Power analysis of transcriptome-wide association study: Implications for practical protocol choice

Abstract

The transcriptome-wide association study (TWAS) has emerged as one of several promising techniques for integrating multi-scale ‘omics’ data into traditional genome-wide association studies (GWAS). Unlike GWAS, which associates phenotypic variance directly with genetic variants, TWAS uses a reference dataset to train a predictive model for gene expressions, which allows it to associate phenotype with variants through the mediating effect of expressions. Although effective, this core innovation of TWAS is poorly understood, since the predictive accuracy of the genotype-expression model is generally low and further bounded by expression heritability. This raises the question: to what degree does the accuracy of the expression model affect the power of TWAS? Furthermore, would replacing predictions with actual, experimentally determined expressions improve power? To answer these questions, we compared the power of GWAS, TWAS, and a hypothetical protocol utilizing real expression data. We derived non-centrality parameters (NCPs) for linear mixed models (LMMs) to enable closed-form calculations of statistical power that do not rely on specific protocol implementations. We examined two representative scenarios: causality (genotype contributes to phenotype through expression) and pleiotropy (genotype contributes directly to both phenotype and expression), and also tested the effects of various properties including expression heritability. Our analysis reveals two main outcomes: (1) Under pleiotropy, the use of predicted expressions in TWAS is superior to actual expressions. This explains why TWAS can function with weak expression models, and shows that TWAS remains relevant even when real expressions are available. (2) GWAS outperforms TWAS when expression heritability is below a threshold of 0.04 under causality, or 0.06 under pleiotropy. Analysis of existing publications suggests that TWAS has been misapplied in place of GWAS, in situations where expression heritability is low.

TIGAR-V2: Efficient TWAS tool with nonparametric Bayesian eQTL weights of 49 tissue types from GTEx V8

“…Additionally, a recent power analysis of TWASs suggested useful threshold of expression heritability >0.04 for a causal model where gene expression is directly causal with respect to the phenotype, and a threshold of expression heritability >0.06 for a pleiotropy model where true causal SNPs of the phenotype are also true causal eQTLs with respect to gene expression, 74 which allowed a TWAS that had higher power than a single-variant GWAS for a simulation cohort with sample size 2,504 that was used as both training and test data. We would only suggest TWAS as a secondary analysis to standard single-variant GWAS, instead of as a competing analysis.…”

webTWAS: a resource for disease candidate susceptibility genes identified by transcriptome-wide association study

“…S-PrediXcan is an extension of PrediXcan which allows PrediXcan's results to be computed from summary statistics ( 25 ). Although the Top1 model is underpowered according to previous studies ( 5 , 42 ), we have retained the Top1 model as it is integrated in the TWAS-FUSION software package. To remind users of this issue, we highlight the Top1 model with the label ‘The Top1 model is underpowered according to previous studies, and is included in webTWAS for reference purposes’.…”

“…This problem is due to linkage disequilibrium (LD) between causal and non-causal variants, which masks the effects of causal variants on the phenotype of interest ( 4 ). TWAS mitigates this interpretation issue by prioritizing potential causal genes in addition to genetic variants ( 4 , 5 ).…”

Identification of shared and differentiating genetic risk for autism spectrum disorder, attention deficit hyperactivity disorder and case subgroups

“…For the construction of the transcriptomic imputation models we used EpiXcan 20 , an elastic net based method, which weighs SNPs based on available epigenetic annotation information 72 . EpiXcan was recently shown to increase power to identify genes under a causality model when compared to TWAS approaches that don’t integrate epigenetic information 73 . We use this model (924 samples from DLPFC) due to power considerations 20 ; in comparison, brain gene expression imputation models based on GTEx V8 74 are trained in 205 or fewer samples.…”