Robust normalization and transformation techniques for constructing gene coexpression networks from RNA-seq data

Background Constructing gene coexpression networks is a powerful approach for analyzing high-throughput gene expression data towards module identification, gene function prediction, and disease-gene prioritization. While optimal workflows for constructing coexpression networks, including good choices for data pre-processing, normalization, and network transformation, have been developed for microarray-based expression data, such well-tested choices do not exist for RNA-seq data. Almost all studies that compare data processing and normalization methods for RNA-seq focus on the end goal of determining differential gene expression. Results Here, we present a comprehensive benchmarking and analysis of 36 different workflows, each with a unique set of normalization and network transformation methods, for constructing coexpression networks from RNA-seq datasets. We test these workflows on both large, homogenous datasets and small, heterogeneous datasets from various labs. We analyze the workflows in terms of aggregate performance, individual method choices, and the impact of multiple dataset experimental factors. Our results demonstrate that between-sample normalization has the biggest impact, with counts adjusted by size factors producing networks that most accurately recapitulate known tissue-naive and tissue-aware gene functional relationships. Conclusions Based on this work, we provide concrete recommendations on robust procedures for building an accurate coexpression network from an RNA-seq dataset. In addition, researchers can examine all the results in great detail at https://krishnanlab.github.io/RNAseq_coexpression to make appropriate choices for coexpression analysis based on the experimental factors of their RNA-seq dataset.

Constructing local cell-specific networks from single-cell data

Gene coexpression networks yield critical insights into biological processes, and single-cell RNA sequencing provides an opportunity to target inquiries at the cellular level. However, due to the sparsity and heterogeneity of transcript counts, it is challenging to construct accurate gene networks. We develop an approach, locCSN, that estimates cell-specific networks (CSNs) for each cell, preserving information about cellular heterogeneity that is lost with other approaches. LocCSN is based on a nonparametric investigation of the joint distribution of gene expression; hence it can readily detect nonlinear correlations, and it is more robust to distributional challenges. Although individual CSNs are estimated with considerable noise, average CSNs provide stable estimates of networks, which reveal gene communities better than traditional measures. Additionally, we propose downstream analysis methods using CSNs to utilize more fully the information contained within them. Repeated estimates of gene networks facilitate testing for differences in network structure between cell groups. Notably, with this approach, we can identify differential network genes, which typically do not differ in gene expression, but do differ in terms of the coexpression networks. These genes might help explain the etiology of disease. Finally, to further our understanding of autism spectrum disorder, we examine the evolution of gene networks in fetal brain cells and compare the CSNs of cells sampled from case and control subjects to reveal intriguing patterns in gene coexpression.

Transcriptomic Analysis Reveals Regulatory Networks for Osmotic Water Stress and Rewatering Response in the Leaves of Ginkgo biloba

To elucidate the transcriptomic regulation mechanisms that underlie the response of Ginkgo biloba to dehydration and rehydration, we used ginkgo saplings exposed to osmotically driven water stress and subsequent rewatering. When compared with a control group, 137, 1453, 1148, and 679 genes were differentially expressed in ginkgo leaves responding to 2, 6, 12, and 24 h of water deficit, and 796 and 1530 genes were differentially expressed responding to 24 and 48 h of rewatering. Upregulated genes participated in the biosynthesis of abscisic acid, eliminating reactive oxygen species (ROS), and biosynthesis of flavonoids and bilobalide, and downregulated genes were involved in water transport and cell wall enlargement in water stress-treated ginkgo leaves. Under rehydration conditions, the genes associated with water transport and cell wall enlargement were upregulated, and the genes that participated in eliminating ROS and the biosynthesis of flavonoids and bilobalide were downregulated in the leaves of G. biloba. Furthermore, the weighted gene coexpression networks were established and correlated with distinct water stress and rewatering time-point samples. Hub genes that act as key players in the networks were identified. Overall, these results indicate that the gene coexpression networks play essential roles in the transcriptional reconfiguration of ginkgo leaves in response to water stress and rewatering.

Generating weighted and thresholded gene coexpression networks using signed distance correlation

Even within well-studied organisms, many genes lack useful functional annotations. One way to generate such functional information is to infer biological relationships between genes or proteins, using a network of gene coexpression data that includes functional annotations. Signed distance correlation has proved useful for the construction of unweighted gene coexpression networks. However, transforming correlation values into unweighted networks may lead to a loss of important biological information related to the intensity of the correlation. Here introduce a principled method to construct \emph{weighted} gene coexpression networks using signed distance correlation. These networks contain weighted edges only between those pairs of genes whose correlation value is higher than a given threshold. We analyse data from different organisms and find that networks generated with our method based on signed distance correlation are more stable and capture more biological information compared to networks obtained from Pearson correlation. Moreover, we show that signed distance correlation networks capture more biological information than unweighted networks based on the same metric. While we use biological data sets to illustrate the method, the approach is general and can be used to construct networks in other domains.

