ARACNe-AP: gene network reverse engineering through adaptive partitioning inference of mutual information

A framework for identifying dysregulated chromatin regulators as master regulators in human cancer

MOTIVATION

Chromatin regulators (CRs) are frequently dysregulated to reprogram the epigenetic landscape of the cancer genome. However, the underpinnings of the dysregulation of CRs and their downstream effectors remain to be elucidated.

RESULTS

Here, we designed an integrated framework based on multi-omics data to identify candidate master regulatory CRs affected by genomic alterations across eight cancer types in The Cancer Genome Atlas. Most of them showed consistent activated or repressed (i.e. oncogenic or tumor-suppressive) roles in cancer initiation and progression. In order to further explore the insight mechanism of the dysregulated CRs, we developed an R package ModReg based on differential connectivity to identify CRs as modulators of transcription factors (TFs) involved in tumorigenesis. Our analysis revealed that the connectivity between TFs and their target genes (TGs) tended to be disrupted in the patients who had a high expression of oncogenic CRs or low-expression of tumor-suppressive CRs. As a proof-of-principle study, 14 (82.4%) of the top-ranked 17 driver CRs in liver cancer were able to be validated by literature mining or experiments including shRNA knockdown and dCas9-based epigenetic editing. Moreover, we confirmed that CR SIRT7 physically interacted with TF NFE2L2, and positively modulated the transcriptional program of NFE2L2 by affecting ∼64% of its TGs.

AVAILABILITY AND IMPLEMENTATION

ModReg is freely accessible at http://cis.hku.hk/software/ModReg.tar.gz.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Linking aberrant chromatin features in chronic lymphocytic leukemia to transcription factor networks

In chronic lymphocytic leukemia (CLL), a diverse set of genetic mutations is embedded in a deregulated epigenetic landscape that drives cancerogenesis. To elucidate the role of aberrant chromatin features, we mapped DNA methylation, seven histone modifications, nucleosome positions, chromatin accessibility, binding of EBF1 and CTCF, as well as the transcriptome of B cells from CLL patients and healthy donors. A globally increased histone deacetylase activity was detected and half of the genome comprised transcriptionally downregulated partially DNA methylated domains demarcated by CTCF CLL samples displayed a H3K4me3 redistribution and nucleosome gain at promoters as well as changes of enhancer activity and enhancer linkage to target genes. A DNA binding motif analysis identified transcription factors that gained or lost binding in CLL at sites with aberrant chromatin features. These findings were integrated into a gene regulatory enhancer containing network enriched for B-cell receptor signaling pathway components. Our study predicts novel molecular links to targets of CLL therapies and provides a valuable resource for further studies on the epigenetic contribution to the disease.

VULCAN integrates ChIP-seq with patient-derived co-expression networks to identify GRHL2 as a key co-regulator of ERa at enhancers in breast cancer

BACKGROUND

VirtUaL ChIP-seq Analysis through Networks (VULCAN) infers regulatory interactions of transcription factors by overlaying networks generated from publicly available tumor expression data onto ChIP-seq data. We apply our method to dissect the regulation of estrogen receptor-alpha activation in breast cancer to identify potential co-regulators of the estrogen receptor's transcriptional response.

RESULTS

VULCAN analysis of estrogen receptor activation in breast cancer highlights the key components of the estrogen receptor complex alongside a novel interaction with GRHL2. We demonstrate that GRHL2 is recruited to a subset of estrogen receptor binding sites and regulates transcriptional output, as evidenced by changes in estrogen receptor-associated eRNA expression and stronger estrogen receptor binding at active enhancers after GRHL2 knockdown.

CONCLUSIONS

Our findings provide new insight into the role of GRHL2 in regulating eRNA transcription as part of estrogen receptor signaling. These results demonstrate VULCAN, available from Bioconductor, as a powerful predictive tool.

Comprehensive characterisation of compartment-specific long non-coding RNAs associated with pancreatic ductal adenocarcinoma

OBJECTIVE

Pancreatic ductal adenocarcinoma (PDA) is a highly metastatic disease with limited therapeutic options. Genome and transcriptome analyses have identified signalling pathways and cancer driver genes with implications in patient stratification and targeted therapy. However, these analyses were performed in bulk samples and focused on coding genes, which represent a small fraction of the genome.

