OpenADMET Science seminar

The OpenADMET Consortium will hold its next internal seminar on February 16, 2026, bringing together internal members and renowned consortium members.

We are especially proud that Dr Ruel Cedeno, Head of Cheminformatics at Novalix, and one of the top performers in the consortium’s recent blind-prediction challenges, has been invited for a deep dive into our AI/ML approaches in ADME prediction. This talk is based on our recent publication in the Journal of Chemical Information and Modeling (JCIM), a top-tier journal in cheminformatics of the American Chemical Society.

AI/ML approaches in ADME prediction

Abstract:
Reliable prediction of ADME properties is critical in drug discovery. While AI and deep learning have seen rapid adoption, it remains unclear whether they consistently outperform classical methods. Thus, blind competitions are crucial, as they offer an unbiased assessment of various approaches on identical datasets. In this talk, I will describe our approach in the recent ASAP-Polaris-OpenADMET Antiviral Challenge. Using rigorous statistical benchmarking, our results show that classical methods remain highly competitive for potency prediction, whereas deep learning excels in ADME modeling. We also highlight the value of careful data curation and feature augmentation with public datasets. These findings provide practical guidance for building robust predictive models and offer insights into the evolving role of computational methods in drug discovery.

Reference:  J. Chem. Inf. Model. 2025, 65, 24, 13115–13131

OpenADMET is an organization with the mission to advance predictive modeling of ADMET properties (Absorption, Distribution, Metabolism, Excretion, and Toxicity). With a growing consortium members (currently involving OMSF, UCSF, Octant Inc, and MSKCC) and funding from ARPA-H, the Gates Foundation, Schrödinger, and the Astera Institute, OpenADMET aims to drive innovation through open datasets, community challenges, and collaborative tools. 

These seminar series will encourage members to evaluate methodologies, question assumptions, and identify shared opportunities for advancing computational drug discovery pipelines.

Looking for more details on ADME-PK? Check our dedicated section.