Multi-tissue integrated Mendelian randomization method identifies disease risk genes

Area of Science:

• Genetics
• Bioinformatics
• Statistical genetics

Background:

• Mendelian randomization (MR) uses genetic variants to infer causality between molecular traits and diseases.
• Identifying specific causal genes for disease risk using MR remains a significant challenge.

Purpose of the Study:

• To present MULTI (Multi-tissue Unified Likelihood-based Transcriptomic Integration), a Bayesian MR framework.
• To improve causal inference accuracy by integrating genetic information across multiple tissues.
• To enhance statistical power and identify tissue-specific causal mechanisms.

Main Methods:

• Developed a Bayesian MR framework named MULTI.
• Integrated genetic information across multiple tissues.
• Employed adaptive tissue information integration to boost power without increasing type I error rates.

Main Results:

• MULTI provides reliable causal inference estimates, even with limited instruments in a single tissue.
• The framework adaptively integrates information from similar tissues, increasing statistical power.
• Simulations confirmed MULTI's robustness across various genetic architectures.
• Real-data applications revealed tissue-specific causal mechanisms and cross-tissue regulation.

Conclusions:

• MULTI offers a practical and extensible framework for causal gene discovery in complex diseases.
• The method aids in elucidating the molecular architecture underlying human diseases.
• Facilitates a deeper understanding of tissue-specific and coordinated regulatory mechanisms in disease etiology.