DESIGN

We developed a computational framework to reconstruct the non-coding transcriptome from cross-sectional RNA-Seq, integrating somatic copy number alterations (SCNA), common germline variants associated to PDA risk and clinical outcome. We validated the results in an independent cohort of paired epithelial and stromal RNA-Seq derived from laser capture microdissected human pancreatic tumours, allowing us to annotate the compartment specificity of their expression. We employed systems and experimental biology approaches to interrogate the function of epithelial long non-coding RNAs (lncRNAs) associated with genetic traits and clinical outcome in PDA.

RESULTS

We generated a catalogue of PDA-associated lncRNAs. We showed that lncRNAs define molecular subtypes with biological and clinical significance. We identified lncRNAs in genomic regions with SCNA and single nucleotide polymorphisms associated with lifetime risk of PDA and associated with clinical outcome using genomic and clinical data in PDA. Systems biology and experimental functional analysis of two epithelial lncRNAs (LINC00673 and FAM83H-AS1) suggest they regulate the transcriptional profile of pancreatic tumour samples and PDA cell lines.

CONCLUSIONS

Our findings indicate that lncRNAs are associated with genetic marks of pancreatic cancer risk, contribute to the transcriptional regulation of neoplastic cells and provide an important resource to design functional studies of lncRNAs in PDA.

MR4Cancer: a web server prioritizing master regulators for cancer

MOTIVATION

During cancer stage transition, a master regulator (MR) refers to the key gene controlling cancer initiation and progression by orchestrating the associated target genes (termed as its regulon). Due to their inherent importance, MRs can serve as critical biomarkers for cancer diagnosis and prognosis, and therapeutic targets. However, it is challenging to infer key MRs that might explain gene expression profile changes between two groups due to lack of context-specific regulons, whose expression level can collectively reflect the activity of likely MRs. There is also a need to design an easy-to-use tool of MR identification for research community.

RESULTS

First, we generated cancer-specific regulons for 26 cancer types by analyzing high-throughput omics data from TCGA, and extracted noncancer-specific regulons from public databases. We subsequently developed a web server MR4Cancer, integrating the regulons with statistical inference to identify and prioritize MRs driving a phenotypic divergence of interest. Based on the input gene list (e.g. differentially expressed genes) or expression profile with two groups, MR4Cancer outputs ranked MRs by enrichment testing against the predefined regulons. Gene Ontology and canonical pathway analyses are also conducted to elucidate the function of likely MRs. Moreover, MR4Cancer provides dynamic network visualization for MR-target relations, and users can interactively interrogate the network to produce new hypotheses and high-quality figures for publication. Finally, the presented case studies highlighted the performance of MR4Cancer. We expect this user-friendly and powerful web tool will provide researchers novel insights into tumorigenesis and therapeutic intervention.

AVAILABILITY AND IMPLEMENTATION

http://cis.hku.hk/MR4Cancer.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Improving GRN re-construction by mining hidden regulatory signals

Inferring gene regulatory networks (GRNs) from gene expression data is an important but challenging issue in systems biology. Here, the authors propose a dictionary learning-based approach that aims to infer GRNs by globally mining regulatory signals, known or latent. Gene expression is often regulated by various regulatory factors, some of which are observed and some of which are latent. The authors assume that all regulators are unknown for a target gene and the expression of the target gene can be mapped into a regulatory space spanned by all the regulators. Specifically, the authors modify the dictionary learning model, k-SVD, according to the sparse property of GRNs for mining the regulatory signals. The recovered regulatory signals are then used as a pool of regulatory factors to calculate a confidence score for a given transcription factor regulating a target gene. The capability of recovering hidden regulatory signals was verified on simulated data. Comparative experiments for GRN inference between the proposed algorithm (OURM) and some state-of-the-art algorithms, e.g. GENIE3 and ARACNE, on real-world data sets show the superior performance of OURM in inferring GRNs: higher area under the receiver operating characteristic curves and area under the precision-recall curves.

