molecular networks, we constructed an F2 intercross comprised of >500 mice from diabetes-resistant (B6) and diabetes-susceptible (BTBR) mouse strains made genetically obese by the Leptin^ob/ob mutation (Lep^ob). High-density genotypes, diabetes-related clinical traits, and whole-transcriptome expression profiling in five tissues (white adipose, liver, pancreatic islets, hypothalamus, and gastrocnemius muscle) were determined for all mice. We performed an integrative analysis to investigate the inter-relationship among genetic factors, expression traits, and plasma insulin, a hallmark diabetes trait. Among five tissues under study, there are extensive protein–protein interactions between genes responding to different loci in adipose and pancreatic islets that potentially jointly participated in the regulation of plasma insulin. We developed a novel ranking scheme based on cross-loci protein-protein network topology and gene expression to assess each gene's potential to regulate plasma insulin. Unique candidate genes were identified in adipose tissue and islets. In islets, the Alzheimer's gene App was identified as a top candidate regulator. Islets from 17-week-old, but not 10-week-old, App knockout mice showed increased insulin secretion in response to glucose or a membrane-permeant cAMP analog, in agreement with the predictions of the network model. Our result provides a novel hypothesis on the mechanism for the connection between two aging-related diseases: Alzheimer's disease and type 2 diabetes.

Alzheimer's disease and type 2 diabetes are two common aging-related diseases. Numerous studies have shown that the two diseases are associated. However, the mechanisms of such connection are not clear. Both diseases are complex diseases that are induced by multiple genetic factors and the environment. To understand the molecular network regulated by complex genetic factors causing type 2 diabetes, we constructed an F2 intercross comprised of >500 mice from diabetes-resistant and diabetic mouse strains. We measured genotypes, clinical traits, and expression profiling in five tissues for each mouse. We then performed an integrative analysis to investigate the inter-relationship among genetic factors, expression traits, and plasma insulin, a hallmark diabetes trait, and developed a novel method for inferring key regulators for regulating plasma insulin. In islets, the Alzheimer's gene App was identified as a top candidate regulator. Islets from 17-week-old, but not 10-week-old, App knockout mice showed increased insulin secretion in response to glucose, in agreement with the predictions of the network model. Our result provides a novel hypothesis on the mechanism for the connection between two aging-related diseases: Alzheimer's disease and type 2 diabetes.

Large-scale signaling network reconstruction

Reconstructing the topology of a signaling network by means of RNA interference (RNAi) technology is an underdetermined problem especially when a single gene in the network is knocked down or observed. In addition, the exponential search space limits the existing methods to small signaling networks of size 10-15 genes. In this paper, we propose integrating RNAi data with a reference physical interaction network. We formulate the problem of signaling network reconstruction as finding the minimum number of edit operations on a given reference network. The edit operations transform the reference network to a network that satisfies the RNAi observations. We show that using a reference network does not simplify the computational complexity of the problem. Therefore, we propose two methods which provide near optimal results and can scale well for reconstructing networks up to hundreds of components. We validate the proposed methods on synthetic and real data sets. Comparison with the state of the art on real signaling networks shows that the proposed methodology can scale better and generates biologically significant results.

atBioNet--an integrated network analysis tool for genomics and biomarker discovery

Background

Large amounts of mammalian protein-protein interaction (PPI) data have been generated and are available for public use. From a systems biology perspective, Proteins/genes interactions encode the key mechanisms distinguishing disease and health, and such mechanisms can be uncovered through network analysis. An effective network analysis tool should integrate different content-specific PPI databases into a comprehensive network format with a user-friendly platform to identify key functional modules/pathways and the underlying mechanisms of disease and toxicity.

Results

atBioNet integrates seven publicly available PPI databases into a network-specific knowledge base. Knowledge expansion is achieved by expanding a user supplied proteins/genes list with interactions from its integrated PPI network. The statistically significant functional modules are determined by applying a fast network-clustering algorithm (SCAN: a Structural Clustering Algorithm for Networks). The functional modules can be visualized either separately or together in the context of the whole network. Integration of pathway information enables enrichment analysis and assessment of the biological function of modules. Three case studies are presented using publicly available disease gene signatures as a basis to discover new biomarkers for acute leukemia, systemic lupus erythematosus, and breast cancer. The results demonstrated that atBioNet can not only identify functional modules and pathways related to the studied diseases, but this information can also be used to hypothesize novel biomarkers for future analysis.

