J. Anim. Sci. 2004. 82:E300-E312
© 2004 American Society of Animal Science
Quantitative genomics: Exploring the genetic architecture of complex trait predisposition1,2
D. Pomp3,
M. F. Allan and
S. R. Wesolowski
Department of Animal Science, University of Nebraska, Lincoln 68583-0908
 |
Abstract
|
|---|
Most phenotypes with agricultural or biomedical relevance are multifactorial traits controlled by complex contributions of genetics and environment. Genetic predisposition results from combinations of relatively small effects due to variations within a large number of genes, known as QTL. Well over 200 QTL have been reported for growth and body composition traits in the mouse, which likely represent at least 50 to 100 distinct genes. Molecular biology has yielded significant advances in understanding these traits at the metabolic and physiological levels; however, little has been learned regarding the identity and nature of the underlying polygenes. In addition to the significantly poor precision inherent to QTL localization, it is very difficult to differentiate between co-localization and coincidence when comparing QTL with other QTL and with potential candidate genes. The wide gap between our knowledge of physiological mechanisms underlying complex traits and the nature of genetic predisposition significantly impairs discovery of genes underlying QTL. Identification and genetic mapping of key transcriptional, proteomic, metabolomic, and endocrine events will uncover large lists of significant positional candidate genes for growth and body composition. However, integration of experimental approaches to jointly evaluate predisposition and physiology will increase success of QTL identification by merging the power of recombination with functional analysis. Measuring physiologically relevant subphenotypes within a structured QTL mapping population will not only facilitate pathway-specific prioritization among candidate genes, but may also directly identify genes underlying QTL. This would advance our understanding of the genetic architecture of complex traits by testing the central hypothesis that genes controlling predisposition to a quantitative trait are primarily involved in trans-regulation of the primary physiological pathways that regulate the trait.
Key Words: Body Fat • Body Weight • Complex Traits • Gene Expression • Genetic Architecture • Quantitative Trait Loci
|
Introduction
|
|---|
Most quantitative traits are exceptionally complex, with varying contributions of genetic susceptibility and interacting environmental factors. Predisposition to a phenotypic range for a complex trait, such as body weight (BW), results from combinations of relatively small effects of DNA variations within a large number of unidentified polygenes, known as QTL. Over 200 QTL have been reported for growth and body composition traits in the mouse, likely representing at least 50 to 100 distinct genes (Figure 1
). Although molecular biology has yielded significant gains in understanding complex traits, such as weight regulation at the metabolic and physiological levels (e.g. leptin, melanocortin, and insulin pathways), the genetic architecture of obesity predisposition remains essentially undefined. This large gap between our extensive knowledge of the physiological mechanisms underlying BW and our embryonic understanding of how genetic predisposition is manifested impairs identification of genes underlying relevant QTL and inhibits gene-based development of diagnostic and therapeutic tools.
View larger version (53K):
[in this window]
[in a new window]
|
Figure 1. Mouse body weight predisposition map. The current status of mapped QTL for fatness (orange), BW, weight gain (violet), serum cholesterol (green), heat loss (red), and levels of mRNA or proteins for candidate genes with physiological relevance to obesity (black). All BW and weight gains are from animals of 6 wk of age or older. Lengths of chromosomes and QTL map positions are according to the Mouse Genome Database. It is emphasized that QTL map positions may be inherently inaccurate (in some cases, 95% confidence intervals for a QTL include a majority of a chromosome). Only QTL mapped with at least 5% genome-wide significance levels were included in the map, but several suggestive QTL were included when their map positions reaffirmed map positions of other QTL from independent experiments. Symbols for QTL are as presented in the literature (see Elo [2003] for full list of references). If authors did not provide symbols, then those suggested by Chagnon et al. (2003) were used. Adapted from Elo (2003).
|
|
We propose a central hypothesis that the majority of genes controlling predisposition to complex traits, such as BW and obesity, are involved in trans-regulation of the primary physiological pathways directly regulating energy balance phenotypes. This hypothesis has been formulated based on several areas of accumulated data. First, few causative mutations have been found within energy balance candidate genes despite significant detection efforts in humans (Chagnon et al., 2003). Second, studies localizing QTL regulating mRNA or protein levels of such candidate genes have primarily identified trans-acting regulation. And third, recent genome-wide evaluations have found that trans-acting loci are the primary drivers of variation in gene expression in yeast (Brem et al., 2002; Yvert et al., 2003), Drosophila (Montooth et al., 2003), and mice (Schadt et al., 2003a). We contend that the paradigm of "quantitative genomics," whereby large-scale subphenotyping at the transcriptional, proteomic, and metabolomic levels is performed within the context of a QTL mapping population, will be a powerful force in dissecting the genetic architecture of complex trait predisposition.
|
Significance
|
|---|
Carcass and body composition traits constitute extremely important considerations of modern livestock production systems, where consumer health concerns and marketing perspectives play increasingly prominent roles. The problem of excess fat in livestock and poultry carcasses is ubiquitous and has serious consequences for the animal industry on four levels: health perceptions of consumers, wasteful production of an undesired biological component, labor costs associated with trimming waste fat, and lower biological efficiencies of fatter animals (Eisen, 1989).
An estimated 65% of U.S. adults are overweight, and 31% are obese, with higher percentages in female minority populations (Flegal et al., 2002). Overweight and obese conditions substantially increase risk of hypertension, dyslipidemia, type 2 diabetes, coronary heart disease, stroke, gallbladder disease, sleep apnea, osteoarthritis, respiratory problems, and endometrial, breast, prostate, and colon cancers, combining to form the single largest cause of death in developed countries (NHLBI, 1998).
Understanding the roles of specific loci in genetic susceptibility to obesity is critical to improving human health and quality of life. As Comuzzie and Allison (1998) stated, "One of the greatest challenges in biomedical research today is the elucidation of the underlying genetic architecture of complex phenotypes such as obesity." Unfortunately, little progress has been made despite significant research attention. A tremendous gap exists between our embryonic knowledge of the nature of genetic predisposition to obesity and our bourgeoning understanding of its physiological and molecular underpinnings. Given the approximately $75 billion annual health costs associated with obesity (Finkelstein et al., 2003), finding novel targets for pharmacological intervention and pharmacogenomic management is critical. Enhanced understanding of how complex traits are controlled will also aid in elucidating the nature of predisposition for other diseases, including certain forms of cancer, and will be broadly applicable to many important phenotypes.
|
Background
|
|---|
A significant heritable component for disease was formally recognized as long ago as the turn of the 20th century, when Sir Archibald Garrod (1902) described the heritable nature of alkaptonuria. Since then, well over 5,000 genetic disorders exhibiting Mendelian inheritance have been described, and the molecular basis for many of these has been identified. These diseases have phenotypes that fall into discrete classes (e.g. afflicted vs. not afflicted). In contrast, most economically relevant traits in animal agriculture, and most common human maladies such as obesity, exhibit continuous phenotypic variation and a predominantly multifactorial and polygenic basis (Festing, 1979; Rich, 1990). Although certain rare mutations have been identified that account for a small minority of extreme phenotypes (e.g., in animals: McPherron and Lee, 1996; in humans: Chagnon et al., 2003), the actual identity of genes segregating and contributing to common phenotypes in populations is essentially unknown.
