Dataset Information

Monte Carlo simulation of OLS and linear mixed model inference of phenotypic effects on gene expression.

ABSTRACT: BACKGROUND:Self-contained tests estimate and test the association between a phenotype and mean expression level in a gene set defined a priori. Many self-contained gene set analysis methods have been developed but the performance of these methods for phenotypes that are continuous rather than discrete and with multiple nuisance covariates has not been well studied. Here, I use Monte Carlo simulation to evaluate the performance of both novel and previously published (and readily available via R) methods for inferring effects of a continuous predictor on mean expression in the presence of nuisance covariates. The motivating data are a high-profile dataset which was used to show opposing effects of hedonic and eudaimonic well-being (or happiness) on the mean expression level of a set of genes that has been correlated with social adversity (the CTRA gene set). The original analysis of these data used a linear model (GLS) of fixed effects with correlated error to infer effects of Hedonia and Eudaimonia on mean CTRA expression. METHODS:The standardized effects of Hedonia and Eudaimonia on CTRA gene set expression estimated by GLS were compared to estimates using multivariate (OLS) linear models and generalized estimating equation (GEE) models. The OLS estimates were tested using O'Brien's OLS test, Anderson's permutation [Formula: see text]-test, two permutation F-tests (including GlobalAncova), and a rotation z-test (Roast). The GEE estimates were tested using a Wald test with robust standard errors. The performance (Type I, II, S, and M errors) of all tests was investigated using a Monte Carlo simulation of data explicitly modeled on the re-analyzed dataset. RESULTS:GLS estimates are inconsistent between data sets, and, in each dataset, at least one coefficient is large and highly statistically significant. By contrast, effects estimated by OLS or GEE are very small, especially relative to the standard errors. Bootstrap and permutation GLS distributions suggest that the GLS results in downward biased standard errors and inflated coefficients. The Monte Carlo simulation of error rates shows highly inflated Type I error from the GLS test and slightly inflated Type I error from the GEE test. By contrast, Type I error for all OLS tests are at the nominal level. The permutation F-tests have ?1.9X the power of the other OLS tests. This increased power comes at a cost of high sign error (?10%) if tested on small effects. DISCUSSION:The apparently replicated pattern of well-being effects on gene expression is most parsimoniously explained as "correlated noise" due to the geometry of multiple regression. The GLS for fixed effects with correlated error, or any linear mixed model for estimating fixed effects in designs with many repeated measures or outcomes, should be used cautiously because of the inflated Type I and M error. By contrast, all OLS tests perform well, and the permutation F-tests have superior performance, including moderate power for very small effects.

SUBMITTER: Walker JA

PROVIDER: S-EPMC5068350 | biostudies-literature | 2016

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Monte Carlo simulation of OLS and linear mixed model inference of phenotypic effects on gene expression.

Walker Jeffrey A JA

PeerJ 20161011

<h4>Background</h4>Self-contained tests estimate and test the association between a phenotype and mean expression level in a gene set defined <i>a priori</i>. Many self-contained gene set analysis methods have been developed but the performance of these methods for phenotypes that are continuous rather than discrete and with multiple nuisance covariates has not been well studied. Here, I use Monte Carlo simulation to evaluate the performance of both novel and previously published (and readily av ...[more]

PMID: 27761350

Similar Datasets

Project description:BACKGROUND: Different high-dimensional regression methodologies exist for the selection of variables to predict a continuous variable. To improve the variable selection in case clustered observations are present in the training data, an extension towards mixed-effects modeling (MM) is requested, but may not always be straightforward to implement.In this article, we developed such a MM extension (GA-MM-MMI) for the automated variable selection by a linear regression based genetic algorithm (GA) using multi-model inference (MMI). We exemplify our approach by training a linear regression model for prediction of resistance to the integrase inhibitor Raltegravir (RAL) on a genotype-phenotype database, with many integrase mutations as candidate covariates. The genotype-phenotype pairs in this database were derived from a limited number of subjects, with presence of multiple data points from the same subject, and with an intra-class correlation of 0.92. RESULTS: In generation of the RAL model, we took computational efficiency into account by optimizing the GA parameters one by one, and by using tournament selection. To derive the main GA parameters we used 3 times 5-fold cross-validation. The number of integrase mutations to be used as covariates in the mixed effects models was 25 (chrom.size). A GA solution was found when R2MM > 0.95 (goal.fitness). We tested three different MMI approaches to combine the results of 100 GA solutions into one GA-MM-MMI model. When evaluating the GA-MM-MMI performance on two unseen data sets, a more parsimonious and interpretable model was found (GA-MM-MMI TOP18: mixed-effects model containing the 18 most prevalent mutations in the GA solutions, refitted on the training data) with better predictive accuracy (R2) in comparison to GA-ordinary least squares (GA-OLS) and Least Absolute Shrinkage and Selection Operator (LASSO). CONCLUSIONS: We have demonstrated improved performance when using GA-MM-MMI for selection of mutations on a genotype-phenotype data set. As we largely automated setting the GA parameters, the method should be applicable on similar datasets with clustered observations.

