Testing and controlling for horizontal pleiotropy with probabilistic Mendelian randomization in transcriptome-wide association studies

Yuan, Zhongshang; Zhu, Huanhuan; Zeng, Ping; Yang, Sheng; Sun, Shuang; Yang, Can; Liu, Jin; Zhou, Xiang

doi:10.1038/s41467-020-17668-6

Cited by 97 publications

(123 citation statements)

References 69 publications

Supporting

Mentioning

122

Contrasting

Order By: Relevance

“…A recently derived collaborative mixed model (CoMM) [ 20 , 21 ] likewise accounts for the uncertainty of eQTL effect size estimates from reference transcriptome data by jointly modeling reference and test data within a linear mixed-model framework. The PMR-Egger [ 22 ] and moPMR-Egger [ 23 ] also take the approach of jointly modeling reference and test data to address the uncertainty of eQTL effect size estimates from the reference data. However, these methods that are based on the maximum likelihood inference framework and implement likelihood ratio tests, are derived only for quantitative phenotypes and could suffer computation burden for testing thousands of cis-SNPs per gene, which often happens in practice particularly when considering imputed SNP data or whole-genome sequencing data for TWAS.…”

Section: Introductionmentioning

confidence: 99%

Novel Variance-Component TWAS method for studying complex human diseases with applications to Alzheimer’s dementia

et al. 2021

View full text Add to dashboard Cite

Transcriptome-wide association studies (TWAS) have been widely used to integrate transcriptomic and genetic data to study complex human diseases. Within a test dataset lacking transcriptomic data, traditional two-stage TWAS methods first impute gene expression by creating a weighted sum that aggregates SNPs with their corresponding cis-eQTL effects on reference transcriptome. Traditional TWAS methods then employ a linear regression model to assess the association between imputed gene expression and test phenotype, thereby assuming the effect of a cis-eQTL SNP on test phenotype is a linear function of the eQTL’s estimated effect on reference transcriptome. To increase TWAS robustness to this assumption, we propose a novel Variance-Component TWAS procedure (VC-TWAS) that assumes the effects of cis-eQTL SNPs on phenotype are random (with variance proportional to corresponding reference cis-eQTL effects) rather than fixed. VC-TWAS is applicable to both continuous and dichotomous phenotypes, as well as individual-level and summary-level GWAS data. Using simulated data, we show VC-TWAS is more powerful than traditional TWAS methods based on a two-stage Burden test, especially when eQTL genetic effects on test phenotype are no longer a linear function of their eQTL genetic effects on reference transcriptome. We further applied VC-TWAS to both individual-level (N = ~3.4K) and summary-level (N = ~54K) GWAS data to study Alzheimer’s dementia (AD). With the individual-level data, we detected 13 significant risk genes including 6 known GWAS risk genes such as TOMM40 that were missed by traditional TWAS methods. With the summary-level data, we detected 57 significant risk genes considering only cis-SNPs and 71 significant genes considering both cis- and trans- SNPs; these findings also validated our findings with the individual-level GWAS data. Our VC-TWAS method is implemented in the TIGAR tool for public use.

show abstract

Section: Introductionmentioning

confidence: 99%

Novel Variance-Component TWAS method for studying complex human diseases with applications to Alzheimer’s dementia

et al. 2021

View full text Add to dashboard Cite

show abstract

“…Furthermore, several expression quantitative trait loci (eQTLs) studies have illustrated that the information on expression regulation may play a pivotal role in disease development ( Albert and Kruglyak, 2015 ). Transcriptome-wide association study (TWAS) is widely utilized in integrating GWAS with eQTL studies for investigating the causal genes associated with complex traits or diseases ( Gamazon et al, 2015 ; Gusev et al, 2016 ; Yuan et al, 2020 ). Therefore, TWAS analysis may help us to identify novel genes associated with IPF.…”

