Exploiting naturally occurring DNA variation and molecular profiling data to dissect disease and drug response traits

Genetic variation in histone modifications and gene expression identifies regulatory variants in the mammary gland of cattle

BACKGROUND

Causal variants for complex traits, such as eQTL are often found in non-coding regions of the genome, where they are hypothesised to influence phenotypes by regulating gene expression. Many regulatory regions are marked by histone modifications, which can be assayed by chromatin immunoprecipitation followed by sequencing (ChIP-seq). Sequence reads from ChIP-seq form peaks at putative regulatory regions, which may reflect the amount of regulatory activity at this region. Therefore, eQTL which are also associated with differences in histone modifications are excellent candidate causal variants.

RESULTS

We assayed the histone modifications H3K4Me3, H3K4Me1 and H3K27ac and mRNA in the mammary gland of up to 400 animals. We identified QTL for peak height (histone QTL), exon expression (eeQTL), allele specific expression (aseQTL) and allele specific binding (asbQTL). By intersecting these results, we identify variants which may influence gene expression by altering regulatory regions of the genome, and may be causal variants for other traits. Lastly, we find that these variants are found in putative transcription factor binding sites, identifying a mechanism for the effect of many eQTL.

CONCLUSIONS

We find that allele specific and traditional QTL analysis often identify the same genetic variants and provide evidence that many eQTL are regulatory variants which alter activity at regulatory regions of the bovine genome. Our work provides methodological and biological updates on how regulatory mechanisms interplay at multi-omics levels.

Scientific and Artful Voices of Resilience

Resilience is a fluid trait that is triggered by personal experience. It is, arguably, a necessity for a scientist. What is it? You know it, when you see it. One thing is for certain: resilience reflects the dynamic toggle between change and an individual’s identity.

An update on clonality: what smooth muscle cell type makes up the atherosclerotic plaque?

Almost 50 years ago, Earl Benditt and his son John described the clonality of the atherosclerotic plaque. This led Benditt to propose that the atherosclerotic lesion was a smooth muscle neoplasm, similar to the leiomyomata seen in the uterus of most women. Although the observation of clonality has been confirmed many times, interest in the idea that atherosclerosis might be a form of neoplasia waned because of the clinical success of treatments for hyperlipemia and because animal models have made great progress in understanding how lipid accumulates in the plaque and may lead to plaque rupture. Four advances have made it important to reconsider Benditt's observations. First, we now know that clonality is a property of normal tissue development. Second, this is even true in the vessel wall, where we now know that formation of clonal patches in that wall is part of the development of smooth muscle cells that make up the tunica media of arteries. Third, we know that the intima, the "soil" for development of the human atherosclerotic lesion, develops before the fatty lesions appear. Fourth, while the cells comprising this intima have been called "smooth muscle cells", we do not have a clear definition of cell type nor do we know if the initial accumulation is clonal. As a result, Benditt's hypothesis needs to be revisited in terms of changes in how we define smooth muscle cells and the quite distinct developmental origins of the cells that comprise the muscular coats of all arterial walls. Finally, since clonality of the lesions is real, the obvious questions are do these human tumors precede the development of atherosclerosis, how do the clones develop, what cell type gives rise to the clones, and in what ways do the clones provide the soil for development and natural history of atherosclerosis?

HTR7 Mediates Serotonergic Acute and Chronic Itch

Chronic itch is a prevalent and debilitating condition for which few effective therapies are available. We harnessed the natural variation across genetically distinct mouse strains to identify transcripts co-regulated with itch behavior. This survey led to the discovery of the serotonin receptor HTR7 as a key mediator of serotonergic itch. Activation of HTR7 promoted opening of the ion channel TRPA1, which in turn triggered itch behaviors. In addition, acute itch triggered by serotonin or a selective serotonin reuptake inhibitor required both HTR7 and TRPA1. Aberrant serotonin signaling has long been linked to a variety of human chronic itch conditions, including atopic dermatitis. In a mouse model of atopic dermatitis, mice lacking HTR7 or TRPA1 displayed reduced scratching and skin lesion severity. These data highlight a role for HTR7 in acute and chronic itch and suggest that HTR7 antagonists may be useful for treating a variety of pathological itch conditions.