Altered Gene Regulatory Networks Are Associated With the Transition From C3 to Crassulacean Acid Metabolism in Erycina (Oncidiinae: Orchidaceae)

Crassulacean acid metabolism (CAM) photosynthesis is a modification of the core C3 photosynthetic pathway that improves the ability of plants to assimilate carbon in water-limited environments. CAM plants fix CO2 mostly at night, when transpiration rates are low. All of the CAM pathway genes exist in ancestral C3 species, but the timing and magnitude of expression are greatly altered between C3 and CAM species. Understanding these regulatory changes is key to elucidating the mechanism by which CAM evolved from C3. Here, we use two closely related species in the Orchidaceae, Erycina pusilla (CAM) and Erycina crista-galli (C3), to conduct comparative transcriptomic analyses across multiple time points. Clustering of genes with expression variation across the diel cycle revealed some canonical CAM pathway genes similarly expressed in both species, regardless of photosynthetic pathway. However, gene network construction indicated that 149 gene families had significant differences in network connectivity and were further explored for these functional enrichments. Genes involved in light sensing and ABA signaling were some of the most differently connected genes between the C3 and CAM Erycina species, in agreement with the contrasting diel patterns of stomatal conductance in C3 and CAM plants. Our results suggest changes to transcriptional cascades are important for the transition from C3 to CAM photosynthesis in Erycina.

Intrinsic-overlapping co-expression module detection with application to Alzheimer's Disease

Genes interact with each other and may cause perturbation in the molecular pathways leading to complex diseases. Often, instead of any single gene, a subset of genes interact, forming a network, to share common biological functions. Such a subnetwork is called a functional module or motif. Identifying such modules and central key genes in them, that may be responsible for a disease, may help design patient-specific drugs. In this study, we consider the neurodegenerative Alzheimer's Disease (AD) and identify potentially responsible genes from functional motif analysis. We start from the hypothesis that central genes in genetic modules are more relevant to a disease that is under investigation and identify hub genes from the modules as potential marker genes. Motifs or modules are often non-exclusive or overlapping in nature. Moreover, they sometimes show intrinsic or hierarchical distributions with overlapping functional roles. To the best of our knowledge, no prior work handles both the situations in an integrated way. We propose a non-exclusive clustering approach, CluViaN (Clustering Via Network) that can detect intrinsic as well as overlapping modules from gene co-expression networks constructed using microarray expression profiles. We compare our method with existing methods to evaluate the quality of modules extracted. CluViaN reports the presence of intrinsic and overlapping motifs in different species not reported by any other research. We further apply our method to extract significant AD specific modules using CluViaN and rank them based the number of genes from a module involved in the disease pathways. Finally, top central genes are identified by topological analysis of the modules. We use two different AD phenotype data for experimentation. We observe that central genes, namely PSEN1, APP, NDUFB2, NDUFA1, UQCR10, PPP3R1 and a few more, play significant roles in the AD. Interestingly, our experiments also find a hub gene, PML, which has recently been reported to play a role in plasticity, circadian rhythms and the response to proteins which can cause neurodegenerative disorders. MUC4, another hub gene that we find experimentally is yet to be investigated for its potential role in AD. A software implementation of CluViaN in Java is available for download at https://sites.google.com/site/swarupnehu/publications/resources/CluViaN Software.rar.

Variability of Betweenness Centrality and Its Effect on Identifying Essential Genes

This paper begins to build a theoretical framework that would enable the pharmaceutical industry to use network complexity measures as a way to identify drug targets. The variability of a betweenness measure for a network node is examined through different methods of network perturbation. Our results indicate a robustness of betweenness centrality in the identification of target genes.

Single-Cell RNA Sequencing Analysis Reveals Sequential Cell Fate Transition during Human Spermatogenesis

Spermatogenesis generates mature male gametes and is critical for the proper transmission of genetic information between generations. However, the developmental landscapes of human spermatogenesis remain unknown. Here, we performed single-cell RNA sequencing (scRNA-seq) analysis for 2,854 testicular cells from donors with normal spermatogenesis and 174 testicular cells from one nonobstructive azoospermia (NOA) donor. A hierarchical model was established, which was characterized by the sequential and stepwise development of three spermatogonia subtypes, seven spermatocyte subtypes, and four spermatid subtypes. Further analysis identified several stage-specific marker genes of human germ cells, such as HMGA1, PIWIL4, TEX29, SCML1, and CCDC112. Moreover, we identified altered gene expression patterns in the testicular somatic cells of one NOA patient via scRNA-seq analysis, paving the way for further diagnosis of male infertility. Our work allows for the reconstruction of transcriptional programs inherent to sequential cell fate transition during human spermatogenesis and has implications for deciphering male-related reproductive disorders.