The location of polygenic factors controlling inheritance of quantitative traits can be established by tracking the segregation of closely linked markers (QTL mapping). This dates back to early work by Sturtevant (1913) with Drosophila, Haldane et al. (1915) with mice, and Sax (1923) with Phaseolus vulgaris. For the subsequent 70 yr, analyses continued to use visible phenotypic markers and protein variants. However, DNA marker analysis of quantitative traits has recently gained prominence with development of ubiquitous and polymorphic marker systems (e.g., Dietrich et al., 1996) and powerful statistical methodologies (e.g., Lander and Botstein, 1989; Zeng, 1993, 1994; Haley et al., 1994). The first extensive use of DNA markers was in plants, where large suitable families could be generated (Paterson et al., 1988; Martin et al., 1989). The feasibility of performing QTL mapping in mammals was first demonstrated using rats (Hilbert et al., 1991; Jacob et al., 1991), but the mouse has subsequently dominated research in polygenic analysis, including BW regulation (Brockmann and Bevova, 2002).
Over 200 QTL for growth and obesity-related traits have been localized in the mouse, representing a comprehensive and exhaustive portrait of the predisposition map for mouse BW regulation. For simplicity and convenience, we have graphically summarized the approximately 50 articles contributing this information in Figure 1
(see also Table 3 in each of Rocha et al., 2004a,b). However, few (if any) of these QTL have been unequivocally cloned, dramatically limiting the ability to harness this critical mass of information for the betterment of human health. This scenario is all the more surprising due to the fact that advancements in knowledge on the endocrine, biochemical and molecular underpinnings of obesity represent a significant success story in modern biology, sparked originally by study of spontaneous mutations causing obesity (e.g., Zhang et al., 1994; Michuad et al., 1994; reviewed by Chua, 1997) and later by targeted single gene mutations in a multitude of genes with relevance to energy balance (see reviews by Bray and Tartaglia, 2000; Barsh and Schwartz, 2002). This frustrating gap in our knowledge of physiological mechanisms underlying obesity, and the nature of genetic predisposition to obesity, was insightfully summarized by Bray and Bouchard (1997), who wrote: "Unfortunately, the spectacular gains in understanding the biology of energy balance of the last few years have not yet translated into significant advances on the genetic front. This is particularly striking when one realizes that so far there is not one single obese human being whose excess body fat can be explained by a specific mutation in one of the genes exerting its effects in relevant energy balance pathways."
In recent years, there have been a few reports of single gene mutations resulting in human obesity (e.g., Farooqi et al., 2003; reviewed by Chagnon et al., 2003), and we have seen a somewhat promising trend in the ability to elucidate the underlying identities of QTL including those underlying obesity-related traits (see review by Korstanje and Paigen, 2002). However, the painstaking and difficult nature of the process bodes poorly for rapid progress (Nadeau and Frankel, 2000). This dilemma grows in parallel with the continued rapid pace of discovery in understanding energy balance mechanisms (e.g., Schwartz et al., 2003; Wang et al., 2003b).
|
Transcriptome Mapping: A New Experimental Paradigm for Analysis of Complex Trait Genetics
|
|---|
A new paradigm for bridging the gap between our knowledge of the physiology and predisposition of obesity is the combining of QTL mapping with large-scale gene expression analysis. Transcriptome mapping (Williams et al., 2002b), also called "genetical genomics" (Jansen and Nap 2001; Jansen 2003), treats gene expression levels of any particular gene measured across different individuals as an expression-level polymorphism that in principle reflects the underlying genetic variation (Dumas et al., 2000; Jansen and Nap, 2001; Doerge, 2002). This type of analysis was actually pioneered by Damerval et al. (1994) and de Vienne et al. (1994) using proteomic evaluation (and later extended by the same group to the transcriptome (Consoli et al., 2002) in an F2 population of maize. Transcriptome mapping has been highlighted as a powerful mechanism to dissect complex traits and make more efficient the selection of candidate genes underlying predisposition loci, with recent successful implementation in yeast (Brem et al., 2002; Yvert et al., 2003), Drosophila (Wayne and McIntyre, 2002; Montooth et al., 2003) and mice (Hitzemann et al., 2003; Schadt et al., 2003a, b).
A critically important question that transcriptome mapping promises to help answer regards the underlying nature of obesity QTL. Are primary obesity QTL represented by sequence variation within genes with major roles in energy balance pathways or, as first proposed several years ago (Pomp, 1999), do they regulate such genes in a trans-acting manner? Although a plethora of candidate gene analyses in humans has provided mixed results (e.g., MC3R: Farooqi et al., 2003; UCP2: Esterbauer et al., 2001; LEPR: Heo et al., 2001), the latter speculation is strongly supported when the transcriptome/proteome-mapping paradigm has been applied on a modest basis by examining one or a few transcripts/proteins at a time. For example, mapping for determinants of levels of plasma leptin, IGF-1, and IGF-1 binding proteins has identified multiple QTL for each trait, but none coinciding with map positions of the structural genes themselves in mice (e.g., Mehrabian et al., 1998; Brockmann et al., 2000; Rosen et al., 2000) or humans (Hixson et al., 1999). Koza et al. (2000) identified several trans-acting QTL for induced Ucp1 mRNA levels in mice, and we have found similar results for control of hypothalamic mRNA levels of Rpl3, Oxt, and Timp2 (Wesolowski et al., 2003; Wesolowski and Pomp, unpublished data). In pigs, Rohrer et al. (2001) identified only trans-acting QTL influencing plasma FSH concentrations.
