TIGAR: An Improved Bayesian Tool for Transcriptomic Data Imputation Enhances Gene Mapping of Complex Traits.

Area of Science:

• Genetics
• Bioinformatics
• Computational Biology

Background:

• Transcriptome-wide association studies (TWASs) identify genetic risk loci for complex traits by associating traits with imputed gene expression.
• Current TWAS tools like PrediXcan and FUSION use parametric imputation models that struggle with complex genetic architectures of transcriptomic data.

Purpose of the Study:

• To develop and evaluate a novel nonparametric Bayesian method for gene expression imputation in TWASs.
• To improve the power of TWASs for discovering genetic risk loci.

Main Methods:

• Employed a nonparametric Bayesian method with a data-driven prior for cis-eQTL effect sizes, generalizing existing parametric models.
• Implemented both parametric and nonparametric methods in the TIGAR software for imputation and TWAS analysis.

Main Results:

• The nonparametric Bayesian model demonstrated improved imputation R-squared and TWAS power over PrediXcan under specific conditions.
• Real-world applications showed the nonparametric method successfully imputed models for 57.8% more genes than PrediXcan, boosting TWAS power.

Conclusions:

• The nonparametric Bayesian approach offers a flexible and powerful alternative for gene expression imputation in TWASs.
• The TIGAR software facilitates the application of these methods for enhanced genetic discovery in complex traits.