Fast and accurate construction of confidence intervals for heritability

Abstract

Estimation of heritability is fundamental in genetic studies. In recent years, heritability estimation using linear mixed models (LMMs) has gained popularity, because these estimates can be obtained from unrelated individuals collected in genome wide association studies. Typically, heritability estimation under LMMs uses either the maximum likelihood (ML) or the restricted maximum likelihood (REML) approach.Existing methods for the construction of confidence intervals and estimators of standard errors for both ML and REML rely on asymptotic properties. However, these assumptions are often violated due to the bounded parameter space, statistical dependencies, and limited sample size, leading to biased estimates, and inflated or deflated confidence intervals. Here, we show that often the probability that the genetic component is estimated as zero is high even when the true heritability is bounded away from zero, emphasizing the need for accurate confidence intervals. We further show that the estimation of confidence intervals by state-of-the-art methods is highly inaccurate, especially when the true heritability is either relatively low or relatively high. Such biases are present, for example, in estimates of heritability of gene expression in the GTEx study, and of lipid profiles in the LURIC study. We propose a computationally efficient method, Accurate LMM-B ased confidence I ntervals (ALBI), for the estimation of the distribution of the heritability estimator, and for the construction of accurate confidence intervals. Our method can be used as an add-on to existing methods for heritability and variance components estimation, such as GCTA, FaST-LMM, GEMMA, or EMMA. ALBI is available at

The Genetic Architecture of Gene Expression in Peripheral Blood

“…Furthermore, the heritability estimates from the constrained REML algorithm are potentially biased due to the bounded variance component parameter space, which is alleviated by the reporting of the estimates from the unconstrained REML algorithm. Schweiger et al 42 showed that the reported standard errors from the constrained REML algorithm led to the construction of confidence intervals with inaccurate coverage probabilities. However, the reported mean standard error from the constrained REML algorithm is a meaningful measure of the uncertainty in these estimates due to the law of large numbers.…”

Environmental factors dominate over host genetics in shaping human gut microbiota composition

“…We therefore first asked whether biome-explainability can be reliably estimated with our sample size. To this end, we computed confidence intervals (CIs) for a large number of biome-explainability values via parametric bootstrap, which constructs CIs by repeatedly drawing random realizations of the phenotypes according to the principle of test inversion 48 . We find that biome-explainability can be estimated more accurately than genetic heritability in our sample, with an average 95% biome-explainability CI width of 32.8% (averaged over different biome-explainability levels), compared with 98.7% for a genetic heritability CI in our cohort ( Supplementary Table 11).…”

Efficient Construction of Test Inversion Confidence Intervals Using Quantile Regression

“…Since the distribution of the estimator is unknown, U cannot be inferred analytically and has to be estimated by a Monte-Carlo simulation. The straight forward way to carry such a simulation is to interpolate sample quantile of Θ(X), as in Schweiger et al [2016]. This approach is briefly described in Algorithm 1, resulting inÛ -an estimate of U .…”