Although these very limited findings provide a tantalizing glimpse into the nature of complex trait predisposition and the underlying QTL, transcriptome mapping offers the potential for several orders-of-magnitude greater power and resolution. In yeast, for example, Brem et al. (2002) found QTL for 570 expressed genes in a cross between laboratory and wild strains. Of these, 36% were apparently due to polymorphisms within the gene themselves, whereas the remainder was controlled by a small group of trans-acting modulator loci each regulating from 7 to 94 genes of related function. Similar findings were reported by Schadt et al. (2003a), who, in their mouse study, further found that the stronger the evidence for an expression QTL, the more likely it would map within the structural gene itself. This latter result is expected as DNA mutations within a gene and affecting the expression of that gene should be easier to identify than second-order effects. However, it should be noted that in such cases, the causal polymorphism may be closely linked to, but not part of, the gene whose expression is being evaluated. Also, such results could possibly be artificially be created by polymorphisms influencing hybridization efficiency, although this would be more likely when shorter oligonucleotide probes are employed in a microarray.
Recently, Yvert et al. (2003) determined that most gene expression differences in a cross between laboratory and wild strains of yeast mapped to trans-acting loci. Furthermore, trans-regulatory variation was broadly dispersed across different classes of genes with a wide variety of functions. And further compelling evidence that obesity QTL may be represented by sequence variation within trans-acting loci that regulate genes with major roles in energy balance pathways is provided by the recent study of Montooth et al. (2003) using Drosophila. In that experiment, trans-regulatory variation was found in metabolic enzyme activity for each key determinant of metabolism and respiration measured.
The study by Schadt et al. (2003a) specifically targeted dietary-induced obesity in mice and identified two promising candidates potentially representing the MMU2 QTL described by Lembertas et al. (1977). This distal region of mouse chromosome 2 is one of the most relevant to obesity predisposition in the mouse genome (see Figure 1 and Chagnon et al., 2003). Not only is this region well populated with multiple BW and fatness QTL from crosses employing different approaches and genetic backgrounds (Lembertas et al., 1997; Mehrabian et al., 1998; Rocha et al., 2003a,b), QTL harbored in this region have among the largest effects of any BW and obesity polygenes ever localized (Pomp, 1997; Rocha et al., 2003a,b).
In regard to regulation of BW and obesity, critical questions remain to be answered after these pioneering, yet quite preliminary, initial studies using the transcriptome mapping paradigm. For example, will expression QTL represent genes with primary (cis) or secondary (trans) roles in energy balance? A second important question to address is whether expression QTL directly underlies obesity QTL? And more globally, can the transcriptome mapping paradigm indeed create a more successful environment for routine cloning of obesity predisposition genes?
|
Expected Outcomes from the Transcriptome Mapping Paradigm
|
|---|
Establishment of a robust system that can genetically map loci that modulate the steady-state levels of any gene showing transcript level variation between populations with divergent phenotypes is expected to provide a wealth of information regarding genetic architecture of complex traits. A hypothetical snapshot of the types of outcomes this research paradigm can provide is summarized in Figure 2, where the x-axis represents the known genetic map position of each gene represented on a large-scale expression microarray, and the y-axis represents the estimated genetic map location of the single QTL that explains the most variation in expression levels for each gene on the x-axis. Based on the early transcriptome mapping efforts, three patterns of transcript regulation are being revealed, with a fourth pattern possible.
View larger version (24K):
[in this window]
[in a new window]
|
Figure 2. Hypothesized results of mRNA expression profiling across a genotyped QTL mapping population. The x-axis is the general map position of each expressed sequence tag (EST) or gene in the expression array. The y-axis is the general map position of the QTL that explains the most variation in expression levels of each EST or gene in the expression array. Four generalized scenarios are described. Predominantly, levels of expressed genes will be controlled by trans-acting QTL (scattered yellow diamonds). The cis-acting QTL (diagonal red circles) may represent genetic variation within the regulatory or coding regions of the expressed genes themselves. In addition, a single QTL may result in changes in expression levels of many unlinked genes (horizontal blue triangles), either due to direct pleiotropy or to multiple changes in a regulatory cascade resulting from alteration of expression in a single key gene. Finally, clusters (small green dots) of gene expression changes could result from changes due to linkage of multiple expressed genes to a single regulatory QTL, or alternatively as a result of coordinated expression neighborhoods.
|
|
First, a large number of genes plotted along the diagonal will suggest that their transcript levels are cis-regulated (i.e., the location of their QTL transcript regulators genetically map to the physical location of the genes themselves). We would speculate that these are due to promoter polymorphisms or other variants within the genes themselves that affect transcript level. This "cis-diagonal" (Williams et al., 2002b) can immediately uncover high-quality candidate genes potentially representing QTL for growth and obesity phenotypes, especially when the map position of the expression QTL falls under the QTL peak for the endpoint phenotype (e.g., weight, fat, intake). A second class of genes would be those controlled by unlinked trans-regulators. These will be evidenced by the genetic locations of the controlling QTL being different than the physical location of the genes they regulate. Evaluation of the trans-regulating patterns of expression phenotypes will add immense value to the selection of candidate genes representing QTL by implicating pathways and mechanisms underlying the mechanism of action of the QTL. The third, and potentially most interesting class of genes, is QTL transcript modulators that would regulate the steady-state transcript levels of tens or even hundreds of genes spread across the genome. These master transcript modulators would be identified by the horizontal strips of plotted QTL modulators, indicating the presence of one or a few tightly linked regulatory genes. This class of results will represent two important findings. First, the QTL may be in a key gene within a pathway that, when perturbed by a polymorphism, causes a cascade of effects that are evidenced by multiple expression changes in other genes. Second, the QTL may represent a key genetic control switch, such as a transcription factor or helicase, a polymorphism within which could cause a multitude of changes in gene expression throughout the genome. A speculative fourth class of genes would be those representing potential "expression neighborhoods" (Oliver et al., 2002; Spellman and Rubin, 2002), although evidence for such results has not yet been observed in the transcriptome mapping experiments conducted to date.
We expect these efforts to begin to enable development of an initial framework for understanding the genetic architecture of obesity predisposition. Such studies should greatly facilitate testing our hypothesized structure for such architecture (Figure 3), whereby genes controlling predisposition to complex traits such as BW and obesity are, for the most part, involved in trans-regulation of the primary physiological pathways directly regulating phenotypes involved in energy balance.