Evolving toward a human-cell based and multiscale approach to drug discovery for CNS disorders

A disruptive approach to therapeutic discovery and development is required in order to significantly improve the success rate of drug discovery for central nervous system (CNS) disorders. In this review, we first assess the key factors contributing to the frequent clinical failures for novel drugs. Second, we discuss cancer translational research paradigms that addressed key issues in drug discovery and development and have resulted in delivering drugs with significantly improved outcomes for patients. Finally, we discuss two emerging technologies that could improve the success rate of CNS therapies: human induced pluripotent stem cell (hiPSC)-based studies and multiscale biology models. Coincident with advances in cellular technologies that enable the generation of hiPSCs directly from patient blood or skin cells, together with methods to differentiate these hiPSC lines into specific neural cell types relevant to neurological disease, it is also now possible to combine data from large-scale forward genetics and post-mortem global epigenetic and expression studies in order to generate novel predictive models. The application of systems biology approaches to account for the multiscale nature of different data types, from genetic to molecular and cellular to clinical, can lead to new insights into human diseases that are emergent properties of biological networks, not the result of changes to single genes. Such studies have demonstrated the heterogeneity in etiological pathways and the need for studies on model systems that are patient-derived and thereby recapitulate neurological disease pathways with higher fidelity. In the context of two common and presumably representative neurological diseases, the neurodegenerative disease Alzheimer's Disease, and the psychiatric disorder schizophrenia, we propose the need for, and exemplify the impact of, a multiscale biology approach that can integrate panomic, clinical, imaging, and literature data in order to construct predictive disease network models that can (i) elucidate subtypes of syndromic diseases, (ii) provide insights into disease networks and targets and (iii) facilitate a novel drug screening strategy using patient-derived hiPSCs to discover novel therapeutics for CNS disorders.

Dominant effects of the Huntington's disease HTT CAG repeat length are captured in gene-expression data sets by a continuous analysis mathematical modeling strategy

In Huntington's disease (HD), the size of the expanded HTT CAG repeat mutation is the primary driver of the processes that determine age at onset of motor symptoms. However, correlation of cellular biochemical parameters also extends across the normal repeat range, supporting the view that the CAG repeat represents a functional polymorphism with dominant effects determined by the longer allele. A central challenge to defining the functional consequences of this single polymorphism is the difficulty of distinguishing its subtle effects from the multitude of other sources of biological variation. We demonstrate that an analytical approach based upon continuous correlation with CAG size was able to capture the modest (∼21%) contribution of the repeat to the variation in genome-wide gene expression in 107 lymphoblastoid cell lines, with alleles ranging from 15 to 92 CAGs. Furthermore, a mathematical model from an iterative strategy yielded predicted CAG repeat lengths that were significantly positively correlated with true CAG allele size and negatively correlated with age at onset of motor symptoms. Genes negatively correlated with repeat size were also enriched in a set of genes whose expression were CAG-correlated in human HD cerebellum. These findings both reveal the relatively small, but detectable impact of variation in the CAG allele in global data in these peripheral cells and provide a strategy for building multi-dimensional data-driven models of the biological network that drives the HD disease process by continuous analysis across allelic panels of neuronal cells vulnerable to the dominant effects of the HTT CAG repeat.

A systematic approach to multifactorial cardiovascular disease: causal analysis

The combination of systems biology and large data sets offers new approaches to the study of cardiovascular diseases. These new approaches are especially important for the common cardiovascular diseases that have long been described as multifactorial. This promise is undermined by biologists' skepticism of the spider web-like network diagrams required to analyze these large data sets. Although these spider webs resemble composites of the familiar biochemical pathway diagrams, the complexity of the webs is overwhelming. As a result, biologists collaborate with data analysts whose mathematical methods seem much like those of experts using Ouija boards. To make matters worse, it is not evident how to design experiments when the network implies that many molecules must be part of the disease process. Our goal is to remove some of this mystery and suggest a simple experimental approach to the design of experiments appropriate for such analysis. We will attempt to explain how combinations of data sets that include all possible variables, graphical diagrams, complementation of different data sets, and Bayesian analyses now make it possible to determine the causes of multifactorial cardiovascular disease. We will describe this approach using the term causal analysis. Finally, we will describe how causal analysis is already being used to decipher the interactions among cytokines as causes of cardiovascular disease.

