Integrating genetic and network analysis to characterize genes related to mouse weight.
Ghazalpour A, Doss S, Zhang B, Wang S, Plaisier C, Castellanos R, Brozell A, Schadt EE, Drake TA, Lusis AJ, Horvath S. · 2006
Read Integrating genetic and network analysis to characterize genes related to mouse weight. by Ghazalpour A, Doss S, Zhang B, Wang S, Plaisier C, Castellanos R, Brozell A, Schadt EE, Drake TA, Lusis AJ, Horvath S. (2006) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc1550283.txt.
Introduction
Gene activity webs can reveal which genes drive body weight. Ghazalpour and co-authors built such a web in 2006 from liver data of crossed mice. They grouped co-active genes into modules, then tied modules to 22 body traits.
Some chromosome spots, dubbed module QTL, perturbed whole modules at once. For a weight-linked module, the team modeled which genes matter using both those spots and each gene's web centrality. Genes that were both central and under a spot's sway predicted weight best. Ghazalpour argues that joining genetics with network maps can pinpoint the genes that steer complex traits.
Catalog record
| CREATOR | Ghazalpour A, Doss S, Zhang B, Wang S, Plaisier C, Castellanos R, Brozell A, Schadt EE, Drake TA, Lusis AJ, Horvath S. |
|---|---|
| YEAR | 2006 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 64 KB |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Integrating Genetic and Network Analysis to Characterize Genes Related to Mouse Weight By Anatole Ghazalpour, Sudheer Doss, Bin Zhang, Susanna Wang, Christopher Plaisier, Ruth Castellanos, Alec Brozell, Eric E Schadt, Thomas A Drake, Aldons J Lusis, Steve Horvath, Greg Gibson Published in: PLoS Genetics DOI: 10.1371/journal.pgen.0020130 ABSTRACT Abstract Systems biology approaches that are based on the genetics of gene expression have been fruitful in identifying genetic regulatory loci related to complex traits. We use microarray and genetic marker data from an F2 mouse intercross to examine the large-scale organization of the gene co-expression network in liver, and annotate several gene modules in terms of 22 physiological traits. We identify chromosomal loci (referred to as module quantitative trait loci, mQTL) that perturb the modules and describe a novel approach that integrates network properties with genetic marker information to model gene/trait relationships. Specifically, using the mQTL and the intramodular connectivity of a body weight–related module, we describe which factors determine the relationship between gene expression profiles and weight. Our approach results in the identification of genetic targets that influence gene modules (pathways) that are related to the clinical phenotypes of interest. INTRODUCTION The identification of pathways and genes underlying complex traits using standard mapping techniques has been difficult due to genetic heterogeneity, epistatic interactions, and environmental factors [1,2]. One promising approach to this problem involves the integration of genetics and gene expression [3–7]. Network methods have been applied to identify and characterize various biological interactions [8–11], and have helped to predict gene function in lower eukaryotes [12,13]. Recently, gene network methods have been used in the analysis of complex traits in higher organisms [14,15]. Here, we present a novel approach for using gene co-expression networks to study the genetics of complex physiological traits that are relevant to metabolic syndrome (obesity, insulin resistance, and dyslipidemia). The traditional quantitative trait locus (QTL) mapping approach [16] relates clinical traits to genetic markers directly. Several authors have proposed a strategy that uses genome-wide gene expression data to help map clinical traits [3–7,17–25] The strategy uses the genetics of gene expression to reconstruct metabolic or regulatory pathways, and utilizes gene expression as a quantitative trait, thereby mapping expression QTL (eQTL). Using this method, several groups have found “hotspot” regions (regulatory gene regions) in the genome that are involved in regulating the expression of many genes [3,18,26]. These hotspots highlight the genomic loci that determine the relationship between the phenotype and groups of functionally related genes [26]. Here, we propose a novel approach for integrating network properties with genetic information to determine the relationship between clinical traits and groups of physiologically relevant genes. The following steps summarize our overall approach: (1) A gene co-expression network is constructed from genome-wide expression data from a segregating population. (2) The biochemical and physiological significance of the network modules are determined. (3) Genetic loci regulating gene modules within the network are