View larger version (35K):
[in this window]
[in a new window]
|
Figure 3. This figure shows hypothesized genetic architecture for obesity predisposition. Simplified and generalized components of example pathways (out of many that combine to regulate obesity) are illustrated. We hypothesize that each of these pathways consist of a complex regulatory cascade with the coordinated and interactive expression of genes playing significant roles in the physiology of the pathway, under the regulation of predisposition genes (QTL) and nongenetic (environmental) influences. In other words, key genes in physiological pathways would, for the most part, not possess relevant heritable variation; such variation would instead reside within loci that regulate expression levels of these key physiological genes. There could be several potential modes of regulation and complexity for QTL that control expression or activity of physiological genes. QTLA (blue): QTL can regulate transcription of a physiological gene. For example, a transcription factor within which genetic variation leads to variability in activity or expression levels. QTLB (green): QTL can regulate post-translational modification of the physiological gene. For example, a kinase or a phosphatase within which genetic variation alters activity and subsequently changes the activity of the physiological gene. The QTLC (yellow): QTL can be pleiotropic factors that regulate multiple physiological genes. The QTLD (red) and QTLE (orange): QTL can act epistatically to regulate other QTL. For example, certain genetic variation within a QTL is required in order for variation at a second QTL to exert an influence on the expression or activity of a physiological gene (QTLD), or effects of multiple QTL must combine in order to regulate a physiological gene (QTLE). The QTLF (pink): cases where heritable genetic variation does exist directly in the physiological genes, either within regulatory or coding regions, contributing to phenotypic variability in expression or activity of that gene (e.g., POMC, MC4R).
|
|
|
Future Directions for Transcriptome Mapping
|
|---|
No published study to date has evaluated multiple tissues in transcriptome mapping. We expect that evaluation of highly relevant yet diverse tissue types will expose different sets of transcript QTL, even when analyzing the same transcript populations. Discovery of similarities (or differences) across tissues will add power to experimental findings, providing validation especially for cis-acting QTL, and revealing important underlying biology for the traits of interest.
Extension of the paradigm of transcriptome mapping to the proteome and metabolome would also be unique. Although experimental issues (i.e., lack of genome-wide reagents) would render such extensions initially to be on a limited scale, it would represent an important test to determine if QTL profiles underlying protein levels parallel those controlling transcription. The question of whether microarrays are valid indicators of actual protein levels (and hence biological activity) is still of major importance, and such results would test and extend this question in regard to comparison of the underlying polygenic control at each step of the central dogma of biology. In regard to metabolomics, discovery of QTL controlling carbohydrate, glucose, and lipid and fatty acid metabolism would have particular relevance to obesity research. Our preliminary efforts in this regard have revealed a large number of QTL underlying de novo fatty acid synthesis (Allan, 2003; our unpublished data).
|
Statistical Issues
|
|---|
Although transcriptome mapping does not present the need for development of new statistical paradigms relative to traditional transcriptome analysis and QTL mapping, several sophisticated analyses will be required to extract full value from the enormous amount of collected data, and gain valuable insight into genetic control of gene expression. As recently noted by Darvasi (2003), "I expect that the combining of genetic information and gene expression will hasten the day when genomics delivers on its promise to improve health care. But we must continue striving to develop and apply sophisticated analytical tools for interpreting the vast, complex data sets that are being produced with modern genomic technologies."
Traditionally, these would include analysis of sex interaction in genetic control of the transcriptome, determination of the role of genomic imprinting in control of genome-wide gene expression, and evaluation of within- and between-founder line genetic variance. Perhaps more importantly, transcriptome mapping represents an extremely challenging scenario for thorough implementation of multiple trait analyses. Initially, this may be best implemented for specific situations, such as genes that are part of the same known pathway, and genes measured in different tissues. Also, when single-trait expression QTL seem to map to the same region, multi-trait analyses can be used to improve precision and significance.
Most QTL analyses have ignored the potential role of gene interactions in the control of trait variation. However, there is mounting evidence that analyses specifically testing for epistasis can both identify QTL that are not otherwise found and explain a greater proportion of the genetic variation (Shimomura et al., 2001; Leamy et al., 2002; Carlborg et al., 2003). In the context of understanding the genetic control of the transcriptome and proteome, it is critical that we extend analyses to include epistasis as this is likely to play an important role in interpretation of the network of gene interactions that contribute to obesity.
When transcriptome mapping is implemented within a very large structured pedigree, the opportunity exists to merge traditional quantitative genetic analyses, such as genetic parameter estimation, with QTL analysis. This would enable estimation of genetic correlations among, and heritabilities of, the sub- (e.g., transcriptional, proteomic, endocrine) and endpoint phenotypic traits. As with multitrait analysis, data reduction is likely required to make this effort feasible and to enable extraction of meaningful information. For example, heritabilities can be measured for all endpoint phenotypes and for all subphenotypes for which at least one significant QTL is identified. Genetic correlations can then be estimated for the following sets of traits: 1) between each pair of endpoint phenotypes and 2) between each subphenotype for which heritability is estimated significantly different from zero; and between traits in categories 1 and 2.
Genetic parameter estimation will add unique value to this research paradigm. A strong heritability of a subphenotype should coincide with the presence of strong evidence for QTL, providing validation of the process. More importantly, genetic correlation between a subphenotype and body fat levels will be critical to differentiate among, and rank, multiple linked transcriptional and/or proteomic QTL that could represent positional candidate genes for obesity predisposition. Finally, it is interesting to speculate that genetic correlation analysis can be a useful method for clustering of array results and identification of biologically relevant pathways. This clustering would be an extension of the stratification of obesity phenotypic classes based on combined expression phenotypes (Schadt et al., 2003a) and the use of principal components analysis recently proposed by Lan et al. (2003).
Both QTL analysis and genetic parameter estimation, in the context of transcriptome mapping, have immense computational requirements. It is likely that such efforts, when carried out on a large scale, will require recoding of existing programs to run on supercomputers.
|
Implementation
|
|---|
Analysis of complex trait genetic architecture using "quantitative genomics," manifested through the transcriptome (and/or proteome and metabolome) mapping paradigm, can be applied to essentially any QTL mapping experiment where samples with spatial and temporal biological relevance have been stored in an appropriate manner. With regard to obesity, we are applying this approach with the polygenic obese M16 line of mice (Hanrahan et al., 1973; Eisen, 1986) and its nonobese ICR control line (Figure 4). Having identified differences between the lines for a wide variety of traits, including transcriptional, proteomic, and metabolomic phenotypes with relevance to energy balance (Pomp et al., 2002; Allan, 2003), we have established a large F2 QTL mapping population and have phenotyped approximately 1,200 mice for growth, fatness, and feed consumption. Tissues with relevance to energy balance have been stored for subphenotyping, whereas DNA has been extracted and an initial panel of 80 genome-wide informative microsatellites has been genotyped. Furthermore, the F2 was designed to provide for large 3/4- and 1/2-sib families to enable genetic parameter estimation for all phenotypes measured across the population. By applying large-scale transcriptional (e.g., Affymetrix, Santa Clara, CA), proteomic (e.g., BD Biosciences [BD Powerblot], San Diego, CA), and metabolomic (e.g., Lipomics Technologies, West Sacramento, CA) phenotyping to the M16 x ICR F2 QTL mapping population, our goal is to advance the positional cloning of obesity polygenes and begin to understand the genetic architecture of obesity predisposition.
