Meta-imputation of transcriptome from genotypes across multiple datasets by leveraging publicly available...

Area of Science:

• Genetics
• Bioinformatics
• Systems Biology

Background:

• Transcriptome-wide association studies (TWAS) link genetic variants to complex traits by analyzing gene expression.
• Current TWAS methods often rely on single-tissue imputation models, limiting accuracy and discovery.
• Existing multi-tissue imputation methods require extensive data access and cannot utilize non-overlapping datasets.

Purpose of the Study:

• To develop a flexible method for combining multiple transcriptome imputation models using summary-level data.
• To improve the accuracy of gene expression imputation and enhance the power of TWAS.
• To leverage diverse expression datasets across tissues and individuals without requiring joint data.

Main Methods:

• Proposed a novel method, SWAM (Summary-level Weighted Average Meta-imputation), for linearly optimizing transcriptome imputation accuracy.
• Integrated 49 tissue-specific gene expression imputation models from GTEx and DGN projects.
• Extended the meta-imputation approach to meta-TWAS using summary-level statistics for multi-tissue TWAS.

Main Results:

• SWAM demonstrated improved imputation accuracy by combining models across tissues and datasets.
• The method successfully leveraged multiple tissues and datasets, even with non-overlapping individuals.
• Meta-TWAS analysis showed the benefit of integrating multiple tissues for identifying genetic variant regulatory impacts.

Conclusions:

• SWAM offers a flexible and accurate approach to transcriptome imputation using summary-level data.
• Integrating multiple tissues and datasets significantly enhances TWAS power and biological interpretation.
• This work highlights the importance of multi-tissue integration for understanding genetic regulation of complex traits.