Identification of cis-regulatory variation influencing protein abundance levels in human plasma

Proteins are central to almost all cellular processes, and dysregulation of expression and function is associated with a range of disorders. A number of studies in human have recently shown that genetic factors significantly contribute gene expression variation. In contrast, very little is known about the genetic basis of variation in protein abundance in man. Here, we assayed the abundance levels of proteins in plasma from 96 elderly Europeans using a new aptamer-based proteomic technology and performed genome-wide local (cis-) regulatory association analysis to identify protein quantitative trait loci (pQTL). We detected robust cis-associations for 60 proteins at a false discovery rate of 5%. The most highly significant single nucleotide polymorphism detected was rs7021589 (false discovery rate, 2.5 × 10(-12)), mapped within the gene coding sequence of Tenascin C (TNC). Importantly, we identified evidence of cis-regulatory variation for 20 previously disease-associated genes encoding protein, including variants with strong evidence of disease association show significant association with protein abundance levels. These results demonstrate that common genetic variants contribute to the differences in protein abundance levels in human plasma. Identification of pQTLs will significantly enhance our ability to discover and comprehend the biological and functional consequences of loci identified from genome-wide association study of complex traits. This is the first large-scale genetic association study of proteins in plasma measured using a novel, highly multiplexed slow off-rate modified aptamer (SOMAmer) proteomic platform.

Environmental and genetic perturbations reveal different networks of metabolic regulation

Measurement of metabolic and physiological parameters in replicated crosses of Drosophila melanogaster inbred lines reveals that environmental and genetic perturbations uncover substantially different networks of metabolic regulation.

We collected extensive data on enzyme activities and physiological parameters from replicated crosses of D. melanogaster inbred lines.

We implemented a multivariate hierarchical Bayesian model to separately assess genetic and environmental covariation among system components and infer metabolic regulatory networks.

Networks revealed by both environmental and genetic perturbations are similar among populations and between sexes.

Environmental and genetic networks differ substantially, suggesting that environmental changes and mutations would have different systemic effects even when their primary targets are the same.

Progress in systems biology depends on accurate descriptions of biological networks. Connections in a regulatory network are identified as correlations of gene expression across a set of environmental or genetic perturbations. To use this information to predict system behavior, we must test how the nature of perturbations affects topologies of networks they reveal. To probe this question, we focused on metabolism of Drosophila melanogaster. Our source of perturbations is a set of crosses among 92 wild-derived lines from five populations, replicated in a manner permitting separate assessment of the effects of genetic variation and environmental fluctuation. We directly assayed activities of enzymes and levels of metabolites. Using a multivariate Bayesian model, we estimated covariance among metabolic parameters and built fine-grained probabilistic models of network topology. The environmental and genetic co-regulation networks are substantially the same among five populations. However, genetic and environmental perturbations reveal qualitative differences in metabolic regulation, suggesting that environmental shifts, such as diet modifications, produce different systemic effects than genetic changes, even if the primary targets are the same.

The age of the "ome": genome, transcriptome and proteome data set collection and analysis

The current state of human genetic studies is both a marvel and a morass. A marvel in that with the completion of the human genome sequence, projects that used to take years now take months or weeks; however, this creates a wealth of data concomitant to a black hole of meaning. In terms of the well used analogy: the human genome sequence is a library in an ancient language with no Rosetta stone. Researchers have readily exploited the human genome map and thousands of candidate gene studies for a multitude of diseases have been performed. However, many of those studies have found that the variants associated with disease risk are not obvious coding changes. The question now becomes: what do these associations mean? One approach to the downstream mapping of associations is to use additional information to map which variant might truly be causative of risk and what that risk variant is doing. This review will summarize the current state of both data set collection and analysis for the understanding of DNA variants and their downstream effects on transcripts and proteins. This article is part of a Special Issue entitled 'Transcriptome'.

Causal modeling using network ensemble simulations of genetic and gene expression data predicts genes involved in rheumatoid arthritis

