Separating the genetics of disease, treatment and treatment response using graphical modeling and large-scale electronic health records

The preprint operates within the established paradigm of genome-wide association studies (GWAS) and their application to pharmacogenomics. It adheres to the standard assumptions of this paradigm, such as the common disease-common variant hypothesis and the additive effect of genetic variants. It utilizes established methodologies for genomic data analysis and statistical inference.

This preprint is classified as Normal Science because it utilizes established methodologies and concepts within the field of pharmacogenomics and extends them by addressing recognized limitations. Its primary focus is refining the tools and methods employed within the existing paradigm, rather than challenging its fundamental tenets. The study develops a novel graphical model to account for and dissect the influence of time, treatment, and treatment response on the genetic components of complex traits, such as blood pressure. While it reports novel findings regarding potential pharmacogenetic markers, these discoveries are presented as extensions and refinements of existing knowledge, rather than paradigm-shifting revelations. The paper acknowledges the limitations of statistical power for certain analyses, especially those concerning binary treatment indicators, which is a common challenge addressed within Normal Science. Despite its inability to identify any associations with post-treatment blood pressure, it emphasizes this negative finding as a significant contribution within the existing paradigm, highlighting the limitations of current methods in detecting genetic effects on drug response while underscoring the need for further research and larger sample sizes, all typical hallmarks of Normal Science.