Weighted gene coexpression network analysis strategies applied to mouse weight

Systems-oriented genetic approaches that incorporate gene expression and genotype data are valuable in the quest for genetic regulatory loci underlying complex traits. Gene coexpression network analysis lends itself to identification of entire groups of differentially regulated genes—a highly relevant endeavor in finding the underpinnings of complex traits that are, by definition, polygenic in nature. Here we describe one such approach based on liver gene expression and genotype data from an F₂ mouse intercross utilizing weighted gene coexpression network analysis (WGCNA) of gene expression data to identify physiologically relevant modules. We describe two strategies: single-network analysis and differential network analysis. Single-network analysis reveals the presence of a physiologically interesting module that can be found in two distinct mouse crosses. Module quantitative trait loci (mQTLs) that perturb this module were discovered. In addition, we report a list of genetic drivers for this module. Differential network analysis reveals differences in connectivity and module structure between two networks based on the liver expression data of lean and obese mice. Functional annotation of these genes suggests a biological pathway involving epidermal growth factor (EGF). Our results demonstrate the utility of WGCNA in identifying genetic drivers and in finding genetic pathways represented by gene modules. These examples provide evidence that integration of network properties may well help chart the path across the gene–trait chasm. Electronic supplementary material The online version of this article (doi: ) contains supplementary material, which is available to authorized users. Tova F. Fuller, Anatole Ghazalpour contributed equally to this work.     

A Systems-Genetics Approach and Data Mining Tool to Assist in the Discovery of Genes Underlying Complex Traits in Oryza sativa

Many traits of biological and agronomic significance in plants are controlled in a complex manner where multiple genes and environmental signals affect the expression of the phenotype. In Oryza sativa (rice), thousands of quantitative genetic signals have been mapped to the rice genome. In parallel, thousands of gene expression profiles have been generated across many experimental conditions. Through the discovery of networks with real gene co-expression relationships, it is possible to identify co-localized genetic and gene expression signals that implicate complex genotype-phenotype relationships. In this work, we used a knowledge-independent, systems genetics approach, to discover a high-quality set of co-expression networks, termed Gene Interaction Layers (GILs). Twenty-two GILs were constructed from 1,306 Affymetrix microarray rice expression profiles that were pre-clustered to allow for improved capture of gene co-expression relationships. Functional genomic and genetic data, including over 8,000 QTLs and 766 phenotype-tagged SNPs (p-value < = 0.001) from genome-wide association studies, both covering over 230 different rice traits were integrated with the GILs. An online systems genetics data-mining resource, the GeneNet Engine, was constructed to enable dynamic discovery of gene sets (i.e. network modules) that overlap with genetic traits. GeneNet Engine does not provide the exact set of genes underlying a given complex trait, but through the evidence of gene-marker correspondence, co-expression, and functional enrichment, site visitors can identify genes with potential shared causality for a trait which could then be used for experimental validation. A set of 2 million SNPs was incorporated into the database and serve as a potential set of testable biomarkers for genes in modules that overlap with genetic traits. Herein, we describe two modules found using GeneNet Engine, one with significant overlap with the trait amylose content and another with significant overlap with blast disease resistance.     

Molecular-Marker-Facilitated Investigations of Quantitative-Trait Loci in Maize. I. Numbers, Genomic Distribution and Types of Gene Action

Individual genetic factors which underlie variation in quantitative traits of maize were investigated in each of two F2 populations by examining the mean trait expressions of genotypic classes at each of 17-20 segregating marker loci. It was demonstrated that the trait expression of marker locus classes could be interpreted in terms of genetic behavior at linked quantitative trait loci (QTLs). For each of 82 traits evaluated, QTLs were detected and located to genomic sites. The numbers of detected factors varied according to trait, with the average trait significantly influenced by almost two-thirds of the marked genomic sites. Most of the detected associations between marker loci and quantitative traits were highly significant, and could have been detected with fewer than the 1800-1900 plants evaluated in each population. The cumulative, simple effects of marker-linked regions of the genome explained between 8 and 40% of the phenotypic variation for a subset of 25 traits evaluated. Single marker loci accounted for between 0.3% and 16% of the phenotypic variation of traits. Individual plant heterozygosity, as measured by marker loci, was significantly associated with variation in many traits. The apparent types of gene action at the QTLs varied both among traits and between loci for given traits, although overdominance appeared frequently, especially for yield-related traits. The prevalence of apparent overdominance may reflect the effects of multiple QTLs within individual marker-linked regions, a situation which would tend to result in overestimation of dominance. Digenic epistasis did not appear to be important in determining the expression of the quantitative traits evaluated. Examination of the effects of marked regions on the expression of pairs of traits suggests that genomic regions vary in the direction and magnitudes of their effects on trait correlations, perhaps providing a means of selecting to dissociate some correlated traits. Marker-facilitated investigations appear to provide a powerful means of examining aspects of the genetic control of quantitative traits. Modifications of the methods employed herein will allow examination of the stability of individual gene effects in varying genetic backgrounds and environments.     