Tumor necrosis factor α (TNF-α) is a key regulator of inflammation and rheumatoid arthritis (RA). TNF-α blocker therapies can be very effective for a substantial number of patients, but fail to work in one third of patients who show no or minimal response. It is therefore necessary to discover new molecular intervention points involved in TNF-α blocker treatment of rheumatoid arthritis patients. We describe a data analysis strategy for predicting gene expression measures that are critical for rheumatoid arthritis using a combination of comprehensive genotyping, whole blood gene expression profiles and the component clinical measures of the arthritis Disease Activity Score 28 (DAS28) score. Two separate network ensembles, each comprised of 1024 networks, were built from molecular measures from subjects before and 14 weeks after treatment with TNF-α blocker. The network ensemble built from pre-treated data captures TNF-α dependent mechanistic information, while the ensemble built from data collected under TNF-α blocker treatment captures TNF-α independent mechanisms. In silico simulations of targeted, personalized perturbations of gene expression measures from both network ensembles identify transcripts in three broad categories. Firstly, 22 transcripts are identified to have new roles in modulating the DAS28 score; secondly, there are 6 transcripts that could be alternative targets to TNF-α blocker therapies, including CD86 - a component of the signaling axis targeted by Abatacept (CTLA4-Ig), and finally, 59 transcripts that are predicted to modulate the count of tender or swollen joints but not sufficiently enough to have a significant impact on DAS28.

The collection and analysis of clinical data has played a key role in providing insights into the diagnosis, prognosis and treatment of disease. However, it is imperative that molecular and genetic data also be collected and integrated into the creation of network models, which capture underlying mechanisms of disease and can be interrogated to elucidate previously unknown biology. Bringing data from the clinic to the bench completes the cycle of translational research, which we demonstrate with this work. We built disease models from genetics, whole blood gene expression profiles and the component clinical measures of rheumatoid arthritis using a data-driven approach that leverages supercomputing. Genetic factors can be utilized as a source of perturbation to the system such that causal connections between genetics, molecular entities and clinical outcomes can be inferred. The existing TNF-α blocker treatments for rheumatoid arthritis are only effective for approximately 2/3 of the affected population. We identified novel therapeutic intervention points that may lead to the development of alternatives to TNF-α blocker treatments. We believe this approach will provide improved drug discovery programs, new insights into disease progression, increased drug efficacy and novel biomarkers for chronic and complex diseases.

Identification of quantitative trait loci underlying proteome variation in human lymphoblastoid cells

Population-based variability in protein expression patterns, especially in humans, is often observed but poorly understood. Moreover, very little is known about how interindividual genetic variation contributes to protein expression patterns. To begin to address this, we describe elements of technical and biological variations contributing to expression of 544 proteins in a population of 24 individual human lymphoblastoid cell lines that have been extensively genotyped as part of the International HapMap Project. We determined that expression levels of 10% of the proteins were tightly correlated to cell doubling rates. Using the publicly available genotypes for these lymphoblastoid cell lines, we applied a genetic association approach to identify quantitative trait loci associated with protein expression variation. Results identified 24 protein forms corresponding to 15 proteins for which genetic elements were responsible for >50% of the expression variation. The genetic variation associated with protein expression levels were located in cis with the gene coding for the transcript of the protein for 19 of these protein forms. Four of the genetic elements identified were coding non-synonymous single nucleotide polymorphisms that resulted in migration pattern changes in the two-dimensional gel. This is the first description of large scale proteomics analysis demonstrating the direct relationship between genome and proteome variations in human cells.

Classical twin design in modern pharmacogenomics studies

Response to medication is highly variable, unpredictable and, at times, may be fatal. All drugs are more effective in certain groups of the population while showing no or minimal benefit in other groups. Although the current data on the subject are piecemeal, anecdotal evidence suggests that, in line with other common multifactorial traits, a myriad of genomic as well as environmental factors underpin population variability in drug response. Pharmacogenomics is the study of how variations in the human genome affect the variability in response to medication. Efforts to personalize treatment based on results from pharmacogenomics studies have the potential to increase efficacy, lower the overall cost of treatment, and decrease the incidence of adverse drug reactions, and are one of the major challenges of the modern era. The classical twin design has traditionally been used to assess the relative contribution of genetic and environmental factors to population variation in common, complex phenotypes, including drug response. Twins are not commonly regarded as providing the optimal design in genomic studies. However, we argue that, through their precise ‘matching’ for confounding variables (age, sex, cohort and common environmental effects), their amenability to numerous nonclassical study designs (genome-wide association studies or the role of epigenetic factors), and the availability of large, established registries worldwide, the twin model represents a flexible study design for systems-biology studies of drug response in humans. In this review, we describe the ‘classical twin model’ and its application in traditional pharmacogenetics studies, discuss the value of the twin design in the modern systems biology era, and highlight the potential of existing twin registries in formulating future strategies in pharmacogenomics research. We argue that the usefulness of this design goes beyond its traditional applications. Moreover, the flexibility of the model in concert with the amenability of large, established registries of twins worldwide to the collecting of new phenotypes will mean that the study of identical and nonidentical twins will play a considerable role in shaping our understanding of the important factors that underpin population variability in common, complex phenotypes, including response to medication.