Project description:Ionic transport in solid electrolytes can often be approximated as ions performing a sequence of hops between distinct lattice sites. If these hops are uncorrelated, quantitative relationships can be derived that connect microscopic hopping rates to macroscopic transport coefficients; i.e. tracer diffusion coefficients and ionic conductivities. In real materials, hops are uncorrelated only in the dilute limit. At non-dilute concentrations, the relationships between hopping frequency, diffusion coefficient and ionic conductivity deviate from the random walk case, with this deviation quantified by single-particle and collective correlation factors, f and fI, respectively. These factors vary between materials, and depend on the concentration of mobile particles, the nature of the interactions, and the host lattice geometry. Here, we study these correlation effects for the garnet lattice using lattice-gas Monte Carlo simulations. We find that, for non-interacting particles (volume exclusion only), single-particle correlation effects are more significant than for any previously studied three-dimensional lattice. This is attributed to the presence of two-coordinate lattice sites, which causes correlation effects intermediate between typical three-dimensional and one-dimensional lattices. Including nearest-neighbour repulsion and on-site energies produces more complex single-particle correlations and introduces collective correlations. We predict particularly strong correlation effects at xLi=3 (from site energies) and xLi=6 (from nearest-neighbour repulsion), where xLi=9 corresponds to a fully occupied lithium sublattice. Both effects are consequences of ordering of the mobile particles. Using these simulation data, we consider tuning the mobile-ion stoichiometry to maximize the ionic conductivity, and show that the 'optimal' composition is highly sensitive to the precise nature and strength of the microscopic interactions. Finally, we discuss the practical implications of these results in the context of lithium garnets and other solid electrolytes.

Project description:Mixture models of item response theory (IRT) can be used to detect inappropriate category use. Data collected by panel surveys where attitudes and traits are typically assessed by short scales with many response categories are prone to response styles indicating inappropriate category use. However, the application of mixed IRT models to this data type can be challenging because of many threshold parameters within items. Up to now, there is very limited knowledge about the sample size required for an appropriate performance of estimation methods as well as goodness-of-fit criteria of mixed IRT models in this case. The present Monte Carlo simulation study examined these issues for two mixed IRT models [the restricted mixed generalized partial credit model (rmGPCM) and the mixed partial credit model (mPCM)]. The population parameters of the simulation study were taken from a real application to survey data which is challenging (a 5-item scale with an 11-point rating scale, and three latent classes). Additional data conditions (e.g., long tests, a reduced number of response categories, and a simple latent mixture) were included in this simulation study to improve the generalizability of the results. Under this challenging data condition, for each model, data were generated based on varying sample sizes (from 500 to 5,000 observations with a 500-step). For the additional conditions, only three sample sizes (consisting of 1,000, 2,500, and 4,500 observations) were examined. The effect of sample size on estimation problems and accuracy of parameter and standard error estimates were evaluated. Results show that the two mixed IRT models require at least 2,500 observations to provide accurate parameter and standard error estimates under the challenging data condition. The rmGPCM produces more estimation problems than the more parsimonious mPCM, mostly because of the sparse tables arising due to many response categories. These models exhibit similar trends of estimation accuracy across sample sizes. Under the additional conditions, no estimation problems are observed. Both models perform well with a smaller sample size when long tests were used or a true latent mixture includes two classes. For model selection, the AIC3 and the SABIC are the most reliable information criteria.

Dataset Information

Monte Carlo simulation of OLS and linear mixed model inference of phenotypic effects on gene expression.

Publications

Monte Carlo simulation of OLS and linear mixed model inference of phenotypic effects on gene expression.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets