Research on how interactions between candidate genes and environmental factors influence psychiatric illnesses has generated enthusiasm but not many replicable findings. The authors discovered that ...the more closely a replication study matched the original research, the less likely it was to have similar results. Publication bias toward positive findings was apparent both in reports of novel findings and in replication studies. Another contributor to false discoveries in many studies of candidate genes is low statistical power due to small study groups or other design factors.
Objective:Gene-by-environment interaction (G×E) studies in psychiatry have typically been conducted using a candidate G×E (cG×E) approach, analogous to the candidate gene association approach used to test genetic main effects. Such cG×E research has received widespread attention and acclaim, yet cG×E findings remain controversial. The authors examined whether the many positive cG×E findings reported in the psychiatric literature were robust or if, in aggregate, cG×E findings were consistent with the existence of publication bias, low statistical power, and a high false discovery rate.
Method:The authors conducted analyses on data extracted from all published studies (103 studies) from the first decade (2000–2009) of cG×E research in psychiatry.
Results:Ninety-six percent of novel cG×E studies were significant compared with 27% of replication attempts. These findings are consistent with the existence of publication bias among novel cG×E studies, making cG×E hypotheses appear more robust than they actually are. There also appears to be publication bias among replication attempts because positive replication attempts had smaller average sample sizes than negative ones. Power calculations using observed sample sizes suggest that cG×E studies are underpowered. Low power along with the likely low prior probability of a given cG×E hypothesis being true suggests that most or even all positive cG×E findings represent type I errors.
Conclusions:In this new era of big data and small effects, a recalibration of views about groundbreaking findings is necessary. Well-powered direct replications deserve more attention than novel cG×E findings and indirect replications.
Inbreeding depression, which refers to reduced fitness among offspring of related parents, has traditionally been studied using pedigrees. In practice, pedigree information is difficult to obtain, ...potentially unreliable, and rarely assessed for inbreeding arising from common ancestors who lived more than a few generations ago. Recently, there has been excitement about using SNP data to estimate inbreeding (F) arising from distant common ancestors in apparently "outbred" populations. Statistical power to detect inbreeding depression using SNP data depends on the actual variation in inbreeding in a population, the accuracy of detecting that with marker data, the effect size, and the sample size. No one has yet investigated what variation in F is expected in SNP data as a function of population size, and it is unclear which estimate of F is optimal for detecting inbreeding depression. In the present study, we use theory, simulated genetic data, and real genetic data to find the optimal estimate of F, to quantify the likely variation in F in populations of various sizes, and to estimate the power to detect inbreeding depression. We find that F estimated from runs of homozygosity (Froh), which reflects shared ancestry of genetic haplotypes, retains variation in even large populations (e.g., SD=0.5% when Ne=10,000) and is likely to be the most powerful method of detecting inbreeding effects from among several alternative estimates of F. However, large samples (e.g., 12,000-65,000) will be required to detect inbreeding depression for likely effect sizes, and so studies using Froh to date have probably been underpowered.
Candidate gene × environment (G × E) interaction research tests the hypothesis that the effects of some environmental variable (e.g., childhood maltreatment) on some outcome measure (e.g., ...depression) depend on a particular genetic polymorphism. Because this research is inherently nonexperimental, investigators have been rightly concerned that detected interactions could be driven by confounders (e.g., ethnicity, gender, age, socioeconomic status) rather than by the specified genetic or environmental variables per se. In an attempt to eliminate such alternative explanations for detected G × E interactions, investigators routinely enter the potential confounders as covariates in general linear models. However, this practice does not control for the effects these variables might have on the G × E interaction. Rather, to properly control for confounders, researchers need to enter the covariate × environment and the covariate × gene interaction terms in the same model that tests the G × E term. In this manuscript, I demonstrate this point analytically and show that the practice of improperly controlling for covariates is the norm in the G × E interaction literature to date. Thus, many alternative explanations for G × E findings that investigators had thought were eliminated have not been.
