Efficient molecular mendelian randomization screens with LaScaMolMR.jl

Abstract

SummaryThe ever-growing genetic cohorts lead to an increase in scale of molecular Quantitative Trait Loci (QTL) studies, creating opportunities for more extensive two samples Mendelian randomization (MR) investigations aiming to identify causal relationships between molecular traits and diseases. This increase led to the identification of multiple causal candidates and potential drug targets over time. However, the increase in scale of such studies and higher dimension multi-omic data come with computational challenges. We present "LArge SCAle MOLecular Mendelian Randomization with Julia" (LaScaMolMR.jl), an open-sourced integrated Julia package optimized for Omic-wide Mendelian Randomization (OWMR) Studies. This versatile package eliminates the two-language problem and implements fast algorithms for instrumental variable selection approaches with both cis and trans instruments and performs the most popular regression estimators for MR studies with molecular exposures. It reduces the compute time via meta-programming allowing easy deployment of multi-threaded approach and the internalization of linkage disequilibrium investigation of potential instrumental variables. Via its integrated approach and high-computational performance, LaScaMolMR.jl allows users who have minimal programming experience to perform large scale OWMR studies. Implementation and availabilityLaScaMolMR is freely available at github.com/SamuelMathieu-code/LaScaMolMR.jl.