In Search of Biomarkers for Pathogenesis and Control of Leishmaniasis by Global Analyses of Leishmania-Infected Macrophages

Leishmaniasis is a vector-borne, neglected tropical disease with a worldwide distribution that can present in a variety of clinical forms, depending on the parasite species and host genetic background. The pathogenesis of this disease remains far from being elucidated because the involvement of a complex immune response orchestrated by host cells significantly affects the clinical outcome. Among these cells, macrophages are the main host cells, produce cytokines and chemokines, thereby triggering events that contribute to the mediation of the host immune response and, subsequently, to the establishment of infection or, alternatively, disease control. There has been relatively limited commercial interest in developing new pharmaceutical compounds to treat leishmaniasis. Moreover, advances in the understanding of the underlying biology of Leishmania spp. have not translated into the development of effective new chemotherapeutic compounds. As a result, biomarkers as surrogate disease endpoints present several potential advantages to be used in the identification of targets capable of facilitating therapeutic interventions considered to ameliorate disease outcome. More recently, large-scale genomic and proteomic analyses have allowed the identification and characterization of the pathways involved in the infection process in both parasites and the host, and these analyses have been shown to be more effective than studying individual molecules to elucidate disease pathogenesis. RNA-seq and proteomics are large-scale approaches that characterize genes or proteins in a given cell line, tissue, or organism to provide a global and more integrated view of the myriad biological processes that occur within a cell than focusing on an individual gene or protein. Bioinformatics provides us with the means to computationally analyze and integrate the large volumes of data generated by high-throughput sequencing approaches. The integration of genomic expression and proteomic data offers a rich multi-dimensional analysis, despite the inherent technical and statistical challenges. We propose that these types of global analyses facilitate the identification, among a large number of genes and proteins, those that hold potential as biomarkers. The present review focuses on large-scale studies that have identified and evaluated relevant biomarkers in macrophages in response to Leishmania infection.

A Human Adult Stem Cell Signature Marks Aggressive Variants across Epithelial Cancers

Cancer progression to an aggressive phenotype often co-opts aspects of stem cell biology. Here, we developed gene signatures for normal human stem cell populations to understand the relationship between epithelial cancers and stem cell transcriptional programs. Using a pan-cancer approach, we reveal that aggressive epithelial cancers are enriched for a transcriptional signature shared by epithelial adult stem cells. The adult stem cell signature selected for epithelial cancers with worse overall survival and alterations of oncogenic drivers. Lethal small cell neuroendocrine lung, prostate, and bladder cancers transcriptionally converged onto the adult stem cell signature and not other stem cell signatures tested. We found that DNA methyltransferase expression correlated with adult stem cell signature status and was enriched in small cell neuroendocrine cancers. DNA methylation analysis uncovered a shared epigenomic profile between small cell neuroendocrine cancers. These pan-cancer findings establish a molecular link between human adult stem cells and aggressive epithelial cancers.

The Spring of Systems Biology-Driven Breeding

Genetics and molecular biology have contributed to the development of rationalized plant breeding programs. Recent developments in both high-throughput experimental analyses of biological systems and in silico data processing offer the possibility to address the whole gene regulatory network (GRN) controlling a given trait. GRN models can be applied to identify topological features helping to shortlist potential candidate genes for breeding purposes. Time-series data sets can be used to support dynamic modelling of the network. This will enable a deeper comprehension of network behaviour and the identification of the few elements to be genetically rewired to push the system towards a modified phenotype of interest. This paves the way to design more efficient, systems biology-based breeding strategies.

