Mendelian randomization in health research: Using appropriate genetic variants and avoiding biased estimates

Mendelian randomization methods, which use genetic variants as instrumental variables for exposures of interest to overcome problems of confounding and reverse causality, are becoming widespread for assessing causal relationships in epidemiological studies. The main purpose of this paper is to demonstrate how results can be biased if researchers select genetic variants on the basis of their association with the exposure in their own dataset, as often happens in candidate gene analyses. This can lead to estimates that indicate apparent “causal” relationships, despite there being no true effect of the exposure. In addition, we discuss the potential bias in estimates of magnitudes of effect from Mendelian randomization analyses when the measured exposure is a poor proxy for the true underlying exposure. We illustrate these points with specific reference to tobacco research.     

Integrating epidemiology and genetic association: the challenge of gene-environment interaction

Recent advances in human genomics have made it possible to better understand the genetic basis of disease. In addition, genetic association studies can also elucidate the mechanisms by which "non-genetic" exogenous and endogenous exposures influence the risk of disease. This is true both of studies that assess the marginal effect of a single gene and studies that look at the joint effect of genes and environmental exposures. For example, gene variants that are known to alter enzyme function or level can serve as surrogates for long-term biomarker levels that are impractical or impossible to measure on many subjects. Evidence that genetic variants modify the effect of an established risk factor may help specify the risk factor's biologically active components. We illustrate these ideas with several examples and discuss design and analysis challenges, particularly for studies of gene-environment interaction. We argue that to increase the power to detect interaction effects and limit the number of false positive results, large sample sizes will be needed, which are currently only available through planned collaborative efforts. Such collaborations also ensure a common approach to measuring variation at a genetic locus, avoiding a problem that has led to difficulties when comparing results from genetic association studies.     

The complex genetic and molecular basis of a model quantitative trait

Quantitative traits are often influenced by many loci with small effects. Identifying most of these loci and resolving them to specific genes or genetic variants is challenging. Yet, achieving such a detailed understanding of quantitative traits is important, as it can improve our knowledge of the genetic and molecular basis of heritable phenotypic variation. In this study, we use a genetic mapping strategy that involves recurrent backcrossing with phenotypic selection to obtain new insights into an ecologically, industrially, and medically relevant quantitative trait—tolerance of oxidative stress, as measured based on resistance to hydrogen peroxide. We examine the genetic basis of hydrogen peroxide resistance in three related yeast crosses and detect 64 distinct genomic loci that likely influence the trait. By precisely resolving or cloning a number of these loci, we demonstrate that a broad spectrum of cellular processes contribute to hydrogen peroxide resistance, including DNA repair, scavenging of reactive oxygen species, stress-induced MAPK signaling, translation, and water transport. Consistent with the complex genetic and molecular basis of hydrogen peroxide resistance, we show two examples where multiple distinct causal genetic variants underlie what appears to be a single locus. Our results improve understanding of the genetic and molecular basis of a highly complex, model quantitative trait.     

Genetics and cardiovascular system: influence of human genetic variants on vascular function

Candidate gene association studies in cardiovascular diseases have provided evidence on the molecular basis of phenotypic differences between individuals. The comprehension of how inherited genetic variants are able to affect protein functions has increased the knowledge of how genes interact with environment in order to modulate a particular phenotype. Although it is known that the human genome contains more than 10 million SNPs, only a minor part of them are supposed to be functional. A causative SNP in a particular gene may confer a small to moderate effect in complex phenotypes, such as functions important to cardiovascular homeostasis. This paper is a selective review of the literature on the evidence for interactions between vascular function and naturally occurring genetic variants in endothelial nitric oxide synthase (eNOS) and beta-2 adrenergic receptor (ADRB2), two genes among those influencing vascular phenotype and examples for which there is a strong evidence base. eNOS and ADRB2 will be characterized, as well as the mechanisms by which the enzyme and the receptor work to control vascular responses will be described. Understanding the molecular mechanisms underlying gene-mediated vascular function and their modification by genetic variants is expected to result in a better comprehension about individual's phenotypic differences.     