Given that natural selection is so powerful at optimizing complex adaptations, why does it seem unable to eliminate genes (susceptibility alleles) that predispose to common, harmful, heritable mental ...disorders, such as schizophrenia or bipolar disorder? We assess three leading explanations for this apparent paradox from evolutionary genetic theory: (1) ancestral neutrality (susceptibility alleles were not harmful among ancestors), (2) balancing selection (susceptibility alleles sometimes increased fitness), and (3) polygenic mutation-selection balance (mental disorders reflect the inevitable mutational load on the thousands of genes underlying human behavior). The first two explanations are commonly assumed in psychiatric genetics and Darwinian psychiatry, while mutation-selection has often been discounted. All three models can explain persistent genetic variance in some traits under some conditions, but the first two have serious problems in explaining human mental disorders. Ancestral neutrality fails to explain low mental disorder frequencies and requires implausibly small selection coefficients against mental disorders given the data on the reproductive costs and impairment of mental disorders. Balancing selection (including spatio-temporal variation in selection, heterozygote advantage, antagonistic pleiotropy, and frequency-dependent selection) tends to favor environmentally contingent adaptations (which would show no heritability) or high-frequency alleles (which psychiatric genetics would have already found). Only polygenic mutation-selection balance seems consistent with the data on mental disorder prevalence rates, fitness costs, the likely rarity of susceptibility alleles, and the increased risks of mental disorders with brain trauma, inbreeding, and paternal age. This evolutionary genetic framework for mental disorders has wide-ranging implications for psychology, psychiatry, behavior genetics, molecular genetics, and evolutionary approaches to studying human behavior.
A central aim for studying runs of homozygosity (ROHs) in genome-wide SNP data is to detect the effects of autozygosity (stretches of the two homologous chromosomes within the same individual that ...are identical by descent) on phenotypes. However, it is unknown which current ROH detection program, and which set of parameters within a given program, is optimal for differentiating ROHs that are truly autozygous from ROHs that are homozygous at the marker level but vary at unmeasured variants between the markers.
We simulated 120 Mb of sequence data in order to know the true state of autozygosity. We then extracted common variants from this sequence to mimic the properties of SNP platforms and performed ROH analyses using three popular ROH detection programs, PLINK, GERMLINE, and BEAGLE. We varied detection thresholds for each program (e.g., prior probabilities, lengths of ROHs) to understand their effects on detecting known autozygosity.
Within the optimal thresholds for each program, PLINK outperformed GERMLINE and BEAGLE in detecting autozygosity from distant common ancestors. PLINK's sliding window algorithm worked best when using SNP data pruned for linkage disequilibrium (LD).
Our results provide both general and specific recommendations for maximizing autozygosity detection in genome-wide SNP data, and should apply equally well to research on whole-genome autozygosity burden or to research on whether specific autozygous regions are predictive using association mapping methods.
Inbreeding increases the risk of certain Mendelian disorders in humans but may also reduce fitness through its effects on complex traits and diseases. Such inbreeding depression is thought to occur ...due to increased homozygosity at causal variants that are recessive with respect to fitness. Until recently it has been difficult to amass large enough sample sizes to investigate the effects of inbreeding depression on complex traits using genome-wide single nucleotide polymorphism (SNP) data in population-based samples. Further, it is difficult to infer causation in analyses that relate degree of inbreeding to complex traits because confounding variables (e.g., education) may influence both the likelihood for parents to outbreed and offspring trait values. The present study used runs of homozygosity in genome-wide SNP data in up to 400,000 individuals in the UK Biobank to estimate the proportion of the autosome that exists in autozygous tracts-stretches of the genome which are identical due to a shared common ancestor. After multiple testing corrections and controlling for possible sociodemographic confounders, we found significant relationships in the predicted direction between estimated autozygosity and three of the 26 traits we investigated: age at first sexual intercourse, fluid intelligence, and forced expiratory volume in 1 second. Our findings corroborate those of several published studies. These results may imply that these traits have been associated with Darwinian fitness over evolutionary time. However, some of the autozygosity-trait relationships were attenuated after controlling for background sociodemographic characteristics, suggesting that alternative explanations for these associations have not been eliminated. Care needs to be taken in the design and interpretation of ROH studies in order to glean reliable information about the genetic architecture and evolutionary history of complex traits.