Genome-wide association: a promising start to a long race

A recent study by Cheung et al. demonstrates how to identify expression quantitative trait loci (eQTLs) underlying gene expression phenotypes through a combination of genome-wide linkage analysis and subsequent fine mapping or by genome-wide association (GWA) analysis. This study emphasizes the complexity of human traits, highlighting the challenges faced by investigators--in particular, insufficient linkage disequilibrium between the trait and marker variant, genetic heterogeneity and correcting for multiple testing will all adversely impact the power to detect loci by association. These issues must be considered carefully if the GWA approach is to succeed in mapping complex phenotypes.     

Expression QTL modules as functional components underlying higher-order phenotypes

Systems genetics studies often involve the mapping of numerous regulatory relations between genetic loci and expression traits. These regulatory relations form a bipartite network consisting of genetic loci and expression phenotypes. Modular network organizations may arise from the pleiotropic and polygenic regulation of gene expression. Here we analyzed the expression QTL (eQTL) networks derived from expression genetic data of yeast and mouse liver and found 65 and 98 modules respectively. Computer simulation result showed that such modules rarely occurred in randomized networks with the same number of nodes and edges and same degree distribution. We also found significant within-module functional coherence. The analysis of genetic overlaps and the evidences from biomedical literature have linked some eQTL modules to physiological phenotypes. Functional coherence within the eQTL modules and genetic overlaps between the modules and physiological phenotypes suggests that eQTL modules may act as functional units underlying the higher-order phenotypes.     

A Computational Systems Biology Study for Understanding Salt Tolerance Mechanism in Rice

Salinity is one of the most common abiotic stresses in agriculture production. Salt tolerance of rice (Oryza sativa) is an important trait controlled by various genes. The mechanism of rice salt tolerance, currently with limited understanding, is of great interest to molecular breeding in improving grain yield. In this study, a gene regulatory network of rice salt tolerance is constructed using a systems biology approach with a number of novel computational methods. We developed an improved volcano plot method in conjunction with a new machine-learning method for gene selection based on gene expression data and applied the method to choose genes related to salt tolerance in rice. The results were then assessed by quantitative trait loci (QTL), co-expression and regulatory binding motif analysis. The selected genes were constructed into a number of network modules based on predicted protein interactions including modules of phosphorylation activity, ubiquity activity, and several proteinase activities such as peroxidase, aspartic proteinase, glucosyltransferase, and flavonol synthase. All of these discovered modules are related to the salt tolerance mechanism of signal transduction, ion pump, abscisic acid mediation, reactive oxygen species scavenging and ion sequestration. We also predicted the three-dimensional structures of some crucial proteins related to the salt tolerance QTL for understanding the roles of these proteins in the network. Our computational study sheds some new light on the mechanism of salt tolerance and provides a systems biology pipeline for studying plant traits in general.     

A stochastic expectation and maximization algorithm for detecting quantitative trait-associated genes

MOTIVATION: Most biological traits may be correlated with the underlying gene expression patterns that are partially determined by DNA sequence variation. The correlations between gene expressions and quantitative traits are essential for understanding the functions of genes and dissecting gene regulatory networks. RESULTS: In the present study, we adopted a novel statistical method, called the stochastic expectation and maximization (SEM) algorithm, to analyze the associations between gene expression levels and quantitative trait values and identify genetic loci controlling the gene expression variations. In the first step, gene expression levels measured from microarray experiments were assigned to two different clusters based on the strengths of their association with the phenotypes of a quantitative trait under investigation. In the second step, genes associated with the trait were mapped to genetic loci of the genome. Because gene expressions are quantitative, the genetic loci controlling the expression traits are called expression quantitative trait loci. We applied the same SEM algorithm to a real dataset collected from a barley genetic experiment with both quantitative traits and gene expression traits. For the first time, we identified genes associated with eight agronomy traits of barley. These genes were then mapped to seven chromosomes of the barley genome. The SEM algorithm and the result of the barley data analysis are useful to scientists in the areas of bioinformatics and plant breeding. Availability and implementation: The R program for the SEM algorithm can be downloaded from our website: http://www.statgen.ucr.edu.     

Integrated multiple population analysis of leaf architecture traits in maize (Zea mays L.)

