Skip to main content

EDITORIAL article

Front. Pharmacol., 29 August 2023
Sec. Pharmacogenetics and Pharmacogenomics
This article is part of the Research Topic The Potential of Machine-learning in Pharmacogenetics, Pharmacogenomics and Pharmacoepidemiology: Volume II View all 5 articles

Editorial: The Potential of Machine-learning in Pharmacogenetics, Pharmacogenomics and Pharmacoepidemiology: Volume II

  • 1University of California San Francisco, San Francisco, CA, United States
  • 2University Institute of Molecular Pathology Biomarkers, Universidad de Extremadura, Caceres, Spain
  • 3Institute for Applied Medical Informatics, University of Tübingen, Tübingen, Germany

The advent of novel deep learning methods shows great promise in diverse fields of knowledge in pharmacology. Specifically, its capability to quickly process vast amounts of data, discovering complex patterns that would otherwise be imperceptible, will permit crucial advances for drug discovery and repurposing and personalized medicine. In this Frontiers Research Topic, we invited authors to submit their latest research in deep learning applied to pharmacogenetics, pharmacogenomics, and pharmacoepidemiology.

Deep learning methods are valuable tools in discovering the mechanisms of hitherto not fully understood adverse drug reactions. In their study, Akimoto et al. aimed to design a more accurate model to assess drug interactions in drug-induced liver injury (DILI). They found that the combinations diclofenac-famotidine, acetaminophen-ambroxol, and aspirin-cilostazol showed relatively higher excess risk due to a significant interaction. Moreover, both diclofenac and famotidine individually increase the risk of developing DILI. They also found that extreme gradient boosting outperformed more traditional learning algorithms, and was able to reduce overfitting.

Another field where AI can significantly improve clinical pipelines is sifting through vast amounts of inconsistent and sparse data. In Breitenstein et al. study, the authors aimed to identify patterns in medication switches and add-ons in a cohort of epilepsy individuals aged 65 or older by looking at their medication consumption, where despite clinical recommendations being clear, there was lacking evidence on how closely these guidelines were being followed. Their approach was able to capture 92% of all add-ons and 88% of all switches.

In their study, Fang et al. researched the genomic aspects of pathogenesis and drug repurposing in acute type A aortic dissection (ATAAD). Relevant expressed genes were identified, and based on these findings, calcium channel blockers and glucocorticoid receptor agonists were flagged as potentially repurposable drugs to treat ATAAD. An empirical Bayesian test was used to identify the differentially expressed genes through which potential drugs were identified, highlighting the relevance of deep learning techniques in pharmacogenomics for drug repurposing.

Finally, Grant et al. present a study which identifies multi-dimensional *-omics factors relevant to measuring antidepressant response in patients with major depressive disorder with a history of attempted suicide, aiming to discover differential factors that could explain differences in drug effectiveness. In this case, a least squares regression model was employed instead. The authors found the most significant differences between groups in the circadian genes CLOCK and ARNTL.

In this Frontiers Research Topic, we selected four manuscripts for their innovative approach to applying deep learning techniques to vast and multimodal data sources that can be leveraged for pharmacogenetics and pharmacogenomics, particularly in the fields of drug repurposing, -omics and personal factors in drug efficacy, and identifying complex adverse drug events. We are confident that as AI, particularly large autoregressive models, become more efficient, accessible, and adaptable to multiple data types, novel discoveries in these and other fields of pharmacogenetics and pharmacogenomics that were hitherto unattainable will become possible in the coming years. As guest editors for Frontiers in Pharmacology, we are excited to learn what advances scientists across the globe will achieve in pharmacogenetics, pharmacogenomics and pharmacoepidemiology, especially in a context where novel drug research is becoming both more relevant and challenging.

Author contributions

AG-A: Writing–original draft, Writing–review and editing. EG-M: Writing–original draft, Writing–review and editing. CE: Writing–original draft, Writing–review and editing.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The author(s) declared that they were an editorial board member of Frontiers, at the time of submission. This had no impact on the peer review process and the final decision.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Keywords: pharmacogenetics, pharmacogenomics, pharmacoepidemiology, machine learning, deep learning

Citation: Garcia-Agundez A, Garcia-Martin E and Eickhoff C (2023) Editorial: The Potential of Machine-learning in Pharmacogenetics, Pharmacogenomics and Pharmacoepidemiology: Volume II. Front. Pharmacol. 14:1277561. doi: 10.3389/fphar.2023.1277561

Received: 14 August 2023; Accepted: 24 August 2023;
Published: 29 August 2023.

Edited and reviewed by:

Luis Abel Quiñones, University of Chile, Chile

Copyright © 2023 Garcia-Agundez, Garcia-Martin and Eickhoff. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

*Correspondence: Augusto Garcia-Agundez, YXVndXN0by5nYXJjaWFAdWNzZi5lZHU=

Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.