Schizophrenia is a complex disorder caused by both genetic and environmental factors. Using 9,087 affected individuals, 12,171 controls and 915,354 imputed SNPs from the Schizophrenia Psychiatric ...Genome-Wide Association Study (GWAS) Consortium (PGC-SCZ), we estimate that 23% (s.e. = 1%) of variation in liability to schizophrenia is captured by SNPs. We show that a substantial proportion of this variation must be the result of common causal variants, that the variance explained by each chromosome is linearly related to its length (r = 0.89, P = 2.6 × 10(-8)), that the genetic basis of schizophrenia is the same in males and females, and that a disproportionate proportion of variation is attributable to a set of 2,725 genes expressed in the central nervous system (CNS; P = 7.6 × 10(-8)). These results are consistent with a polygenic genetic architecture and imply more individual SNP associations will be detected for this disease as sample size increases.
Many traits are subject to assortative mating, with recent molecular genetic findings confirming longstanding theoretical predictions that assortative mating induces long range dependence across ...causal variants. However, all marker-based heritability estimators implicitly assume mating is random. We provide mathematical and simulation-based evidence demonstrating that both method-of-moments and likelihood-based estimators are biased in the presence of assortative mating and derive corrected heritability estimators for traits subject to assortment. Finally, we demonstrate that the empirical patterns of estimates across methods and sample sizes for real traits subject to assortative mating are congruent with expected assortative mating-induced biases. For example, marker-based heritability estimates for height are 14% - 23% higher than corrected estimates using UK Biobank data.
The causes of temporal fluctuations in adult traits are poorly understood. Here, we investigate the genetic determinants of within-person trait variability of 8 repeatedly measured anthropometric ...traits in 50,117 individuals from the UK Biobank. We found that within-person (non-directional) variability had a SNP-based heritability of 2-5% for height, sitting height, body mass index (BMI) and weight (P
2.4 × 10
). We also analysed longitudinal trait change and show a loss of both average height and weight beyond about 70 years of age. A variant tracking the Alzheimer's risk APOE-
allele (rs429358) was significantly associated with weight loss (
= -0.047 kg per yr, s.e. 0.007, P = 2.2 × 10
), and using 2-sample Mendelian Randomisation we detected a relationship consistent with causality between decreased lumbar spine bone mineral density and height loss (b
= 0.011, s.e. 0.003, P = 3.5 × 10
). Finally, population-level variance quantitative trait loci (vQTL) were consistent with within-person variability for several traits, indicating an overlap between trait variability assessed at the population or individual level. Our findings help elucidate the genetic influence on trait-change within an individual and highlight disease risks associated with these changes.
Attributing the similarity between individuals to genetic and non-genetic factors is central to genetic analyses. In this paper we use the genomic relationship (Formula: see text) among 417,060 ...individuals to investigate the phenotypic covariance between pairs of individuals for 32 traits across the spectrum of relatedness, from unrelated pairs through to identical twins. We find linear relationships between phenotypic covariance and Formula: see text that agree with the SNP-based heritability (Formula: see text) in unrelated pairs (Formula: see text), and with pedigree-estimated heritability in close relatives (Formula: see text). The covariance increases faster than Formula: see text in distant relatives (Formula: see text), and we attribute this to imperfect linkage disequilibrium between causal variants and the common variants used to construct Formula: see text. We also examine the effect of assortative mating on heritability estimates from different experimental designs. We find that full-sib identity-by-descent regression estimates for height (0.66 s.e. 0.07) are consistent with estimates from close relatives (0.82 s.e. 0.04) after accounting for the effect of assortative mating.