View larger version (71K):
[in this window]
[in a new window]
|
Figure 4. Application of the transcriptome mapping approach to BW and obesity using an F2 cross between a line selected for rapid weight gain (M16, top left) and its unselected control line (ICR, top right). Individuals in the segregating F2 population (images here are graphical representations created from pictures of the two F0 mice) have been phenotyped for BW, body composition and feed intake. Additionally, tissue samples from F2 individuals will be assayed using a microarray containing most expressed murine genes. Alternative parental line forms of DNA markers will segregate and can be tracked in the F2 population, facilitating a QTL analysis for both weight and composition end-point phenotypes and gene expression sub-phenotypes.
|
|
A much broader and extremely powerful platform would be provided by development of a large cohort of recombinant inbred lines developed from a multiway cross of strains representing the majority of phenotypic diversity available in mice (Williams et al., 2002a). A set of 1,000 recombinant inbred lines originating from a cross of eight inbred lines (Figure 5; also see Vogel, 2003) would theoretically achieve 0.1 cM precision (approximately 100,000 bp) when mapping QTL with additive effects of >0.25 SD. Applying transcriptome mapping to this organized assortment of defined recombinational breakpoints would dramatically increase the rate of positional cloning of genes underlying QTL, and would significantly enhance the understanding of the genetic architecture of complex trait predisposition for a wide variety of agriculturally and biomedically relevant phenotypes.
View larger version (59K):
[in this window]
[in a new window]
|
Figure 5. Outline of the production of a recombinant inbred strain originating from a cross of eight inbred lines. Production of approximately 1,000 such recombinant inbred strains would enable very high resolution mapping of QTL, effective dissection of epistatic interactions, and powerful analysis of gene x environment interactions. For specific details, see Williams et al. (2002a; www.complextrait.org/Workshop1.pdf).
|
|
It is prudent to acknowledge that, despite the potential power and breadth of the transcriptome mapping approach, it has important drawbacks and limitations that can and will restrict its utility. We will use examples from recent and well-characterized gene discoveries in livestock species to generally illustrate the limitations of transcriptome mapping, assuming hypothetically that this approach were able to have been applied in each case.
One important issue is that some QTL may not be manifested by changes in steady-state levels of mRNA. Not only would transcriptome mapping fail to identify correct candidate genes underlying such QTL, it may in fact mislead the investigator into examining the wrong candidates. For example, the double muscling phenotype in cattle is caused by mutations in the myostatin gene, but these mutations are not manifested by changes in mRNA levels, but rather by alterations in protein function (Kambadur et al., 1997). Transcriptome mapping would not have identified myostatin as a candidate gene in a resource population segregating the double muscling phenotype, whereas any other expression QTL falling on the cis-diagonal (see Figure 2) in the chromosomal region where double muscling had been mapped, may have been falsely identified as candidate genes. The approach would still, however, provide important information on transcriptional changes that are downstream from the QTL effect and which are important in the context of understanding the overall genetic architecture of the trait.
Because the transcriptome mapping approach relies on gene expression phenotypes, it is critical that selection of both spatial (what tissue) and temporal (when the tissue is collected) coordinates captures as much significant biology as possible. An example of this is clearly demonstrated in the recent finding of a QTL represented by a regulatory mutation in IGF-2 causing a major effect on muscle growth in pigs (Van Laere et al., 2003). Given that this mutation is manifested by gene expression changes in postnatal skeletal and cardiac muscle, but not in fetal muscle or postnatal liver, transcriptome mapping would have been of immediate assistance in finding this mutation only if postnatal muscle was evaluated. In cases such as obesity, multiple tissues are implicated in control of specific pathways that contribute to the end-phenotype, and expression of many genes will vary over time and across environments (e.g., diet). Thus, a thorough transcriptome mapping effort would constitute a massive undertaking involving multiple tissues, time points, environments, and genetic backgrounds that, combined with the high cost of microarrays, is likely beyond the scope of most research budgets.
Given that significant effort and expense are invested in data collection using transcriptome mapping, it is therefore imperative that data and results be made broadly available to the research community. One such powerful and useful environment is provided by WebQTL (Wang et al., 2003a; http://www.webqtl.org/). WebQTL is a Web-based package for complex trait analysis, and a tool for multidimensional searches among large datasets derived from high-throughput analysis techniques. Furthermore, exploring these data sets in a systematic way will be a challenge. This challenge has already been partially addressed by those describing the molecular biology and genetics of simpler eukaryotes. For example, GRID is a database for genetic or molecular interactions among products of yeast genes (Breitkreutz et al., 2003). The same group developed an Osprey Java-based interaction viewer that will display selected subsets of interaction data from GRID in graphical forms. These tools and others like them will be directly applicable to exploration of obesity-related interactions uncovered in transcriptome mapping.
|
Implications
|
|---|
The paradigm of "quantitative genomics," whereby large-scale subphenotyping at the transcriptional, proteomic, and metabolomic levels is performed within the context of a quantitative trait loci mapping population, will be a powerful force in dissecting the genetic architecture of complex trait predisposition. Carcass and body composition traits constitute extremely important considerations for modern livestock production systems where consumer health concerns and marketing perspectives play increasingly prominent roles. An estimated 65% of U.S. adults are overweight, and 31% are obese, substantially increasing the risk for numerous other heath concerns. Understanding the roles of specific loci in genetic susceptibility to obesity is critical to improving human health and quality of life. Enhanced understanding of how complex traits are controlled will also aid in elucidating the nature of predisposition for other diseases, including certain forms of cancer, and will be broadly applicable to many important relevant phenotypes in agriculture and biomedicine.
|
Footnotes
|
|---|
1 This research is a contribution of the Univ. of Nebraska Agric. Res. Div., Lincoln (Journal Series No. 14321) and was supported in part by funds provided through the Hatch Act. The authors are grateful to D. Van Vleck, J. Rocha, K. Elo, R. Williams, D. Threadgill, K. Manly, and C. Haley for useful discussions. Some research in progress discussed in this paper is part of a fruitful and ongoing collaboration with G. Eisen at North Carolina State University. 
2 This article was presented at the 2003 Joint ASAS-ADSA-WSASAS Meeting as part of the Breeding and Genetics symposium.
3 Correspondence: A218 Animal Science (phone: 402-472-6416; fax: 402-472-6362; e-mail: dpomp{at}unlnotes.unl.edu).
Received for publication October 26, 2003.
Accepted for publication January 28, 2004.
|
Literature Cited
|
|---|
Allan, M. F. 2003. Integration of genomic, proteomic and metabolomic analyses towards polygene discovery in mice selected for 3 to 6 week weight gain. Ph.D. Diss., Univ. of Nebraska, Lincoln.