Conclusion

atBioNet is a free web-based network analysis tool that provides a systematic insight into proteins/genes interactions through examining significant functional modules. The identified functional modules are useful for determining underlying mechanisms of disease and biomarker discovery. It can be accessed at: http://www.fda.gov/ScienceResearch/BioinformaticsTools/ucm285284.htm.

SteinerNet: a web server for integrating 'omic' data to discover hidden components of response pathways

High-throughput technologies including transcriptional profiling, proteomics and reverse genetics screens provide detailed molecular descriptions of cellular responses to perturbations. However, it is difficult to integrate these diverse data to reconstruct biologically meaningful signaling networks. Previously, we have established a framework for integrating transcriptional, proteomic and interactome data by searching for the solution to the prize-collecting Steiner tree problem. Here, we present a web server, SteinerNet, to make this method available in a user-friendly format for a broad range of users with data from any species. At a minimum, a user only needs to provide a set of experimentally detected proteins and/or genes and the server will search for connections among these data from the provided interactomes for yeast, human, mouse, Drosophila melanogaster and Caenorhabditis elegans. More advanced users can upload their own interactome data as well. The server provides interactive visualization of the resulting optimal network and downloadable files detailing the analysis and results. We believe that SteinerNet will be useful for researchers who would like to integrate their high-throughput data for a specific condition or cellular response and to find biologically meaningful pathways. SteinerNet is accessible at http://fraenkel.mit.edu/steinernet.

Identification and characterization of lipopolysaccharide binding protein (LBP) as an estrogen receptor α specific serum biomarker

Estrogen Receptor a (ERα) and Estrogen Receptor β (ERβ) are steroid nuclear receptors that transduce estrogen signaling to control diverse physiological processes linked to reproduction, bone remodeling, behavior, immune response and endocrine-related diseases. In order to differentiate between ERα and ERβ mediated effects in vivo, ER subtype selective biomarkers are essential. We utilized ERα knockout (AERKO) and ERβ knockout (BERKO) mouse liver RNA and genome wide profiling to identify novel ERα selective serum biomarker candidates. Results from the gene array experiments were validated using real-time RT-PCR and subsequent ELISA's to demonstrate changes in serum proteins. Here we present data that Lipopolysacharide Binding Protein (LBP) is a novel liver-derived ERα selective biomarker that can be measured in serum.

Delineation of key regulatory elements identifies points of vulnerability in the mitogen-activated signaling network

Drug development efforts against cancer are often hampered by the complex properties of signaling networks. Here we combined the results of an RNAi screen targeting the cellular signaling machinery, with graph theoretical analysis to extract the core modules that process both mitogenic and oncogenic signals to drive cell cycle progression. These modules encapsulated mechanisms for coordinating seamless transition of cells through the individual cell cycle stages and, importantly, were functionally conserved across different cancer cell types. Further analysis also enabled extraction of the core signaling axes that progressively guide commitment of cells to the division cycle. Importantly, pharmacological targeting of the least redundant nodes in these axes yielded a synergistic disruption of the cell cycle in a tissue-type-independent manner. Thus, the core elements that regulate temporally distinct stages of the cell cycle provide attractive targets for the development of multi-module-based chemotherapeutic strategies.

ResponseNet: revealing signaling and regulatory networks linking genetic and transcriptomic screening data

Cellular response to stimuli is typically complex and involves both regulatory and metabolic processes. Large-scale experimental efforts to identify components of these processes often comprise of genetic screening and transcriptomic profiling assays. We previously established that in yeast genetic screens tend to identify response regulators, while transcriptomic profiling assays tend to identify components of metabolic processes. ResponseNet is a network-optimization approach that integrates the results from these assays with data of known molecular interactions. Specifically, ResponseNet identifies a high-probability sub-network, composed of signaling and regulatory molecular interaction paths, through which putative response regulators may lead to the measured transcriptomic changes. Computationally, this is achieved by formulating a minimum-cost flow optimization problem and solving it efficiently using linear programming tools. The ResponseNet web server offers a simple interface for applying ResponseNet. Users can upload weighted lists of proteins and genes and obtain a sparse, weighted, molecular interaction sub-network connecting their data. The predicted sub-network and its gene ontology enrichment analysis are presented graphically or as text. Consequently, the ResponseNet web server enables researchers that were previously limited to separate analysis of their distinct, large-scale experiments, to meaningfully integrate their data and substantially expand their understanding of the underlying cellular response. ResponseNet is available at http://bioinfo.bgu.ac.il/respnet.