Leaf morphology in maize is regulated by developmental patterning along three axes: proximodistal, mediolateral, and adaxial-abaxial. Maize contains homologues of many genes identified as regulators of leaf development in other species, but their relationship to the natural variation of leaf shape remains unknown. In this study, quantitative trait loci (QTLs) for leaf angle, leaf orientation value, leaf length, and leaf width were mapped by a total of 256 F(2:3) families evaluated in three environments. Meta-analysis was used to integrate genetic maps and detect QTLs across several independent QTL studies, on the basis of the previously reported experimental results for leaf architecture traits. Candidate gene sequences for leaf architecture were mapped in the integrated consensus genetic map. In total, 21 QTLs and 17 meta-QTLs (mQTLs) were detected. Among these QTLs, qLA1-1 and qLA2 were consistently detected in five and three populations respectively, and six of seven QTLs with contributions (R(2)) >10% were integrated in mQTLs. Six key mQTLs (mQTL1-1, mQTL2-1, mQTL3-3, mQTL5-1, mQTL7-2, and mQTL8-1) with R(2) of some initial QTLs >10% included 4-6 initial QTLs associated with 2-4 traits. Therefore, the chromosome regions for six mQTLs with high QTL co-localization might be hot spots of the important QTLs for the associated traits. Fifteen key candidate genes controlling leaf architecture traits coincided with 11 corresponding mQTLs, namely DWARF4, KAN3, liguleless1, TAC1, ROT3, AS2/liguleless2, PFL2, yabby9/SE/LIC/yabby15, mwp1, CYCD3;2, and CYCB1. In particular, DWARF4, liguleless1, AS2/liguleless2, yabby9/SE/LIC/yabby15, and CYCD3;2 were mapped within the important mQTL1-1, mQTL2-1, mQTL3-3, mQTL5-1, and mQTL7-2 intervals, respectively. Fine mapping or construction of single chromosome segment lines for genetic regions of these five mQTLs is worth further study and could be put to use in marker-assisted breeding. In conclusion, the results provide useful information for further research and help to reveal the molecular mechanisms with regard to leaf architecture traits.     

Identification of enzymatic and regulatory genes of plant metabolism through QTL analysis in Arabidopsis

The biochemical diversity in the plant kingdom is estimated to well exceed 100,000 distinct compounds (Weckwerth, 2003) and 4000 to 20,000 metabolites per species seem likely (Fernie et al., 2004). In recent years extensive progress has been made towards the identification of enzymes and regulatory genes working in a complex network to generate this large arsenal of metabolites. Genetic loci influencing quantitative traits, e.g. metabolites or biomass, may be mapped to associated molecular markers, a method called quantitative trait locus mapping (QTL mapping), which may facilitate the identification of novel genes in biochemical pathways. Arabidopsis thaliana, as a model organism for seed plants, is a suitable target for metabolic QTL (mQTL) studies due to the availability of highly developed molecular and genetic tools, and the extensive knowledge accumulated on the metabolite profile. While intensely studied, in particular since the availability of its complete sequence, the genome of Arabidopsis still comprises a large proportion of genes with only tentative function based on sequence homology. From a total number of 33,518 genes currently listed (TAIR 9, http://www.arabidopsis.org), only about 25% have direct experimental evidence for their molecular function and biological process, while for more than 30% no biological data are available. Modern metabolomics approaches together with continually extended genomic resources will facilitate the task of assigning functions to those genes. In our previous study we reported on the identification of mQTL (Lisec et al., 2008). In this paper, we summarize the current status of mQTL analyses and causal gene identification in Arabidopsis and present evidence that a candidate gene located within the confidence interval of a fumarate mQTL (AT5G50950) encoding a putative fumarase is likely to be the causal gene of this QTL. The total number of genes molecularly identified based on mQTL studies is still limited, but the advent of multi-parallel analysis techniques for measurement of gene expression, as well as protein and metabolite abundances and for rapid gene identification will assist in the important task of assigning enzymes and regulatory genes to the growing network of known metabolic reactions.     

Genetic analysis of leaf morphology underlying the plant density response by QTL mapping in maize (<Emphasis Type="Italic">Zea mays</Emphasis> L.)

