AI Insight
This study evaluated computational methods for predicting how genetic variants affect drug metabolism, focusing on pharmacogenomic variants including those in CYP2C9, a key drug-metabolizing enzyme. Researchers found that while current prediction tools perform moderately well for loss-of-function variants that reduce drug clearance, they struggle to identify gain-of-function variants associated with faster drug metabolism. The coevolution-based method StructureDCA outperformed newer deep learning approaches like AlphaMissense and protein language models, suggesting that incorporating structural and evolutionary data improves prediction accuracy for pharmacogenomic applications.
Why it matters
Accurate prediction of how genetic variants affect drug metabolism is critical for personalized medicine, helping clinicians anticipate individual patient responses to medications and adjust dosing accordingly. The findings highlight current limitations in computational tools and suggest that combining multiple data types, particularly structural and evolutionary information, could enhance clinical decision-making when genetic testing reveals variants of uncertain significance.
Understand the Science
⚠️ Preprint – Noch nicht peer-reviewed
Dieser Artikel wurde noch nicht von unabhängigen Experten begutachtet. Die Ergebnisse sind vorläufig und sollten mit Vorsicht interpretiert werden.
Accurately predicting the effects of pharmacogenomic variants is essential for the development of personalized therapeutic strategies, as genetic variability can influence drug response differently across patients. Here, we assessed several computational approaches using a dataset of pharmacogenomic variants with either clinical annotations or functional characterization by deep mutational scanning, compiled from the literature, with an additional focus on CYP2C9, a clinically relevant drug-metabolizing enzyme. Our results show that, despite recent methodological advances, substantial room for improvement remains. In particular, current methods struggle to distinguish gain-of-function variants associated with increased drug clearance and fast-metabolizer phenotypes from neutral variants, whereas loss-of-function variants that reduce drug clearance are predicted more accurately. The integration of structural and evolutionary information appears to be a key strategy for improving performance, with the coevolution-based StructureDCA method achieving the highest accuracy compared with classical genetic variant-effect predictors and recent deep learning approaches, including the pathogenic-variant predictor AlphaMissense and general protein language model-based methods. Finally, our results indicate that computational models can complement in vitro experiments in clinical variant interpretation, as StructureDCA predictions showed better agreement with clinically annotated phenotypes than large-scale deep mutational scanning data in several cases.
Source: Assessing Computational Models for Pharmacogenomic Variant Interpretation