Toward the dynamic interactome: it's about time

Dynamic molecular interactions play a central role in regulating the functioning of cells and organisms. The availability of experimentally determined large-scale cellular networks, along with other high-throughput experimental data sets that provide snapshots of biological systems at different times and conditions, is increasingly helpful in elucidating interaction dynamics. Here we review the beginnings of a new subfield within computational biology, one focused on the global inference and analysis of the dynamic interactome. This burgeoning research area, which entails a shift from static to dynamic network analysis, promises to be a major step forward in our ability to model and reason about cellular function and behavior.

Gemcitabine and arabinosylcytosin pharmacogenomics: genome-wide association and drug response biomarkers

Cancer patients show large individual variation in their response to chemotherapeutic agents. Gemcitabine (dFdC) and AraC, two cytidine analogues, have shown significant activity against a variety of tumors. We previously used expression data from a lymphoblastoid cell line-based model system to identify genes that might be important for the two drug cytotoxicity. In the present study, we used that same model system to perform a genome-wide association (GWA) study to test the hypothesis that common genetic variation might influence both gene expression and response to the two drugs. Specifically, genome-wide single nucleotide polymorphisms (SNPs) and mRNA expression data were obtained using the Illumina 550K(R) HumanHap550 SNP Chip and Affymetrix U133 Plus 2.0 GeneChip, respectively, for 174 ethnically-defined "Human Variation Panel" lymphoblastoid cell lines. Gemcitabine and AraC cytotoxicity assays were performed to obtain IC(50) values for the cell lines. We then performed GWA studies with SNPs, gene expression and IC(50) of these two drugs. This approach identified SNPs that were associated with gemcitabine or AraC IC(50) values and with the expression regulation for 29 genes or 30 genes, respectively. One SNP in IQGAP2 (rs3797418) was significantly associated with variation in both the expression of multiple genes and gemcitabine and AraC IC(50). A second SNP in TGM3 (rs6082527) was also significantly associated with multiple gene expression and gemcitabine IC50. To confirm the association results, we performed siRNA knock down of selected genes with expression that was associated with rs3797418 and rs6082527 in tumor cell and the knock down altered gemcitabine or AraC sensitivity, confirming our association study results. These results suggest that the application of GWA approaches using cell-based model systems, when combined with complementary functional validation, can provide insights into mechanisms responsible for variation in cytidine analogue response.

Compendium of gene expression profiles comprising a baseline model of the human liver drug metabolism transcriptome

Oligonucleotide microarrays were used to study the variability of pharmacokinetics and drug metabolism (PKDM)-related gene expression in 75 normal human livers. The objective was to define and use absorption, distribution, metabolism and excretion (ADME) gene expression variability to discern co-regulated genes and potential surrogate biomarkers of inducible gene expression. RNA was prepared from donor tissue and hybridized on Agilent microarrays against an RNA mass balanced pool from all donors. Clustering of PKDM gene sets revealed donors with distinct patterns of gene expression that grouped genes known to be regulated by the nuclear receptor, pregnane X-receptor (PXR). Fold range metrics and frequency distributions from the heterogeneous human population were used to define the variability of individual PKDM genes in the 75 human livers and were placed in context by comparing expression data with basal ADME gene expression variability in an inbred and diet/environment controlled population of 27 Rhesus livers. The most variable genes in the hepatic transcriptome were mainly related to drug metabolism, intermediary metabolism, inflammation and cell cycle control. Unique patterns of expression across 75 individuals of inducible ADME gene expression allowed their expression to be correlated with the expression of many other genes. Correlated genes for AhR, CAR and PXR responsive genes (CYP1A2, CYP2B6 and CYP3A4) were identified that may be co-regulated and, therefore, provide clues to the identity of surrogate gene or protein markers for CYP induction. In conclusion, microarrays were used to define the variable expression of hepatic ADME genes in a diverse human population, the expression variability of ADME genes was compared with the expression variability in an inbred population of Rhesus monkeys, and genes were defined that may be co-regulated with important inducible CYP genes.