Adjusting for Heritable Covariates Can Bias Effect Estimates in Genome-Wide Association Studies

In recent years, a number of large-scale genome-wide association studies have been published for human traits adjusted for other correlated traits with a genetic basis. In most studies, the motivation for such an adjustment is to discover genetic variants associated with the primary outcome independently of the correlated trait. In this report, we contend that this objective is fulfilled when the tested variants have no effect on the covariate or when the correlation between the covariate and the outcome is fully explained by a direct effect of the covariate on the outcome. For all other scenarios, an unintended bias is introduced with respect to the primary outcome as a result of the adjustment, and this bias might lead to false positives. Here, we illustrate this point by providing examples from published genome-wide association studies, including large meta-analysis of waist-to-hip ratio and waist circumference adjusted for body mass index (BMI), where genetic effects might be biased as a result of adjustment for body mass index. Using both theory and simulations, we explore this phenomenon in detail and discuss the ramifications for future genome-wide association studies of correlated traits and diseases.     

G = E: What GWAS Can Tell Us about the Environment

As our understanding of genetics has improved, genome-wide association studies (GWAS) have identified numerous variants associated with lifestyle behaviours and health outcomes. However, what is sometimes overlooked is the possibility that genetic variants identified in GWAS of disease might reflect the effect of modifiable risk factors as well as direct genetic effects. We discuss this possibility with illustrative examples from tobacco and alcohol research, in which genetic variants that predict behavioural phenotypes have been seen in GWAS of diseases known to be causally related to these behaviours. This consideration has implications for the interpretation of GWAS findings.     

Where are the missing pieces of the schizophrenia genetics puzzle?

On the basis of recent data from candidate region/gene and genome-wide association studies (GWAS) and their follow-up investigations, the number of genes potentially implicated in schizophrenia has been estimated to be over 1000. However, with regard to the identified odds ratio, it is likely that genetic variants with more definitive effect on schizophrenia phenotype are still missing. The hunt therefore remains open for the genetic variants that would explain the majority of the missing heritability of schizophrenia. This review aims at summarizing data from recent DNA microarray and target gene/region resequencing in order to propose new insights of where to look next. The review is divided into three sections: GWAS, copy-number variations and rare variant--candidate gene resequencing.     

The evolution of novel host use is unlikely to be constrained by trade‐offs or a lack of genetic variation

The genetic and ecological factors that shape the evolution of animal diets remain poorly understood. For herbivorous insects, the expectation has been that trade‐offs exist, such that adaptation to one host plant reduces performance on other potential hosts. We investigated the genetic architecture of alternative host use by rearing individual Lycaeides melissa butterflies from two wild populations in a crossed design on two hosts (one native and one introduced) and analysing the genetic basis of differences in performance using genomic approaches. Survival during the experiment was highest when butterfly larvae were reared on their natal host plant, consistent with local adaptation. However, cross‐host correlations in performance among families (within populations) were not different from zero. We found that L. melissa populations possess genetic variation for larval performance and variation in performance had a polygenic basis. We documented very few genetic variants with trade‐offs that would inherently constrain diet breadth by preventing the optimization of performance across hosts. Instead, most genetic variants that affected performance on one host had little to no effect on the other host. In total, these results suggest that genetic trade‐offs are not the primary cause of dietary specialization in L. melissa butterflies.     

Identification and functional characterization of new missense SNPs in the coding region of the TP53 gene

Infrequent and rare genetic variants in the human population vastly outnumber common ones. Although they may contribute significantly to the genetic basis of a disease, these seldom-encountered variants may also be miss-identified as pathogenic if no correct references are available. Somatic and germline TP53 variants are associated with multiple neoplastic diseases, and thus have come to serve as a paradigm for genetic analyses in this setting. We searched 14 independent, globally distributed datasets and recovered TP53 SNPs from 202,767 cancer-free individuals. In our analyses, 19 new missense TP53 SNPs, including five novel variants specific to the Asian population, were recurrently identified in multiple datasets. Using a combination of in silico, functional, structural, and genetic approaches, we showed that none of these variants displayed loss of function compared to the normal TP53 gene. In addition, classification using ACMG criteria suggested that they are all benign. Considered together, our data reveal that the TP53 coding region shows far more polymorphism than previously thought and present high ethnic diversity. They furthermore underline the importance of correctly assessing novel variants in all variant-calling pipelines associated with genetic diagnoses for cancer.Subject terms: Tumour-suppressor proteins, Genetics research     

MAINTENANCE OF GENETIC VARIATION IN HUMAN PERSONALITY: TESTING EVOLUTIONARY MODELS BY ESTIMATING HERITABILITY DUE TO COMMON CAUSAL VARIANTS AND INVESTIGATING THE EFFECT OF DISTANT INBREEDING