Barsh, G. S., and M. W. Schwartz. 2002. Genetic approaches to studying energy balance: Perception and integration. Nat. Rev. Genet. 3:589–600.[Medline]
Bray, G. A., and L. A. Tartaglia. 2000. Medicinal strategies in the treatment of obesity. Nature 404:672– 677.[Medline]
Bray, G., and C. Bouchard. 1997. Genetics of human obesity: Research directions. FASEB J. 11:937–945.[Abstract]
Breitkreutz, B. J., C. Stark, and M. Tyers. 2003. The GRID: The general repository for interaction datasets. Genome Biol. 4:R23.[Medline]
Brem, R. B., G. Yvert, R. Clinton, and L. Kruglyak. 2002. Genetic dissection of transcriptional regulation in budding yeast. Science 296:752–755.[Abstract/Free Full Text]
Brockmann, G. A., and M. R. Bevova. 2002. Using mouse models to dissect the genetics of obesity. Trends Genet. 18:367–376.[Medline]
Brockmann, G. A., J. Kratzsch, C. S. Haley, U. Renne, and M. Schwerin. 2000. Single QTL effects, epistasis, and pleiotropy account for two-thirds of the phenotypic F2 variance of growth and obesity in DU6i x DBA/2 mice. Genome Res. 10:1941–1957.[Abstract/Free Full Text]
Carlborg, Ö., S. Kerje, K. Schutz, L. Jacobsson, P. Jensen, and L. Andersson. 2003. A global search reveals epistatic interaction between QTLs for early growth in the chicken. Genome Res. 13:413–421.[Abstract/Free Full Text]
Chagnon, Y. C., T. Rankinen, E. E. Snyder, S. J. Weisnagel, L. Perusse, and C. Bouchard. 2003. The human obesity gene map: The 2002 update. Obes. Res. 11:313–367.[Medline]
Chua, S. C., Jr., 1997. Monogenic models of obesity. Behav. Genet. 27:277–284.[Medline]
Comuzzie, A. G., and D. B. Allison. 1998. The search for human obesity genes. Science 280:1374–1377.[Abstract/Free Full Text]
Consoli, L, A. Lefevre, M. Zivy, D. de Vienne, and C. Damerval. 2002. QTL analysis of proteome and transcriptome variations for dissecting the genetic architecture of complex traits in maize. Plant Mol. Biol. 48:575–581.[Medline]
Darvasi, A. 2003. Genomics: Gene expression meets genetics. Nature. 422:269–270.[Medline]
Damerval, C., A. Maurice, J. M. Josse, and D. de Vienne. 1994. Quantitative trait loci underlying gene product variation: a novel perspective for analyzing regulation of genome expression. Genetics 137:289–301.[Abstract]
de Vienne, D., A. Maurice, J. M. Josse, A. Leonardi, and C. Damerval. 1994. Mapping factors controlling genetic expression. Cell. Mol. Biol. (Noisy-le-grand) 40:29–39.
Dietrich, W. F., J. Miller, R. Steen, M. A. Merchant, D. Damronboles, Z. Husain, R. Dredge, M. J. Daly, K. A. Ingalls, T. J. Oconnor, C. A. Evans, M. M. Deangelis, D. M Levinson, L. Kruglyak, N. Goodman, N. G. Copeland, N. A. Jenkins, T. L. Hawkins, L. Stein, D. C. Page, and E. S. Lander. 1996. A comprehensive genetic map of the mouse genome. Nature 380:149– 152.[Medline]
Doerge, R. W. 2002. Mapping and analysis of quantitative trait loci in experimental populations. Nat. Rev. Genet. 3:43–52.[Medline]
Dumas, P., Y. Sun, G. Corbell, S. Trenblay, Z. Pausova, V. Kren, D. Krenova, M. Pravence, P. Hamet, and J. Treblay. 2000. Mapping of quantitative trait loci (QTL) of differential stress gene expression in rat recombinant inbred strains. J. Hypertens. 18:545–551.[Medline]
Eisen, E. J. 1986. Maturing patterns of organ weights in mice selected for rapid postweaning gain. Theor. Appl. Genet. 73:148–157.
Eisen, E. J. 1989. Selection experiments for body composition in mice and rats: A review. Livest. Prod. Sci. 23:17–32.
Elo, K. 2003. Towards integrated genomic approaches—Lessons from energy metabolism studies. Pages 222–227 in Proc. of the NJF’s 22nd Cong., Turku, Finland.
Esterbauer, H., C. Schneitler, H. Oberkofler, C. Ebenbichler, B. Paulweber, F. Sandhofer, G. Ladurner, E. Hell, A. D. Strosberg, J. R. Patsch, F. Krempler, and W. Patsch. 2001. A common polymorphism in the promoter of UCP2 is associated with decreased risk of obesity in middle-aged humans. Nat. Genet. 28:178–183.[Medline]
Farooqi, I. S., J. M. Keogh, G. S. Yeo, E. J. Lank, T. Cheetham, and S. O’Rahilly. 2003. Clinical spectrum of obesity and mutations in the melanocortin 4 receptor gene. N. Engl. J. Med. 348:1085–1095.[Abstract/Free Full Text]
Festing, M. F. W. 1979. The inheritance of obesity in animal models of obesity. Pages 15–37 in Animal Models of Obesity. M. F. W. Festing, ed. MacMillan Press, London, U.K.
Finkelstein, E. A., I. C. Fiebelkorn, and G. Wang. 2003. National medical spending attributable to overweight and obesity: How much, and who’s paying? Health Affairs W3:219–226.
Flegal, K. M., M. D. Carroll, C. L. Ogden, and C. L. Johnson. 2002. Prevalence and trends in obesity among US adults, 1999–2000. J. Am. Med. Assoc. 288:1723–1727.[Abstract/Free Full Text]
Garrod, A. 1902. The incidence of alkaptonuria: A study in chemical individuality. Lancet 2:1616–1620.
Haldane, J. B., A. D. Sprunt, and N. M. Haldane. 1915. Reduplication in mice. Science 5:133–135.
