|
Abstract:
|
Mendelian Randomization (MR) is an emerging statistical approach that uses genetic variants as instrumental variables to estimate the causal effects of risk factors on outcomes. In contrast to the traditional mediation test requiring individual-level data, applying MR allows us to assess the causality between putative exposure traits and disease outcomes through using the increasingly large amount of summary statistics available from genome-wide association studies (GWAS).To validate therapeutic targets and improve the likelihood of success in drug discovery, we have performed large scale causal inference analyses through MR to provide a systematic and robust validation of putative therapeutic targets and strategies in humans prior to clinical trials. Utilizing the summary level data from nearly 3,000 GWAS, we build a phenotypic trait network that analyzes 2.7 million trait-trait causal effect relationships using multiple canonical MR regression methods. The analysis confirms many published and validated trait relationships, but also discovers some potentially novel associations for future drug development. This framework for understanding the genetic architecture of diseases can b
|