Section: Introductionmentioning

confidence: 99%

Integrative Analysis of Transcriptome-Wide Association Study and mRNA Expression Profiles Identifies Candidate Genes Associated With Idiopathic Pulmonary Fibrosis

et al. 2020

Self Cite

View full text Add to dashboard Cite

Idiopathic pulmonary fibrosis (IPF) is a type of scarring lung disease characterized by a chronic, progressive, and irreversible decline in lung function. The genetic basis of IPF remains elusive. A transcriptome-wide association study (TWAS) of IPF was performed by FUSION using gene expression weights of three tissues combined with a large-scale genome-wide association study (GWAS) dataset, totally involving 2,668 IPF cases and 8,591 controls. Significant genes identified by TWAS were then subjected to gene ontology (GO) and pathway enrichment analysis. The overlapped GO terms and pathways between enrichment analysis of TWAS significant genes and differentially expressed genes (DEGs) from the genome-wide mRNA expression profiling of IPF were also identified. For TWAS significant genes, protein–protein interaction (PPI) network and clustering modules analyses were further conducted using STRING and Cytoscape. Overall, TWAS identified a group of candidate genes for IPF under the Bonferroni corrected P value threshold (0.05/14929 = 3.35 × 10–6), such as DSP (PTWAS = 1.35 × 10–29 for lung tissue), MUC5B (PTWAS = 1.09 × 10–28 for lung tissue), and TOLLIP (PTWAS = 1.41 × 10–15 for whole blood). Pathway enrichment analysis identified multiple candidate pathways, such as herpes simplex infection (P value = 7.93 × 10–5) and antigen processing and presentation (P value = 6.55 × 10–5). 38 common GO terms and 8 KEGG pathways shared by enrichment analysis of TWAS significant genes and DEGs were identified. In the PPI network, 14 genes (DYNLL1, DYNC1LI1, DYNLL2, HLA-DRB5, HLA-DPB1, HLA-DQB2, HLA-DQA2, HLA-DQB1, HLA-DRB1, POLR2L, CENPP, CENPK, NUP133, and NUP107) were simultaneously detected by hub gene and module analysis. In conclusion, through integrative analysis of TWAS and mRNA expression profiles, we identified multiple novel candidate genes, GO terms and pathways for IPF, which contributes to the understanding of the genetic mechanism of IPF.

show abstract

“…where y and ε y are defined as in Equation (2);

is the N × p genotype matrix for p cis-SNPs in the given gene; w m is the same SNP effects on gene expression as defined in Equation (7); the scalar

represents the genetic effect of GReX (i.e.,

) on y and can be interpreted as the causal effect of GReX on y (Yuan et al, 2019 ; Zhu and Zhou, 2020 ); and

is the q -length vector of alternative genetic effects; note that

is not the same genotype matrix as

, and the q SNPs in

are those who show direct horizontal effects on y , such as the trans-eQTLs and SNPs associated with alternative splicing events (Matlin et al, 2005 ).…”

Section: Methods Based On Tissue-specific Genetically Regulated Exprementioning

confidence: 99%

A Review of Statistical Methods for Identifying Trait-Relevant Tissues and Cell Types

2021

Self Cite

View full text Add to dashboard Cite

Genome-wide association studies (GWASs) have identified and replicated many genetic variants that are associated with diseases and disease-related complex traits. However, the biological mechanisms underlying these identified associations remain largely elusive. Exploring the biological mechanisms underlying these associations requires identifying trait-relevant tissues and cell types, as genetic variants likely influence complex traits in a tissue- and cell type-specific manner. Recently, several statistical methods have been developed to integrate genomic data with GWASs for identifying trait-relevant tissues and cell types. These methods often rely on different genomic information and use different statistical models for trait-tissue relevance inference. Here, we present a comprehensive technical review to summarize ten existing methods for trait-tissue relevance inference. These methods make use of different genomic information that include functional annotation information, expression quantitative trait loci information, genetically regulated gene expression information, as well as gene co-expression network information. These methods also use different statistical models that range from linear mixed models to covariance network models. We hope that this review can serve as a useful reference both for methodologists who develop methods and for applied analysts who apply these methods for identifying trait relevant tissues and cell types.

show abstract

Testing and controlling for horizontal pleiotropy with probabilistic Mendelian randomization in transcriptome-wide association studies

Cited by 97 publications

References 69 publications

Novel Variance-Component TWAS method for studying complex human diseases with applications to Alzheimer’s dementia

Novel Variance-Component TWAS method for studying complex human diseases with applications to Alzheimer’s dementia

Integrative Analysis of Transcriptome-Wide Association Study and mRNA Expression Profiles Identifies Candidate Genes Associated With Idiopathic Pulmonary Fibrosis

A Review of Statistical Methods for Identifying Trait-Relevant Tissues and Cell Types

Contact Info

Product

Resources

About