Integrative transcriptome imputation reveals tissue-specific and shared biological mechanisms mediating susceptibility to complex traits

Abstract

Transcriptome-wide association studies integrate gene expression data with common risk variation to identify gene-trait associations. By incorporating epigenome data to estimate the functional importance of genetic variation on gene expression, we generate a small but significant improvement in the accuracy of transcriptome prediction and increase the power to detect significant expression-trait associations. Joint analysis of 14 large-scale transcriptome datasets and 58 traits identify 13,724 significant expression-trait associations that converge on biological processes and relevant phenotypes in human and mouse phenotype databases. We perform drug repurposing analysis and identify compounds that mimic, or reverse, trait-specific changes. We identify genes that exhibit agonistic pleiotropy for genetically correlated traits that converge on shared biological pathways and elucidate distinct processes in disease etiopathogenesis. Overall, this comprehensive analysis provides insight into the specificity and convergence of gene expression on susceptibility to complex traits.

kTWAS: Integrating kernel-machine with transcriptome-wide association studies improves statistical power and reveals novel genes

“…Remarkably, this approach remains effective even when the predictive power of the genotype-expression model is low. As a result, despite having average R 2 values around 1%, the use of genotype-expression models in TWAS has led to significant successes in real data analyses [3,4,[13][14][15][16][17]. Indeed, as demonstrated in our simulations [18], predicted expressions generated by a genotype-phenotype model can perform better than actual expression data when applying TWAS analysis.…”

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

“…Proof. For LMM (1), or equivalently the transformed model (27), the log-likelihood function is Then the MLE estimating equations for 9? and 9@ are:…”

“…Initiated by Gusev et al [19], methods using summary statistics in GWAS datasets (i.e., meta-analysis) to conduct TWAS were also developed [20]. Since then, TWAS has achieved significant success in identifying the genetic basis of complex traits [21][22][23][24][25][26][27] and became a popular method in practice. Inspired by TWAS, researchers have developed multiple similar protocols using other endophenotypes to conduct association mappings.…”

Fine-mapping and QTL tissue-sharing information improve causal gene identification and transcriptome prediction performance

“…28 Also, important methodological progress in fine-mapping [16,17]{Wen2017, Wang2018} 29 and an adaptive shrinkage method that improves effect size estimates across multi- 30 ple experiments [18]{Urbut2019} provide opportunities to further improve quality of 31 downstream associations. 32 In this article, we analyze different transcriptome prediction strategies and compare 33 their strengths both in prediction performance and downstream phenotypic associations. 34 Proximity and linkage disequilibrium (LD) between distinct causal variants can 35 lead to non causal associations between predicted expression and complex traits [9, 36 19]{Barbeira2018, Wainberg:2019kq}.…”

“…Potential developments could 262 rely on fine-mapping methods that jointly incorporate cross-tissue patterns, or consensus 263 between different fine-mapping approaches. Also, epigenetic information has been 264 shown to improve transcriptome prediction [33]{Zhang2019EpiXcan} as well. Future 265 improvements should incorporate this epigenetic information and other biologically-266 informed annotations jointly.…”