Genetic control of human brain transcript expression in Alzheimer disease

We recently surveyed the relationship between the human brain transcriptome and genome in a series of neuropathologically normal postmortem samples. We have now analyzed additional samples with a confirmed pathologic diagnosis of late-onset Alzheimer disease (LOAD; final n = 188 controls, 176 cases). Nine percent of the cortical transcripts that we analyzed had expression profiles correlated with their genotypes in the combined cohort, and approximately 5% of transcripts had SNP-transcript relationships that could distinguish LOAD samples. Two of these transcripts have been previously implicated in LOAD candidate-gene SNP-expression screens. This study shows how the relationship between common inherited genetic variants and brain transcript expression can be used in the study of human brain disorders. We suggest that studying the transcriptome as a quantitative endo-phenotype has greater power for discovering risk SNPs influencing expression than the use of discrete diagnostic categories such as presence or absence of disease.

Natural genetic variation in transcriptome reflects network structure inferred with major effect mutations: insulin/TOR and associated phenotypes in Drosophila melanogaster

Background

A molecular process based genotype-to-phenotype map will ultimately enable us to predict how genetic variation among individuals results in phenotypic alterations. Building such a map is, however, far from straightforward. It requires understanding how molecular variation re-shapes developmental and metabolic networks, and how the functional state of these networks modifies phenotypes in genotype specific way. We focus on the latter problem by describing genetic variation in transcript levels of genes in the InR/TOR pathway among 72 Drosophila melanogaster genotypes.

Results

We observe tight co-variance in transcript levels of genes not known to influence each other through direct transcriptional control. We summarize transcriptome variation with factor analyses, and observe strong co-variance of gene expression within the dFOXO-branch and within the TOR-branch of the pathway. Finally, we investigate whether major axes of transcriptome variation shape phenotypes expected to be influenced through the InR/TOR pathway. We find limited evidence that transcript levels of individual upstream genes in the InR/TOR pathway predict fly phenotypes in expected ways. However, there is no evidence that these effects are mediated through the major axes of downstream transcriptome variation.

Conclusion

In summary, our results question the assertion of the 'sparse' nature of genetic networks, while validating and extending candidate gene approaches in the analyses of complex traits.

Systems biology and its potential role in radiobiology

About a century ago, Conrad Röentgen discovered X-rays, and Henri Becquerel discovered a new phenomenon, which Marie and Pierre Curie later coined as radio-activity. Since their seminal work, we have learned much about the physical properties of radiation and its effects on living matter. Alas, the more we discover, the more we appreciate the complexity of the biological processes that are triggered by radiation exposure and eventually lead (or do not lead) to disease. Equipped with modern biological methods of high-throughput experimentation, imaging, and vastly increased computational prowess, we are now entering an era where we can piece some of the multifold aspects of radiation exposure and its sequelae together, and develop a more systemic understanding of radiogenic effects such as radio-carcinogenesis than has been possible in the past. It is evident from the complexity of even the known processes that such an understanding can only be gained if it is supported by mathematical models. At this point, the construction of comprehensive models is hampered both by technical inadequacies and a paucity of appropriate data. Nonetheless, some initial steps have been taken already and the generally increased interest in systems biology may be expected to speed up future progress. In this context, we discuss in this article examples of relatively small, yet very useful models that elucidate selected aspects of the effects of exposure to ionizing radiation and may shine a light on the path before us.

A systems biology approach to drug discovery

Common human diseases like obesity and diabetes are driven by complex networks of genes and any number of environmental factors. To understand this complexity in hopes of identifying targets and developing drugs against disease, a systematic approach is required to elucidate the genetic and environmental factors and interactions among and between these factors, and to establish how these factors induce changes in gene networks that in turn lead to disease. The explosion of large-scale, high-throughput technologies in the biological sciences has enabled researchers to take a more systems biology approach to study complex traits like disease. Genotyping of hundreds of thousands of DNA markers and profiling tens of thousands of molecular phenotypes simultaneously in thousands of individuals is now possible, and this scale of data is making it possible for the first time to reconstruct whole gene networks associated with disease. In the following sections, we review different approaches for integrating genetic expression and clinical data to infer causal relationships among gene expression traits and between expression and disease traits. We further review methods to integrate these data in a more comprehensive manner to identify common pathways shared by the causal factors driving disease, including the reconstruction of association and probabilistic causal networks. Particular attention is paid to integrating diverse information to refine these types of networks so that they are more predictive. To highlight these different approaches in practice, we step through an example on how Insig2 was identified as a causal factor for plasma cholesterol levels in mice.