A protein network-guided screen for cell cycle regulators in Drosophila

Background

Large-scale RNAi-based screens are playing a critical role in defining sets of genes that regulate specific cellular processes. Numerous screens have been completed and in some cases more than one screen has examined the same cellular process, enabling a direct comparison of the genes identified in separate screens. Surprisingly, the overlap observed between the results of similar screens is low, suggesting that RNAi screens have relatively high levels of false positives, false negatives, or both.

Results

We re-examined genes that were identified in two previous RNAi-based cell cycle screens to identify potential false positives and false negatives. We were able to confirm many of the originally observed phenotypes and to reveal many likely false positives. To identify potential false negatives from the previous screens, we used protein interaction networks to select genes for re-screening. We demonstrate cell cycle phenotypes for a significant number of these genes and show that the protein interaction network is an efficient predictor of new cell cycle regulators. Combining our results with the results of the previous screens identified a group of validated, high-confidence cell cycle/cell survival regulators. Examination of the subset of genes from this group that regulate the G1/S cell cycle transition revealed the presence of multiple members of three structurally related protein complexes: the eukaryotic translation initiation factor 3 (eIF3) complex, the COP9 signalosome, and the proteasome lid. Using a combinatorial RNAi approach, we show that while all three of these complexes are required for Cdk2/Cyclin E activity, the eIF3 complex is specifically required for some other step that limits the G1/S cell cycle transition.

Conclusions

Our results show that false positives and false negatives each play a significant role in the lack of overlap that is observed between similar large-scale RNAi-based screens. Our results also show that protein network data can be used to minimize false negatives and false positives and to more efficiently identify comprehensive sets of regulators for a process. Finally, our data provides a high confidence set of genes that are likely to play key roles in regulating the cell cycle or cell survival.

SVM classifier to predict genes important for self-renewal and pluripotency of mouse embryonic stem cells

Background

Mouse embryonic stem cells (mESCs) are derived from the inner cell mass of a developing blastocyst and can be cultured indefinitely in-vitro. Their distinct features are their ability to self-renew and to differentiate to all adult cell types. Genes that maintain mESCs self-renewal and pluripotency identity are of interest to stem cell biologists. Although significant steps have been made toward the identification and characterization of such genes, the list is still incomplete and controversial. For example, the overlap among candidate self-renewal and pluripotency genes across different RNAi screens is surprisingly small. Meanwhile, machine learning approaches have been used to analyze multi-dimensional experimental data and integrate results from many studies, yet they have not been applied to specifically tackle the task of predicting and classifying self-renewal and pluripotency gene membership.

Results

For this study we developed a classifier, a supervised machine learning framework for predicting self-renewal and pluripotency mESCs stemness membership genes (MSMG) using support vector machines (SVM). The data used to train the classifier was derived from mESCs-related studies using mRNA microarrays, measuring gene expression in various stages of early differentiation, as well as ChIP-seq studies applied to mESCs profiling genome-wide binding of key transcription factors, such as Nanog, Oct4, and Sox2, to the regulatory regions of other genes. Comparison to other classification methods using the leave-one-out cross-validation method was employed to evaluate the accuracy and generality of the classification. Finally, two sets of candidate genes from genome-wide RNA interference screens are used to test the generality and potential application of the classifier.

Conclusions

Our results reveal that an SVM approach can be useful for prioritizing genes for functional validation experiments and complement the analyses of high-throughput profiling experimental data in stem cell research.

Genomic screening with RNAi: results and challenges

RNA interference (RNAi) is an effective tool for genome-scale, high-throughput analysis of gene function. In the past five years, a number of genome-scale RNAi high-throughput screens (HTSs) have been done in both Drosophila and mammalian cultured cells to study diverse biological processes, including signal transduction, cancer biology, and host cell responses to infection. Results from these screens have led to the identification of new components of these processes and, importantly, have also provided insights into the complexity of biological systems, forcing new and innovative approaches to understanding functional networks in cells. Here, we review the main findings that have emerged from RNAi HTS and discuss technical issues that remain to be improved, in particular the verification of RNAi results and validation of their biological relevance. Furthermore, we discuss the importance of multiplexed and integrated experimental data analysis pipelines to RNAi HTS.