Haley, C. S., S. A. Knott, and J. M. Elsen. 1994. Mapping quantitative trait loci in crosses between outbred lines using least squares. Genetics 136:1195–1207.[Abstract]
Hanrahan, J. P., E. J. Eisen, and J. E. Legates. 1973. Effects of population size and selection intensity on short-term response to selection for post-weaning gain in mice. Genetics 73:513–530.[Abstract/Free Full Text]
Heo, M., R. L. Leibel, B. B. Boyer, W. K. Chung, M. Koulu, M. K. Karvonen, U. Pesonen, A. Rissanen, M. Laakso, M. I. Uusitupa, Y. Chagnon, C. Bouchard, P. A. Donohoue, T. L. Burns, A. R. Shuldiner, K. Silver, R. E. Andersen, O. Pedersen, S. Echwald, T. I. Sorensen, P. Behn, M. A. Permutt, K. B. Jacobs, R. C. Elston, D. J. Hoffman, and D. B. Allison. 2001. Pooling analysis of genetic data: the association of leptin receptor (LEPR) polymorphisms with variables related to human adiposity. Genetics 159:1163–1178.[Abstract/Free Full Text]
Hixson, J. E., L. Almasy, S. Cole, S. Birnbaum, B. D. Mitchell, M. C. Mahaney, M. P. Stern, J. W. MacCluer, J. Blangero, and A. G. Comuzzie. 1999. Normal variation in leptin levels is associated with polymorphisms in the proopiomelanocortin gene, POMC. J. Clin. Endocrinol. Metab. 84:3187–3191.[Abstract/Free Full Text]
Hilbert, P., K. Lindpainter, J. S. Beckmann, T. Serikawa, F. Soubrier, C. Dubay, P. Cartwright, B. De Gouyon, C. Julie, S. Takahasi, M. Vincent, D. Ganten, M. Georges, and G. M. Lathrop. 1991. Chromosomal mapping of two genetic loci associated with blood-pressure regulation in hereditary hypertensive rats. Nature 353:521–529.[Medline]
Hitzemann, R., B. Malmanger, C. Reed, M. Lawler, B. Hitzemann, S. Coulombe, K. Buck, B. Rademacher, N. Walter, Y. Polyakov, J. Sikela, R. W. Williams, J. Flint, and C. Talbot. 2003. A strategy for integration of QTL, gene expression, and sequence analyses. Mamm. Genome 14:733–747.[Medline]
Jacob, H. J., K. Lindpainter, S. E. Lincoln, K. Kusumi, R. K. Bunker, Y.-P. Mao, D. Ganten, V. J. Dzau, and E. S. Lander. 1991. Genetic mapping of a gene causing hypertension in the stroke-prone rat. Cell 67:213–224.[Medline]
Jansen, R. C. 2003. Studying complex biological systems using multifactorial perturbation. Nature Rev. Genet. 4:145–151.[Medline]
Jansen, R. C., and J. P. Nap. 2001. Genetical genomics: The added value from segregation. Trends Genet. 17:388–391.[Medline]
Kambadur R., M. Sharma, T. P. Smith, and J. J. Bass. 1997. Mutations in myostatin (GDF8) in double-muscled Belgian Blue and Piedmontese cattle. Genome Res. 7: 910–916.[Abstract/Free Full Text]
Korstanje, R., and B. Paigen. 2002. From QTL to gene: the harvest begins. Nat. Genet. 31:235–236.[Medline]
Koza, R. A., S. M. Hohmann, C. Guerra, M. Rossmeisl, and L. P. Kozak. 2000. Synergistic gene interactions control the induction of the mitochondrial uncoupling protein (Ucp1) gene in white fat tissue. J. Biol. Chem. 275:34486–34492.[Abstract/Free Full Text]
Lan, H., J. P. Stoehr, S. T. Nadler, K. L. Schueler, B. S. Yandell, and A. D. Attie. 2003. Dimension reduction for mapping mRNA abundance as quantitative traits. Genetics 164:1607–1614.[Abstract/Free Full Text]
Lander, E. S., and D. Botstein. 1989. Mapping Mendelian factors underlying quantitative traits using RFLP linkage maps. Genetics 121:185–199.[Abstract/Free Full Text]
Leamy, L. J., E. J. Routman, and J. M. Cheverud. 2002 An epistatic genetic basis for fluctuating asymmetry of mandible size in mice. Evolution Int. J. Org. Evolution. 56:642–653.[Medline]
Lembertas, A. V., L. Perusse, Y. C. Chagnon, J. S. Fisler, and C. H. Warden. 1997. Identification of an obesity quantitative trait locus on mouse chromosome 2 and evidence of linkage to body fat and insulin on the human homologous region 20q. J. Clin. Invest. 100:1240–1247.[Medline]
Martin, B., J. Nienhuis, G. King, and A. Schaefer. 1989. Restriction fragment length polymorphisms associated with water use efficiency in tomato. Science 243:1725–1728.[Abstract/Free Full Text]
McPherron, A. C., and S. J. Lee. 1996. The transforming growth factor b superfamily. Pages 357–393 in Growth Factors and Cytokines in Health and Disease. D. LeRoith and C. Bondy, ed. JAI Press, Greenwich, CT.
Mehrabian, M., P. Z. Wen, J. Fisler, R. C. Davis, and A. J. Lusis. 1998. Genetic loci controlling body fat, lipoprotein metabolism, and insulin levels in a multifactorial mouse model. J. Clin. Invest. 101:2485–2496.[Medline]
Michuad, E. J., S. J. Bultman, M. L. Klebig, M. J. van Vugt, L. S. Stubbs, L. B. Russell, and R. P. Woychik. 1994. A molecular model for the genetic and phenotypic characteristics of the mouse lethal yellow (Ay) mutation. Proc. Natl. Acad. Sci. 91:2562–2566.[Abstract/Free Full Text]
Montooth, K. L., J. H. Marden, and A. G. Clark. 2003. Mapping determinants of variation in energy metabolism, respiration and flight in Drosophila. Genetics 165:623–635.[Abstract/Free Full Text]
Nadeau, J. H., and W. N. Frankel. 2000. The roads from phenotypic variation to gene discovery: mutagenesis versus QTLs. Nat. Genet. 25:381–384.[Medline]
Oliver, B., M. Parisi, and D. Clark. 2002. Gene expression neighborhoods. J. Biol. 1:4.[Medline]
Paterson, A. H., E. S. Lander, J. D. Hewitt, S. Peterson, S. E. Lincoln, and S. D. Tanksley. 1988. Resolution of quantitative traits into Mendelian factors by using a complete linkage map of restriction fragment length polymorphisms. Nature 335:721–726.[Medline]
Pomp, D. 1997. Genetic dissection of obesity in polygenic animal models. Behav. Genet. 27:285–306.[Medline]
Pomp, D. 1999. Animal models of obesity. Mol. Med. Today 5:459–460.[Medline]
Pomp, D., N. Jerez, M. F. Allan, and E. J. Eisen. 2002. Integrated genomic, proteomic and metabolomic dissection of polygenic selection response for murine growth and fatness. Proc. of the 7th World Cong. on Genet. Applied to Livest. Prod., Montpellier, France.