Achieving confidence in mechanism for drug discovery and development

Decisions in drug development are made on the basis of determinations of cause and effect from experimental observations that span drug development phases. Despite advances in our powers of observation, the ability to determine compound mechanisms from large-scale multi-omic technologies continues to be a major bottleneck. This can only be overcome by utilizing computational learning methods that identify from compound data the circuits and connections between drug-affected molecular constituents and physiological observables. The marriage of multi-omics technologies with network inference approaches will provide missing insights needed to improve drug development success rates.

Genetic susceptibility for breast cancer: How many more genes to be found?

Today, breast cancer is the most commonly occurring cancer among women. It accounts for 22% of all female cancers and the estimated annual incidence of breast cancer worldwide is about one million cases. Many risk factors have been identified but a positive family history remains among the most important ones established for breast cancer, with first-degree relatives of patients having an approximately two-fold elevated risk. It is currently estimated that approximately 20-25% of this risk is explained by known breast cancer susceptibility genes, mostly those conferring high risks, such as BRCA1 and BRCA2. However, these genes explain less than 5% of the total breast cancer incidence, even though several studies have suggested that the proportion of breast cancer that can be attributed to a genetic factor may be as high as 30%. It is thus likely that there are still breast cancer susceptibility genes to be found. It is presently not known how many such genes there still are, nor how many will fall into the class of rare high-risk (e.g. BRCAx) or of common low-risk susceptibility genes, nor if and how these factors interact with each other to cause susceptibility (a polygenic model). In this review we will address this question and discuss the different undertaken approaches used in identifying new breast cancer susceptibility genes, such as (genome-wide) linkage analysis, CGH, LOH, association studies and global gene expression analysis.

Moving toward a system genetics view of disease

Testing hundreds of thousands of DNA markers in human, mouse, and other species for association to complex traits like disease is now a reality. However, information on how variations in DNA impact complex physiologic processes flows through transcriptional and other molecular networks. In other words, DNA variations impact complex diseases through the perturbations they cause to transcriptional and other biological networks, and these molecular phenotypes are intermediate to clinically defined disease. Because it is also now possible to monitor transcript levels in a comprehensive fashion, integrating DNA variation, transcription, and phenotypic data has the potential to enhance identification of the associations between DNA variation and diseases like obesity and diabetes, as well as characterize those parts of the molecular networks that drive these diseases. Toward that end, we review methods for integrating expression quantitative trait loci (eQTLs), gene expression, and clinical data to infer causal relationships among gene expression traits and between expression and clinical traits. We further describe methods to integrate these data in a more comprehensive manner by constructing coexpression gene networks that leverage pairwise gene interaction data to represent more general relationships. To infer gene networks that capture causal information, we describe a Bayesian algorithm that further integrates eQTLs, expression, and clinical phenotype data to reconstruct whole-gene networks capable of representing causal relationships among genes and traits in the network. These emerging network approaches, aimed at processing high-dimensional biological data by integrating data from multiple sources, represent some of the first steps in statistical genetics to identify multiple genetic perturbations that alter the states of molecular networks and that in turn push systems into disease states. Evolving statistical procedures that operate on networks will be critical to extracting information related to complex phenotypes like disease, as research goes beyond a single-gene focus. The early successes achieved with the methods described herein suggest that these more integrative genomics approaches to dissecting disease traits will significantly enhance the identification of key drivers of disease beyond what could be achieved by genetic association studies alone.

An integrative genomics strategy for systematic characterization of genetic loci modulating phenotypes

Naturally occurring genetic variations may affect certain phenotypes through influencing transcript levels of the genes that are causally related to those phenotypes. Genomic regions harboring common sequence variants that modulate gene expression can be mapped as quantitative trait loci (QTLs) using a newly developed genetical genomics approach. This enables a new strategy for systematically mapping novel genetic loci underlying various phenotypes. In this work, we started from a seed set of genes with variants that are known to affect behavioral and neurological phenotypes (as recorded in Mammalian Phenotype Ontology Database) and used microarrays to analyze their expression levels in brain samples of a panel of BXD recombinant inbred mouse strains. We then systematically mapped the QTLs controlling the expression of these genes. Candidate causal genes in the QTL intervals were evaluated for evidence of functional genetic polymorphisms. Using this method, we were able to predict novel genetic loci and causal genes for a number of behavioral and neurological phenotypes. Lines of independent evidence supporting some of our results were provided by transcription factor binding site analysis and by biomedical literature. This strategy integrates gene-phenotype relations from decades of experimental mutagenesis studies and new genomic resources to provide an approach to rapidly expand knowledge on genetic loci modulating phenotypes.