Integrated Network Analysis to Identify Key Modules and Potential Hub Genes Involved in Bovine Respiratory Disease: A Systems Biology Approach

Background: Bovine respiratory disease (BRD) is the most common disease in the beef and dairy cattle industry. BRD is a multifactorial disease resulting from the interaction between environmental stressors and infectious agents. However, the molecular mechanisms underlying BRD are not fully understood yet. Therefore, this study aimed to use a systems biology approach to systematically evaluate this disorder to better understand the molecular mechanisms responsible for BRD.Methods: Previously published RNA-seq data from whole blood of 18 healthy and 25 BRD samples were downloaded from the Gene Expression Omnibus (GEO) and then analyzed. Next, two distinct methods of weighted gene coexpression network analysis (WGCNA), i.e., module–trait relationships (MTRs) and module preservation (MP) analysis were used to identify significant highly correlated modules with clinical traits of BRD and non-preserved modules between healthy and BRD samples, respectively. After identifying respective modules by the two mentioned methods of WGCNA, functional enrichment analysis was performed to extract the modules that are biologically related to BRD. Gene coexpression networks based on the hub genes from the candidate modules were then integrated with protein–protein interaction (PPI) networks to identify hub–hub genes and potential transcription factors (TFs).Results: Four significant highly correlated modules with clinical traits of BRD as well as 29 non-preserved modules were identified by MTRs and MP methods, respectively. Among them, two significant highly correlated modules (identified by MTRs) and six nonpreserved modules (identified by MP) were biologically associated with immune response, pulmonary inflammation, and pathogenesis of BRD. After aggregation of gene coexpression networks based on the hub genes with PPI networks, a total of 307 hub–hub genes were identified in the eight candidate modules. Interestingly, most of these hub–hub genes were reported to play an important role in the immune response and BRD pathogenesis. Among the eight candidate modules, the turquoise (identified by MTRs) and purple (identified by MP) modules were highly biologically enriched in BRD. Moreover, STAT1, STAT2, STAT3, IRF7, and IRF9 TFs were suggested to play an important role in the immune system during BRD by regulating the coexpressed genes of these modules. Additionally, a gene set containing several hub–hub genes was identified in the eight candidate modules, such as TLR2, TLR4, IL10, SOCS3, GZMB, ANXA1, ANXA5, PTEN, SGK1, IFI6, ISG15, MX1, MX2, OAS2, IFIH1, DDX58, DHX58, RSAD2, IFI44, IFI44L, EIF2AK2, ISG20, IFIT5, IFITM3, OAS1Y, HERC5, and PRF1, which are potentially critical during infection with agents of bovine respiratory disease complex (BRDC).Conclusion: This study not only helps us to better understand the molecular mechanisms responsible for BRD but also suggested eight candidate modules along with several promising hub–hub genes as diagnosis biomarkers and therapeutic targets for BRD.

Beyond the Biosynthetic Gene Cluster Paradigm: Genome-Wide Coexpression Networks Connect Clustered and Unclustered Transcription Factors to Secondary Metabolic Pathways

There is an urgent need for novel bioactive molecules in both agriculture and medicine. The genomes of fungi are thought to contain vast numbers of metabolic pathways involved in the biosynthesis of secondary metabolites with diverse bioactivities.

Integrative Network-Based Analysis Reveals Gene Networks and Novel Drug Repositioning Candidates for Alzheimer Disease