Rich, S. S. 1990. Mapping genes in diabetes. Diabetes 39:1315–1319.[Abstract]
Rocha, J. L., L. D. Van Vleck, E. J. Eisen, and D. Pomp. 2004a. A large-sample QTL study in mice: I. Growth. Mamm. Genome 15:83–99.
Rocha, J. L., L. D. Van Vleck, E. J. Eisen, and D. Pomp. 2004b. A large-sample QTL study in mice: II. Organ and body composition traits. Mamm. Genome 15:100–113.
Rohrer, G. A., T. H. Wise, D. D. Lunstra, and J. J. Ford. 2001. Identification of genomic regions controlling plasma FSH concentrations in Meishan-White Composite boars. Physiol. Genomics 6:145–151.[Abstract/Free Full Text]
Rosen, C. J., G. A. Churchill, L. R. Donahue, K. L. Shultz, J. K. Burgess, D. R. Powell, C. Ackert, and W. G. Beamer. 2000. Mapping quantitative trait loci for serum insulin-like growth factor-1 levels in mice. Bone 27:521–528.[Medline]
Sax, K. 1923. The association of size differences with seed-coat pattern and pigmentation in Phaseolusvulgaris. Genetics 8:552–560.[Free Full Text]
Schadt, E. E., S. A. Monks, T. A. Drake, A. J. Lusis, N. Che, V. Colinayo, T. G. Ruff, S. B. Milligan, J. R. Lamb, G. Cavet, P. S. Linsley, M. Mao, R. B. Stoughton, and S. H. Friend. 2003a. Genetics of gene expression surveyed in maize, mouse and man. Nature 422:397–402.[Medline]
Schadt, E. E., S. A. Monks, and S. H. Friend. 2003b. A new paradigm for drug discovery: Integrating clinical, genetic, genomic and molecular phenotype data to identify drug targets. Biochem. Soc. Trans. 31:437–443.[Medline]
Schwartz, M. W., S. C. Woods, R. J. Seeley, G. S. Barsh, D. G. Baskin, and R. L. Leibel. 2003. Is the energy homeostasis system inherently biased toward weight gain? Diabetes 52:232–238.[Abstract/Free Full Text]
Shimomura, K., S. S. Low-Zeddues, D. P. King, T. D. L. Steeves, A. Whiteley, J. Kushla, P. D. Zemenides, A. Lin, M. H. Vitaterna, G. A. Churchill, and J. S. Takahashi. 2001. Genome-wide epistatic interaction analysis reveals complex genetic determinants of circadian behavior in mice. Genome Res. 11:959–980.[Abstract/Free Full Text]
Spellman, P. T., and G. M. Rubin. 2002. Evidence for large domains of similarly expressed genes in the Drosophila genome. J. Biol. 1:5.[Medline]
Sturtevant, A. H. 1913. The linear arrangement of six sex-linked factors in Drosophila, as shown by their mode of association. J. Exp. Zool. 14:43–59.
Van Laere, A. S., M. Nguyen, M. Braunschweig, C. Nezer, C. Collette, L. Moreau, A. L. Archibald, C. S. Haley, N. Buys, M. Tally, G. Andersson, M. Georges, and L. A. Andersson. 2003. A regulatory mutation in IGF2 causes a major QTL effect on muscle growth in the pig. Nature 425:832–836.[Medline]
Vogel, G. 2003. Scientists dream of 1001 complex mice. Science 301:456–457.[Abstract/Free Full Text]
Wang, J., R. W. Williams, E. J. Chesler, and K. F. Manly. 2003a. WebQTL: Web-based complex trait analysis. Neuroinformatics 1:299–308.[Medline]
Wang, Y. X., C. H. Lee, S. Tiep, R. T. Yu, J. Ham, H. Kang, and R. M. Evans. 2003b. Peroxisome-Proliferator-activated receptor delta activates fat metabolism to prevent obesity. Cell 113:159–170.[Medline]
Wayne, M. L., and L. M. McIntyre. 2002. Combining mapping and arraying: An approach to candidate gene identification. Proc. Natl. Acad. Sci. 99:14903–14906.[Abstract/Free Full Text]
Wesolowski, S. R., M. F. Allan, M. K. Nielsen, and D. Pomp. 2003. Evaluation of hypothalamic gene expression in mice divergently selected for heat loss. Physiol. Genomics 13:129–137.[Abstract/Free Full Text]
Williams, R. W., K. W. Broman, J. M. Cheverud, G. A. Churchill, R. W. Hitzemann, K. W. Hunter, J. D. Mountz, D. Pomp, R. H. Reeves, L. C. Schalkwyk, and D. W. Threadgill. 2002a. A collaborative cross for high-precision complex trait analysis. 1st Workshop Rep. of the Complex Trait Consortium: September 2002. Available: www.complextrait.org/Workshop1.pdf. Accessed November, 2003.
Williams, R. W., S. Shou, L. Lu, Y. Qu, J. Wang, K. F. Manly, E. J. Chesler, H. C. Hsu, J. D. Mountz, and D. W. Threadgill. 2002b. Genomic analysis of transcriptional networks: Combining microarrays with complex trait analysis. Proc. of the 1st Annu. Complex Trait Consort. Mtg., Memphis, TN. Available: http//www.complextrait.org/ctc2002/williams.html. Accessed November, 2003.
Yvert, G., R. B. Brem, J. Whittle, J. M. Akey, E. Foss, E. N. Smith, R. Mackelprang, and L. Kruglyak. 2003. Trans-acting regulatory variation in Saccharomyces cerevisiae and the role of transcription factors. Nat. Genet. 35:57–64.[Medline]
Zeng, Z.-B. 1993. Theoretical basis for separation of multiple linked gene effects in mapping quantitative trait loci. Proc. Natl. Acad. Sci. USA 90:10972–10976.[Abstract/Free Full Text]
Zeng, Z.-B. 1994. Precision mapping of quantitative trait loci. Genetics 136:1457–1468.[Abstract]
Zhang, Y., R. Proenca, M. Maffei, M. Barone, L. Leopold, and J. Friedman. 1994. Positional cloning of the mouse obese gene and its human homologue. Nature 372:425–432.[Medline]
This article has been cited by other articles:
|
|
|
|
 
E. M. Sellner, J. W. Kim, M. C. McClure, K. H. Taylor, R. D. Schnabel, and J. F. Taylor
BOARD-INVITED REVIEW: Applications of genomic information in livestock
J Anim Sci,
December 1, 2007;
85(12):
3148 - 3158.
[Abstract]
[Full Text]
[P |
Top
Abstract
Introduction