Genetics of global gene expression

A new field of genetic analysis of global gene expression has emerged in recent years, driven by the realization that traditional techniques of linkage and association analysis can be applied to thousands of transcript levels measured by microarrays. Genetic dissection of transcript abundance has shed light on the architecture of quantitative traits, provided a new approach for connecting DNA sequence variation with phenotypic variation, and improved our understanding of transcriptional regulation and regulatory variation.

Thematic review series: Systems Biology Approaches to Metabolic and Cardiovascular Disorders. Reverse engineering gene networks to identify key drivers of complex disease phenotypes

Diseases such as obesity, diabetes, and atherosclerosis result from multiple genetic and environmental factors, and importantly, interactions between genetic and environmental factors. Identifying susceptibility genes for these diseases using genetic and genomic technologies is accelerating, and the expectation over the next several years is that a number of genes will be identified for common diseases. However, the identification of single genes for disease has limited utility, given that diseases do not originate in complex systems from single gene changes. Further, the identification of single genes for disease may not lead directly to genes that can be targeted for therapeutic intervention. Therefore, uncovering single genes for disease in isolation of the broader network of molecular interactions in which they operate will generally limit the overall utility of such discoveries. Several integrative approaches have been developed and applied to reconstructing networks. Here we review several of these approaches that involve integrating genetic, expression, and clinical data to elucidate networks underlying disease. Networks reconstructed from these data provide a richer context in which to interpret associations between genes and disease. Therefore, these networks can lead to defining pathways underlying disease more objectively and to identifying biomarkers and more-robust points for therapeutic intervention.

PolymiRTS Database: linking polymorphisms in microRNA target sites with complex traits

Polymorphism in microRNA Target Site (PolymiRTS) database is a collection of naturally occurring DNA variations in putative microRNA target sites. PolymiRTSs may affect gene expression and cause variations in complex phenotypes. The database integrates sequence polymorphism, phenotype and expression microarray data, and characterizes PolymiRTSs as potential candidates responsible for the quantitative trait locus (QTL) effects. It is a resource for studying PolymiRTSs and their implications in phenotypic variations. PolymiRTS database can be accessed at http://compbio.utmem.edu/miRSNP/.

An integrated in silico gene mapping strategy in inbred mice

In recent years in silico analysis of common laboratory mice has been introduced and subsequently applied, in slightly different ways, as a methodology for gene mapping. Previously we have demonstrated some limitation of the methodology due to sporadic genetic correlations across the genome. Here, we revisit the three main aspects that affect in silico analysis. First, we report on the use of marker maps: we compared our existing 20,000 SNP map to the newly released 140,000 SNP map. Second, we investigated the effect of varying strain numbers on power to map QTL. Third, we introduced a novel statistical approach: a cladistic analysis, which is well suited for mouse genetics and has increased flexibility over existing in silico approaches. We have found that in our examples of complex traits, in silico analysis by itself does fail to uniquely identify quantitative trait gene (QTG)-containing regions. However, when combined with additional information, it may significantly help to prioritize candidate genes. We therefore recommend using an integrated work flow that uses other genomic information such as linkage regions, regions of shared ancestry, and gene expression information to obtain a list of candidate genes from the genome.

Genetics of osteoporosis

Osteoporosis is a common disease with a strong genetic component. In recent years, some progress has been made in understanding the genetic basis of osteoporosis. Genetic factors contribute to osteoporosis by influencing not only bone mineral density but also bone size, bone quality, and bone turnover. Meta-analysis has been used to define the role of several candidate genes in osteoporosis. Some quantitative trait loci that regulate bone mass identified by linkage studies in humans and experimental animals have been replicated in multiple populations. Genes that cause monogenic bone diseases also contribute to regulation of bone mass in the normal population. Genome-wide association studies and functional genomics approaches have recently begun to apply to genetic studies of osteoporosis. In the future, not only single gene but also the entire gene networks involved in osteoporosis and regulation of bone mass will systematically be discovered through integrative genomics.