PREMER: A Tool to Infer Biological Networks

Inferring the structure of unknown cellular networks is a main challenge in computational biology. Data-driven approaches based on information theory can determine the existence of interactions among network nodes automatically. However, the elucidation of certain features-such as distinguishing between direct and indirect interactions or determining the direction of a causal link-requires estimating information-theoretic quantities in a multidimensional space. This can be a computationally demanding task, which acts as a bottleneck for the application of elaborate algorithms to large-scale network inference problems. The computational cost of such calculations can be alleviated by the use of compiled programs and parallelization. To this end, we have developed PREMER (Parallel Reverse Engineering with Mutual information & Entropy Reduction), a software toolbox that can run in parallel and sequential environments. It uses information theoretic criteria to recover network topology and determine the strength and causality of interactions, and allows incorporating prior knowledge, imputing missing data, and correcting outliers. PREMER is a free, open source software tool that does not require any commercial software. Its core algorithms are programmed in FORTRAN 90 and implement OpenMP directives. It has user interfaces in Python and MATLAB/Octave, and runs on Windows, Linux, and OSX (https://sites.google.com/site/premertoolbox/).

Structure Optimization for Large Gene Networks Based on Greedy Strategy

In the last few years, gene networks have become one of most important tools to model biological processes. Among other utilities, these networks visually show biological relationships between genes. However, due to the large amount of the currently generated genetic data, their size has grown to the point of being unmanageable. To solve this problem, it is possible to use computational approaches, such as heuristics-based methods, to analyze and optimize gene network's structure by pruning irrelevant relationships. In this paper we present a new method, called GeSOp, to optimize large gene network structures. The method is able to perform a considerably prune of the irrelevant relationships comprising the input network. To do so, the method is based on a greedy heuristic to obtain the most relevant subnetwork. The performance of our method was tested by means of two experiments on gene networks obtained from different organisms. The first experiment shows how GeSOp is able not only to carry out a significant reduction in the size of the network, but also to maintain the biological information ratio. In the second experiment, the ability to improve the biological indicators of the network is checked. Hence, the results presented show that GeSOp is a reliable method to optimize and improve the structure of large gene networks.

MICMIC: identification of DNA methylation of distal regulatory regions with causal effects on tumorigenesis

Aberrant promoter methylation is a common mechanism for tumor suppressor inactivation in cancer. We develop a set of tools to identify genome-wide DNA methylation in distal regions with causal effect on tumorigenesis called MICMIC. Many predictions are directly validated by dCas9-based epigenetic editing to support the accuracy and efficiency of our tool. Oncogenic and lineage-specific transcription factors are shown to aberrantly shape the methylation landscape by modifying tumor-subtype core regulatory circuitry. Notably, the gene regulatory networks orchestrated by enhancer methylation across different cancer types are seen to converge on a common architecture. MICMIC is available on https://github.com/ZhangJlab/MICMIC .

Graph complexity analysis identifies an ETV5 tumor-specific network in human and murine low-grade glioma

Conventional differential expression analyses have been successfully employed to identify genes whose levels change across experimental conditions. One limitation of this approach is the inability to discover central regulators that control gene expression networks. In addition, while methods for identifying central nodes in a network are widely implemented, the bioinformatics validation process and the theoretical error estimates that reflect the uncertainty in each step of the analysis are rarely considered. Using the betweenness centrality measure, we identified Etv5 as a potential tissue-level regulator in murine neurofibromatosis type 1 (Nf1) low-grade brain tumors (optic gliomas). As such, the expression of Etv5 and Etv5 target genes were increased in multiple independently-generated mouse optic glioma models relative to non-neoplastic (normal healthy) optic nerves, as well as in the cognate human tumors (pilocytic astrocytoma) relative to normal human brain. Importantly, differential Etv5 and Etv5 network expression was not directly the result of Nf1 gene dysfunction in specific cell types, but rather reflects a property of the tumor as an aggregate tissue. Moreover, this differential Etv5 expression was independently validated at the RNA and protein levels. Taken together, the combined use of network analysis, differential RNA expression findings, and experimental validation highlights the potential of the computational network approach to provide new insights into tumor biology.