Personality traits are basic dimensions of behavioral variation, and twin, family, and adoption studies show that around 30% of the between‐individual variation is due to genetic variation. There is rapidly growing interest in understanding the evolutionary basis of this genetic variation. Several evolutionary mechanisms could explain how genetic variation is maintained in traits, and each of these makes predictions in terms of the relative contribution of rare and common genetic variants to personality variation, the magnitude of nonadditive genetic influences, and whether personality is affected by inbreeding. Using genome‐wide single nucleotide polymorphism (SNP) data from > 8000 individuals, we estimated that little variation in the Cloninger personality dimensions (7.2% on average) is due to the combined effect of common, additive genetic variants across the genome, suggesting that most heritable variation in personality is due to rare variant effects and/or a combination of dominance and epistasis. Furthermore, higher levels of inbreeding were associated with less socially desirable personality trait levels in three of the four personality dimensions. These findings are consistent with genetic variation in personality traits having been maintained by mutation–selection balance.     

Association analysis of candidate genes for neuropsychiatric disease: the perpetual campaign

Association studies have been proposed to identify the genetic determinants of complex neuropsychiatric traits. Although such studies of candidate genes offer great potential to identify genetic variants that contribute to the expression of psychiatric disease, no consistent associations have been identified. Studies to date have focused on candidate genes that are selected for analysis on the basis of incomplete information about gene function in the brain, therefore the majority of genes expressed in the brain have been ignored. Additionally, most genetic determinants of psychiatric disease will probably be of modest effect and therefore require association studies of large samples. As genomic technologies advance, massive genotyping of large samples should allow identification of alleles that contribute to psychopathology.     

The sense and nonsense of direct-to-consumer genetic testing for cardiovascular disease

Expectations are high that increasing knowledge of the genetic basis of cardiovascular disease will eventually lead to personalised medicine—to preventive and therapeutic interventions that are targeted to at-risk individuals on the basis of their genetic profiles. Most cardiovascular diseases are caused by a complex interplay of many genetic variants interacting with many non-genetic risk factors such as diet, exercise, smoking and alcohol consumption. Since several years, genetic susceptibility testing for cardiovascular diseases is being offered via the internet directly to consumers. We discuss five reasons why these tests are not useful, namely: (1) the predictive ability is still limited; (2) the risk models used by the companies are based on assumptions that have not been verified; (3) the predicted risks keep changing when new variants are discovered and added to the test; (4) the tests do not consider non-genetic factors in the prediction of cardiovascular disease risk; and (5) the test results will not change recommendations of preventive interventions. Predictive genetic testing for multifactorial forms of cardiovascular disease clearly lacks benefits for the public. Prevention of disease should therefore remain focused on family history and on non-genetic risk factors as diet and physical activity that can have the strongest impact on disease risk, regardless of genetic susceptibility.     

Bayesian variable selection in searching for additive and dominant effects in genome-wide data

Although complex diseases and traits are thought to have multifactorial genetic basis, the common methods in genome-wide association analyses test each variant for association independent of the others. This computational simplification may lead to reduced power to identify variants with small effect sizes and requires correcting for multiple hypothesis tests with complex relationships. However, advances in computational methods and increase in computational resources are enabling the computation of models that adhere more closely to the theory of multifactorial inheritance. Here, a Bayesian variable selection and model averaging approach is formulated for searching for additive and dominant genetic effects. The approach considers simultaneously all available variants for inclusion as predictors in a linear genotype-phenotype mapping and averages over the uncertainty in the variable selection. This leads to naturally interpretable summary quantities on the significances of the variants and their contribution to the genetic basis of the studied trait. We first characterize the behavior of the approach in simulations. The results indicate a gain in the causal variant identification performance when additive and dominant variation are simulated, with a negligible loss of power in purely additive case. An application to the analysis of high- and low-density lipoprotein cholesterol levels in a dataset of 3895 Finns is then presented, demonstrating the feasibility of the approach at the current scale of single-nucleotide polymorphism data. We describe a Markov chain Monte Carlo algorithm for the computation and give suggestions on the specification of prior parameters using commonly available prior information. An open-source software implementing the method is available at http://www.lce.hut.fi/research/mm/bmagwa/ and https://github.com/to-mi/.     