Maize yield increase has been strongly linked to plant population densities over time with changes in plant architecture, but the genetic basis for the plant architecture response to plant density is unknown, as is its stability across environments. To elucidate the genetic basis of the plant architecture response to density in maize, we mapped quantitative trait loci (QTLs) for leaf morphology-related traits in four sets of recombinant inbred line (RIL) populations under two plant density conditions. Forty-five QTLs for six traits were detected in both high and low plant density conditions. Thirty-seven QTLs were only detected when grown under high plant density, and 34 QTLs were only detected when grown under low plant density. Twenty-two meta-QTLs (mQTLs) were identified by meta-analysis, and mQTL1-1, mQTL3-2 and mQTL8 were identified when grown under high and low plant densities, with R ² of some initial QTLs > 10 %, suggesting the mQTLs might be hot spots of the important QTLs for the related traits under planting density stress conditions. The results presented here provide useful information for further research and the marker-assisted selection of varieties targeting increased plant density and will help to reveal the molecular mechanisms related to leaf morphology in response to density.     

Dissection of genotype-phenotype associations in rice grains using metabolome quantitative trait loci analysis

A comprehensive and large‐scale metabolome quantitative trait loci (mQTL) analysis was performed to investigate the genetic backgrounds associated with metabolic phenotypes in rice grains. The metabolome dataset consisted of 759 metabolite signals obtained from the grains of 85 lines of rice (Oryza sativa, Sasanishiki × Habataki back‐crossed inbred lines). Metabolome analysis was performed using four mass spectrometry pipelines to enhance detection of different classes of metabolites. This mQTL analysis of a wide range of metabolites highlighted an uneven distribution of 802 mQTLs on the rice genome, as well as different modes of metabolic trait (m‐trait) control among various types of metabolites. The levels of most metabolites within rice grains were highly sensitive to environmental factors, but only weakly associated with mQTLs. Coordinated control was observed for several groups of metabolites, such as amino acids linked to the mQTL hotspot on chromosome 3. For flavonoids, m‐trait variation among the experimental lines was tightly governed by genetic factors that alter the glycosylation of flavones. Many loci affecting levels of metabolites were detected by QTL analysis, and plausible gene candidates were evaluated by in silico analysis. Several mQTLs profoundly influenced metabolite levels, providing insight into the control of rice metabolism. The genomic region and genes potentially responsible for the biosynthesis of apigenin‐6,8‐di‐C‐α‐l‐ arabinoside are presented as an example of a critical mQTL identified by the analysis.     

The Contribution of Quantitative Trait Loci and Neutral Marker Loci to the Genetic Variances and Covariances among Quantitative Traits in Random Mating Populations

Using Cockerham's approach of orthogonal scales, we develop genetic models for the effect of an arbitrary number of multiallelic quantitative trait loci (QTLs) or neutral marker loci (NMLs) upon any number of quantitative traits. These models allow the unbiased estimation of the contributions of a set of marker loci to the additive and dominance variances and covariances among traits in a random mating population. The method has been applied to an analysis of allozyme and quantitative data from the European oyster. The contribution of a set of marker loci may either be real, when the markers are actually QTLs, or apparent, when they are NMLs that are in linkage disequilibrium with hidden QTLs. Our results show that the additive and dominance variances contributed by a set of NMLs are always minimum estimates of the corresponding variances contributed by the associated QTLs. In contrast, the apparent contribution of the NMLs to the additive and dominance covariances between two traits may be larger than, equal to or lower than the actual contributions of the QTLs. We also derive an expression for the expected variance explained by the correlation between a quantitative trait and multilocus heterozygosity. This correlation explains only a part of the genetic variance contributed by the markers, i.e., in general, a combination of additive and dominance variances and, thus, provides only very limited information relative to the method supplied here.     

A genetical genomics approach to genome scans increases power for QTL mapping

We describe a method for integrating gene expression information into genome scans and show that this can substantially increase the statistical power of QTL mapping. The method has three stages. First, standard clustering methods identify small (size 5-20) groups of genes with similar expression patterns. Second, each gene group is tested for a causative genetic locus shared with the clinical trait of interest. This is done using an EM algorithm approach that treats genotype at the putative causative locus as an unobserved variable and combines expression information from all of the genes in the group to infer genotype information at the locus. Finally, expression QTL (eQTL) are mapped for each gene group that shares a causative locus with the clinical trait. Such eQTL are candidates for the causative locus. Simulation results show that this method has far superior power to standard QTL mapping techniques in many circumstances. We applied this method to existing data on mouse obesity. Our method identified 27 putative body weight QTL, whereas standard QTL mapping produced only one. Furthermore, most gene groups with body weight QTL included cis genes, so candidate genes could be immediately identified. Eleven body weight QTL produced 16 candidate genes that have been previously associated with body weight or body weight-related traits, thus validating our method. In addition, 15 of the 16 other loci produced 32 candidate genes that have not been associated with body weight. Thus, this method shows great promise for finding new causative loci for complex traits.