Cross-Cohort Analysis Identifies a TEAD4-MYCN Positive Feedback Loop as the Core Regulatory Element of High-Risk Neuroblastoma

High-risk neuroblastomas show a paucity of recurrent somatic mutations at diagnosis. As a result, the molecular basis for this aggressive phenotype remains elusive. Recent progress in regulatory network analysis helped us elucidate disease-driving mechanisms downstream of genomic alterations, including recurrent chromosomal alterations. Our analysis identified three molecular subtypes of high-risk neuroblastomas, consistent with chromosomal alterations, and identified subtype-specific master regulator proteins that were conserved across independent cohorts. A 10-protein transcriptional module-centered around a TEAD4-MYCN positive feedback loop-emerged as the regulatory driver of the high-risk subtype associated with MYCN amplification. Silencing of either gene collapsed MYCN-amplified (MYCNAmp) neuroblastoma transcriptional hallmarks and abrogated viability in vitro and in vivo Consistently, TEAD4 emerged as a robust prognostic marker of poor survival, with activity independent of the canonical Hippo pathway transcriptional coactivators YAP and TAZ. These results suggest novel therapeutic strategies for the large subset of MYCN-deregulated neuroblastomas.Significance: Despite progress in understanding of neuroblastoma genetics, little progress has been made toward personalized treatment. Here, we present a framework to determine the downstream effectors of the genetic alterations sustaining neuroblastoma subtypes, which can be easily extended to other tumor types. We show the critical effect of disrupting a 10-protein module centered around a YAP/TAZ-independent TEAD4-MYCN positive feedback loop in MYCNAmp neuroblastomas, nominating TEAD4 as a novel candidate for therapeutic intervention. Cancer Discov; 8(5); 582-99. ©2018 AACR.This article is highlighted in the In This Issue feature, p. 517.

Molecular Characterization and Clinical Relevance of Metabolic Expression Subtypes in Human Cancers

Metabolic reprogramming provides critical information for clinical oncology. Using molecular data of 9,125 patient samples from The Cancer Genome Atlas, we identified tumor subtypes in 33 cancer types based on mRNA expression patterns of seven major metabolic processes and assessed their clinical relevance. Our metabolic expression subtypes correlated extensively with clinical outcome: subtypes with upregulated carbohydrate, nucleotide, and vitamin/cofactor metabolism most consistently correlated with worse prognosis, whereas subtypes with upregulated lipid metabolism showed the opposite. Metabolic subtypes correlated with diverse somatic drivers but exhibited effects convergent on cancer hallmark pathways and were modulated by highly recurrent master regulators across cancer types. As a proof-of-concept example, we demonstrated that knockdown of SNAI1 or RUNX1-master regulators of carbohydrate metabolic subtypes-modulates metabolic activity and drug sensitivity. Our study provides a system-level view of metabolic heterogeneity within and across cancer types and identifies pathway cross-talk, suggesting related prognostic, therapeutic, and predictive utility.

Systems-epigenomics inference of transcription factor activity implicates aryl-hydrocarbon-receptor inactivation as a key event in lung cancer development

Background

Diverse molecular alterations associated with smoking in normal and precursor lung cancer cells have been reported, yet their role in lung cancer etiology remains unclear. A prominent example is hypomethylation of the aryl hydrocarbon-receptor repressor (AHRR) locus, which is observed in blood and squamous epithelial cells of smokers, but not in lung cancer.

Results

Using a novel systems-epigenomics algorithm, called SEPIRA, which leverages the power of a large RNA-sequencing expression compendium to infer regulatory activity from messenger RNA expression or DNA methylation (DNAm) profiles, we infer the landscape of binding activity of lung-specific transcription factors (TFs) in lung carcinogenesis. We show that lung-specific TFs become preferentially inactivated in lung cancer and precursor lung cancer lesions and further demonstrate that these results can be derived using only DNAm data. We identify subsets of TFs which become inactivated in precursor cells. Among these regulatory factors, we identify AHR, the aryl hydrocarbon-receptor which controls a healthy immune response in the lung epithelium and whose repressor, AHRR, has recently been implicated in smoking-mediated lung cancer. In addition, we identify FOXJ1, a TF which promotes growth of airway cilia and effective clearance of the lung airway epithelium from carcinogens.

Conclusions

We identify TFs, such as AHR, which become inactivated in the earliest stages of lung cancer and which, unlike AHRR hypomethylation, are also inactivated in lung cancer itself. The novel systems-epigenomics algorithm SEPIRA will be useful to the wider epigenome-wide association study community as a means of inferring regulatory activity.