Delineating the Hemostaseome as an aid to individualize the analysis of the hereditary basis of thrombotic and bleeding disorders

Next-generation sequencing and genome-wide association studies represent powerful tools to identify genetic variants that confer disease risk within populations. On their own, however, they cannot provide insight into how these variants contribute to individual risk for diseases that exhibit complex inheritance, or alternatively confer health in a given individual. Even in the case of well-characterized variants that confer a significant disease risk, more healthy individuals carry the variant, with no apparent ill effect, than those who manifest disease. Access to low-cost genome sequence data promises to provide an unprecedentedly detailed view of the nature of the hereditary component of complex diseases, but requires the large-scale comparison of sequence data from individuals with and without disease to deliver a clinical calibration. The provision of informatics support remains problematic as there are currently no means to interpret the data generated. Here, we initiate this process, a prerequisite for such a study, by narrowing the focus from an entire genome to that of a single biological system. To this end, we examine the ‘Hemostaseome,’ and more specifically focus on DNA sequence changes pertaining to those human genes known to impact upon hemostasis and thrombosis that can be analyzed coordinately, and on an individual basis, to interrogate how specific combinations of variants act to confer disease predisposition. As a first step, we delineate known members of the Hemostaseome and explore the nature of the genetic variants that may cause disease in individuals whose hemostatic balance has become shifted toward either a prothrombotic or anticoagulant phenotype.     

Polygenic evolution drives species divergence and climate adaptation in corals

Closely related species often show substantial differences in ecological traits that allow them to occupy different environmental niches. For few of these systems is it clear what the genomic basis of adaptation is and whether a few loci of major effect or many genome‐wide differences drive species divergence. Four cryptic species of the tabletop coral Acropora hyacinthus are broadly sympatric in American Samoa; here we show that two common species have differences in key environmental traits such as microhabitat distributions and thermal stress tolerance. We compared gene expression patterns and genetic polymorphism between these two species using RNA‐Seq. The vast majority of polymorphisms are shared between species, but the two species show widespread differences in allele frequencies and gene expression, and tend to host different symbiont types. We find that changes in gene expression are related to changes in the frequencies of many gene regulatory variants, but that many of these differences are consistent with the action of genetic drift. However, we observe greater genetic divergence between species in amino acid replacement polymorphisms compared to synonymous variants. These findings suggest that polygenic evolution plays a major role in driving species differences in ecology and resilience to climate change.     

Supporting the structural basis of prion strains: induction and identification of [PSI] variants

The [PSI] genetic element, which enhances the nonsense suppression efficiency in the yeast Saccharomyces cerevisiae, is thought to be amyloid-like aggregates of the Sup35 protein, and to self-propagate by a prion-like mechanism. Analogous to strains of the mammalian prion, variants of [PSI], with different nonsense suppression efficiencies and mitotic stabilities, can be isolated from the same yeast genetic background. In the framework of the "protein-only" hypothesis, variants of prion are assumed to be distinct conformers of the same prion polypeptide. This study aims to provide further support for the structural basis of [PSI] variation. Three variants of [PSI] were induced and distinguished by a panel of 11 single point mutations of the Sup35 protein. The variant phenotypes are intrinsically associated with [PSI] elements, presumably structurally different amyloids, rather than produced from variations in the genetic background. Differential incorporation to [PSI] variants of a Sup35 point mutation as well as N and C-terminally truncated Sup35 fragments is further demonstrated in vivo, suggesting that distinct patches of amino acid residues are involved in the assembly of [PSI] variants. These results establish a method for [PSI] variant-typing and indicate that heritable variations of amyloid structures can be derived from the same polypeptide.     

CNTNAP2 variants affect early language development in the general population

Early language development is known to be under genetic influence, but the genes affecting normal variation in the general population remain largely elusive. Recent studies of disorder reported that variants of the CNTNAP2 gene are associated both with language deficits in specific language impairment (SLI) and with language delays in autism. We tested the hypothesis that these CNTNAP2 variants affect communicative behavior, measured at 2 years of age in a large epidemiological sample, the Western Australian Pregnancy Cohort (Raine) Study. Singlepoint analyses of 1149 children (606 males and 543 females) revealed patterns of association which were strikingly reminiscent of those observed in previous investigations of impaired language, centered on the same genetic markers and with a consistent direction of effect (rs2710102, P = 0.0239; rs759178, P = 0.0248). On the basis of these findings, we performed analyses of four-marker haplotypes of rs2710102-rs759178-rs17236239-rs2538976 and identified significant association (haplotype TTAA, P = 0.049; haplotype CGAG, [corrected] P = .0014). Our study suggests that common variants in the exon 13-15 region of CNTNAP2 influence early language acquisition, as assessed at age 2, in the general population. We propose that these CNTNAP2 variants increase susceptibility to SLI or autism when they occur together with other risk factors.