Background and ObjectivesTo integrate genome-wide association study data with tissue-specific gene expression information to identify coexpression networks, biological pathways, and drug repositioning candidates for Alzheimer disease.MethodsWe integrated genome-wide association summary statistics for Alzheimer disease with tissue-specific gene coexpression networks from brain tissue samples in the Genotype-Tissue Expression study. We identified gene coexpression networks enriched with genetic signals for Alzheimer disease and characterized the associated networks using biological pathway analysis. The disease-implicated modules were subsequently used as a molecular substrate for a computational drug repositioning analysis, in which we (1) imputed genetically regulated gene expression within Alzheimer disease implicated modules; (2) integrated the imputed gene expression levels with drug-gene signatures from the connectivity map to identify compounds that normalize dysregulated gene expression underlying Alzheimer disease; and (3) prioritized drug compounds and mechanisms of action based on the extent to which they normalize dysregulated expression signatures.ResultsGenetic factors for Alzheimer disease are enriched in brain gene coexpression networks involved in the immune response. Computational drug repositioning analyses of expression changes within the disease-associated networks retrieved known Alzheimer disease drugs (e.g., memantine) as well as biologically meaningful drug categories (e.g., glutamate receptor antagonists).DiscussionOur results improve the biological interpretation of genetic data for Alzheimer disease and provide a list of potential antidementia drug repositioning candidates for which the efficacy should be investigated in functional validation studies.

Physiological Characteristics and Transcriptomic Dissection in Two Root Segments with Contrasting Net Fluxes of Ammonium and Nitrate of Poplar Under Low Nitrogen Availability

To investigate physiological and transcriptomic regulation mechanisms underlying the distinct net fluxes of NH4+ and NO3- in different root segments of Populus species under low nitrogen (N) conditions, we used saplings of P. × canescens supplied with either 500 (normal N) or 50 (low N) μM NH4NO3. The net fluxes of NH4+ and NO3-, and the concentrations of NH4+, amino acids, organic acids and the enzymatic activities of nitrite reductase (NiR) and glutamine synthetase (GS) in root segment II (SII, 35-70 mm to the apex) were lower than those in root segment I (SI, 0-35 mm to the apex). The net NH4+ influxes and the concentrations of organic acids were elevated, whereas the concentrations of NH4+ and NO3-, and the activities of NiR and GS were reduced in SI and SII in response to low N. A number of genes were significantly differentially expressed in SII vs SI and in both segments grown under low vs normal N conditions, and these genes were mainly involved in transport of NH4+ and NO3-, N metabolism, and adenosine triphosphate (ATP) synthesis. Moreover, the hub gene coexpression networks were dissected and correlated with N physiological processes in SI and SII under normal and low N conditions. These results suggest that the hub gene coexpression networks play pivotal roles in regulating N uptake and assimilation, amino acid metabolism as well as the levels of organic acids from TCA cycle in the two root segments of poplars in acclimation to low N availability.

cageminer: an R/Bioconductor package to prioritize candidate genes by integrating GWAS and gene coexpression networks

Summary: Although genome-wide association studies (GWAS) identify variants associated with traits of interest, they often fail in identifying causative genes underlying a given phenotype. Integrating GWAS and gene coexpression networks can help prioritize high-confidence candidate genes, as the expression profiles of trait-associated genes can be used to mine novel candidates. Here, we present cageminer, the first R package to prioritize candidate genes through the integration of GWAS and coexpression networks. Genes are considered high-confidence candidates if they pass all three filtering criteria implemented in cageminer, namely physical proximity to SNPs, coexpression with known trait-associated genes, and significant changes in expression levels in conditions of interest. Prioritized candidates can also be scored and ranked to select targets for experimental validation. By applying cageminer to a real data set, we demonstrate that it can effectively prioritize candidates, leading to >99% reductions in candidate gene lists. Availability and implementation: The package is available at Bioconductor (http://bioconductor.org/packages/cageminer).

The Wild Sugarcane and Sorghum Kinomes: Insights Into Expansion, Diversification, and Expression Patterns

The protein kinase (PK) superfamily is one of the largest superfamilies in plants and the core regulator of cellular signaling. Despite this substantial importance, the kinomes of sugarcane and sorghum have not been profiled. Here, we identified and profiled the complete kinomes of the polyploid Saccharum spontaneum (Ssp) and Sorghum bicolor (Sbi), a close diploid relative. The Sbi kinome was composed of 1,210 PKs; for Ssp, we identified 2,919 PKs when disregarding duplications and allelic copies, and these were related to 1,345 representative gene models. The Ssp and Sbi PKs were grouped into 20 groups and 120 subfamilies and exhibited high compositional similarities and evolutionary divergences. By utilizing the collinearity between the species, this study offers insights into Sbi and Ssp speciation, PK differentiation and selection. We assessed the PK subfamily expression profiles via RNA-Seq and identified significant similarities between Sbi and Ssp. Moreover, coexpression networks allowed inference of a core structure of kinase interactions with specific key elements. This study provides the first categorization of the allelic specificity of a kinome and offers a wide reservoir of molecular and genetic information, thereby enhancing the understanding of Sbi and Ssp PK evolutionary history.