identified. (4) Network properties are integrated with genetic information to explain the biological significance of the module genes. RESULTS Construction of a Weighted Mouse Liver Co-Expression Network The details of the gene co-expression network construction are given in [27] and summarized in Materials and Methods. Briefly, the absolute value of the Pearson correlation coefficient is calculated for all pair-wise comparisons of gene expression values across all microarray samples. This correlation matrix is then transformed into a matrix of connection strengths using a power function (connection strength = |correlation|ß), resulting in a weighted network. The use of weighted networks represents an improvement over unweighted networks that are based on dichotomizing the correlation matrix, because (1) the continuous nature of the gene co-expression information is preserved and (2) the results of weighted network analyses are highly robust with respect to the choice of the parameter ß, whereas unweighted networks display sensitivity to the choice of the cutoff threshold [27]. We applied the network construction algorithm to a subset of gene expression data from an F2 intercross between inbred strains C3H/HeJ and C57BL/6J. Liver gene expression data from 135 female mice were used for this analysis (see Materials and Methods). This cross is described in detail elsewhere [28], and will herein be referred to as the BxH cross. As detailed in the Materials and Methods section, we used only the 3,421 most varying and most connected genes for module detection. This should not lead to major information loss since module genes tend to be highly connected. As shown in Figure 1A and 1B, transcripts were clustered into distinct groups, herein referred to as modules. We were able to identify 12 distinct gene modules or groups of genes with high topological overlap (for details on module construction, see Materials and Methods). To distinguish between modules, we designated each module by an arbitrary color. The number of genes included in the modules ranged from 34 (Light-yellow) to 772 (Red), and their mean overall connectivity (kall) ranged from 6.49 (Salmon) to 27.58 (Brown). We also defined the intramodular connectivity (kin) for each gene based on its Pearson correlation with all other genes in the module (see Materials and Methods). Summary statistics for all gene modules are available in Table S1. Detailed information about all genes and their network properties are available in Protocol S1. [Figure: Visual Representations of the Gene Co-Expression Network(A) Hierarchical clustering of 3,421 genes and visualization of gene module partitioning. The colored bars (below) directly correspond to the module (color) designation for the clusters of genes. One can visualize where in the whole clustering dendrogram the gene modules were defined.(B) Multi-dimensional scaling plot representation of the entire gene expression network. Each gene is represented by a dot, where the color of the dot corresponds to the gene module to which that gene belongs. The distance between each dot is representative of their topological overlap. This representation allows one to visualize the relationships among genes within a gene module, how that module is related to the rest of the network, and how closely any two modules are related.] Since we make use of the intramodular connectivity in the Blue module (534 genes), we studied its robustness with respect to dropping half of the microarrays used in its definition. Specifically, the following procedure was carried out 1,000 times. We randomly omitted half of the microarrays from the analysis when computing intramodular connectivity of the Blue module genes. Next, we correlated the resulting connectivity measures with the original full data intramodular connectivities. The median correlation between the full data connectivity and that of the 1,000 replicate subsets was 0.89 (interquartile range = [0.86,0.91]), which demonstrates that the intramodular connectivity measure is highly robust with respect to randomly omitting half of the microarrays. Biological Significance of Network Modules We examined the functional significance of the gene modules by testing for both enriched biochemical pathways and subcellular compartmentalization of proteins corresponding to genes within a gene module. First, we examined whether these modules were enriched for gene products in a specific subcellular compartment as defined by Gene Ontology (GO). Using the WebGestalt web site (http://genereg.ornl.gov/webgestalt) [29], we calculated the enrichment of any specific GO cellular component in each module. All statistically significant observations are presented in Protocol S2. The results show that each module was enriched with distinct gene