Electronic supplementary material

The online version of this article (doi:10.1186/s13059-017-1366-0) contains supplementary material, which is available to authorized users.

Accelerated parallel algorithm for gene network reverse engineering

Background

The Algorithm for the Reconstruction of Accurate Cellular Networks (ARACNE) represents one of the most effective tools to reconstruct gene regulatory networks from large-scale molecular profile datasets. However, previous implementations require intensive computing resources and, in some cases, restrict the number of samples that can be used. These issues can be addressed elegantly in a GPU computing framework, where repeated mathematical computation can be done efficiently, but requires extensive redesign to apply parallel computing techniques to the original serial algorithm, involving detailed optimization efforts based on a deep understanding of both hardware and software architecture.

Result

Here, we present an accelerated parallel implementation of ARACNE (GPU-ARACNE). By taking advantage of multi-level parallelism and the Compute Unified Device Architecture (CUDA) parallel kernel-call library, GPU-ARACNE successfully parallelizes a serial algorithm and simplifies the user experience from multi-step operations to one step. Using public datasets on comparable hardware configurations, we showed that GPU-ARACNE is faster than previous implementations and is able to reconstruct equally valid gene regulatory networks.

Conclusion

Given that previous versions of ARACNE are extremely resource demanding, either in computational time or in hardware investment, GPU-ARACNE is remarkably valuable for researchers who need to build complex regulatory networks from large expression datasets, but with limited budget on computational resources. In addition, our GPU-centered optimization of adaptive partitioning for Mutual Information (MI) estimation provides lessons that are applicable to other domains.

Electronic supplementary material

The online version of this article (doi:10.1186/s12918-017-0458-5) contains supplementary material, which is available to authorized users.

Master Transcriptional Regulators in Cancer: Discovery via Reverse Engineering Approaches and Subsequent Validation

Reverse engineering of transcriptional networks using gene expression data enables identification of genes that underpin the development and progression of different cancers. Methods to this end have been available for over a decade and, with a critical mass of transcriptomic data in the oncology arena having been reached, they are ever more applicable. Extensive and complex networks can be distilled into a small set of key master transcriptional regulators (MTR), genes that are very highly connected and have been shown to be involved in processes of known importance in disease. Interpreting and validating the results of standardized bioinformatic methods is of crucial importance in determining the inherent value of MTRs. In this review, we briefly describe how MTRs are identified and focus on providing an overview of how MTRs can and have been validated for use in clinical decision making in malignant diseases, along with serving as tractable therapeutic targets. Cancer Res; 77(9); 2186-90. ©2017 AACR.

Diurnal Transcriptome and Gene Network Represented through Sparse Modeling in Brachypodium distachyon

We report the comprehensive identification of periodic genes and their network inference, based on a gene co-expression analysis and an Auto-Regressive eXogenous (ARX) model with a group smoothly clipped absolute deviation (SCAD) method using a time-series transcriptome dataset in a model grass, Brachypodium distachyon. To reveal the diurnal changes in the transcriptome in B. distachyon, we performed RNA-seq analysis of its leaves sampled through a diurnal cycle of over 48 h at 4 h intervals using three biological replications, and identified 3,621 periodic genes through our wavelet analysis. The expression data are feasible to infer network sparsity based on ARX models. We found that genes involved in biological processes such as transcriptional regulation, protein degradation, and post-transcriptional modification and photosynthesis are significantly enriched in the periodic genes, suggesting that these processes might be regulated by circadian rhythm in B. distachyon. On the basis of the time-series expression patterns of the periodic genes, we constructed a chronological gene co-expression network and identified putative transcription factors encoding genes that might be involved in the time-specific regulatory transcriptional network. Moreover, we inferred a transcriptional network composed of the periodic genes in B. distachyon, aiming to identify genes associated with other genes through variable selection by grouping time points for each gene. Based on the ARX model with the group SCAD regularization using our time-series expression datasets of the periodic genes, we constructed gene networks and found that the networks represent typical scale-free structure. Our findings demonstrate that the diurnal changes in the transcriptome in B. distachyon leaves have a sparse network structure, demonstrating the spatiotemporal gene regulatory network over the cyclic phase transitions in B. distachyon diurnal growth.