Functional genomic analysis of glioblastoma multiforme through short interfering RNA screening: a paradigm for therapeutic development

Glioblastoma multiforme (GBM) is a high-grade brain malignancy arising from astrocytes. Despite aggressive surgical approaches, optimized radiation therapy regimens, and the application of cytotoxic chemotherapies, the median survival of patients with GBM from time of diagnosis remains less than 15 months, having changed little in decades. Approaches that target genes and biological pathways responsible for tumorigenesis or potentiate the activity of current therapeutic modalities could improve treatment efficacy. In this regard, several genomic and proteomic strategies promise to impact significantly on the drug discovery process. High-throughput genome-wide screening with short interfering RNA (siRNA) is one strategy for systematically exploring possible therapeutically relevant targets in GBM. Statistical methods and protein-protein interaction network databases can also be applied to the screening data to explore the genes and pathways that underlie the pathological basis and development of GBM. In this study, we highlight several genome-wide siRNA screens and implement these experimental concepts in the T98G GBM cell line to uncover the genes and pathways that regulate GBM cell death and survival. These studies will ultimately influence the development of a new avenue of neurosurgical therapy by placing the drug discovery process in the context of the entire biological system.

HDL lipids and insulin resistance

There is renewed interest in high-density lipoproteins (HDLs) due to recent findings linking atherosclerosis to the formation of dysfunctional HDL. This article focuses on the universe of HDL lipids and their potential protective or proinflammatory roles in vascular disease and insulin resistance. HDL carries a wide array of lipids including sterols, triglycerides, fat-soluble vitamins, and a large number of phospholipids, including phosphatidylcholine, sphingomyelin, and ceramide with many biological functions. Ceramide has been implicated in the pathogenesis of insulin resistance and has many proinflammatory properties. In contrast, sphingosine-1-phosphate, which is transported mainly in HDL, has anti-inflammatory properties that may be atheroprotective and may account for some of the beneficial effects of HDL. However, the complexity of the HDL lipidome is only beginning to reveal itself. The emergence of new analytical technologies should rapidly increase our understanding of the function of HDL lipids and their role in disease states.

Blockade of sphingosine 1-phosphate receptor 2 signaling attenuates streptozotocin-induced apoptosis of pancreatic beta-cells

Sphingosine 1-phosphate (S1P) is a potent sphingolipid mediator that acts through five cognate G protein-coupled receptors (S1P(1)-S1P(5)) and regulates many critical biological processes. Recent studies indicated that S1P at nanomolar concentrations significantly reduces cytokine-induced apoptosis of pancreatic beta-cells in which genes for S1P(1)-S1P(4) are co-expressed. However, the S1P receptor subtype(s) involved in this effect remains to be clarified. In this study, we investigated the potential role of S1P(2) in streptozotocin (STZ)-induced apoptosis of pancreatic beta-cells and progression of diabetes. S1P(2)-deficient (S1P(2)(-/-)) mice displayed a greater survive ability, lower blood glucose levels, and smaller numbers of TUNEL-positive apoptotic beta-cells to administration of a high dose of STZ than wild-type (WT) mice. S1P(2)(-/-) mice showed higher insulin/glucose ratios (an index of relative insulin deficiency) and larger insulin-positive islet areas to administration of a low dose of STZ than WT mice. Moreover, administration of JTE-013, a S1P(2)-specific antagonist, to WT mice ameliorated STZ-induced blood glucose elevation and reduced the incidence of diabetes. Our findings indicate that blockade of S1P(2) signaling attenuates STZ-induced apoptosis of pancreatic beta-cells and decreases the incidence of diabetes.

Toward the dynamic interactome: it's about time

Dynamic molecular interactions play a central role in regulating the functioning of cells and organisms. The availability of experimentally determined large-scale cellular networks, along with other high-throughput experimental data sets that provide snapshots of biological systems at different times and conditions, is increasingly helpful in elucidating interaction dynamics. Here we review the beginnings of a new subfield within computational biology, one focused on the global inference and analysis of the dynamic interactome. This burgeoning research area, which entails a shift from static to dynamic network analysis, promises to be a major step forward in our ability to model and reason about cellular function and behavior.