EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Evaluating Machine Learning Affinity Prediction Methods on Novel Lead Optimization Data: A 15‐PGDH Case Study

Alexander S Rich, Leela S Dodda, Daniel J Price iD

ChemMedChem · 28.10.2026 · 21 (20)

DOI: 10.1002/cmdc.70531 ↗

Kurz gefasst

Fachartikel in der Zeitschrift ChemMedChem, online erschienen am 18.10.2026. Verfasst von Alexander S Rich und 2 weiteren Autor:innen. Beteiligte Einrichtung: Inductive Bio, Brooklyn, New York, USA. Hauptthemen laut MEDLINE: Machine Learning, Hydroxyprostaglandin Dehydrogenases, Enzyme Inhibitors. Die Arbeit nennt 31 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Recent machine learning (ML) methods for protein–ligand affinity prediction, such as Boltz‐2 and AEV‐PLIG, have been reported to approach free energy perturbation (FEP) accuracy on well‐known benchmarks. Even short of that accuracy, ML‐based potency models that fall between FEP and common docking scoring functions in accuracy and speed offer clear potential value. Here, we report a case study using recently disclosed data from four related ligand series targeting 15‐hydroxyprostaglandin dehydrogenase (15‐PGDH). We evaluate each ML method's predictive accuracy with a focus on when it could be used to influence synthetic priorities. AEV‐PLIG performed poorly across most series, and adding literature 15‐PGDH inhibitors to its training did not improve performance. Boltz‐2 performed well on series most similar to published inhibitors in its training, achieving accurate cofolding (median ligand RMSD ≤0.81 Å) and rank order. However, for a series exploring a new region of the binding pocket, Boltz‐2 sometimes generated less accurate poses and predicted all ligands in the series to be less potent than a published exemplar, despite experimental potency gains of up to 500x. These results highlight both the promise in well‐characterized chemical space and limitations in the novel regions where computational guidance is most needed.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Machine LearningHydroxyprostaglandin DehydrogenasesEnzyme InhibitorsLigandsThermodynamicsHumansMolecular StructurePredictive Learning ModelsPrediction AlgorithmsBinding SitesMolecular Docking SimulationStructure-Activity RelationshipProtein Binding

Stichwörter der Autor:innen

binding affinity predictioncofoldingfree energy perturbationlead optimizationmachine learning

Substanzen

LigandsHydroxyprostaglandin Dehydrogenases15-hydroxyprostaglandin dehydrogenaseEnzyme Inhibitors

Autor:innen und Institutionen

  • Inductive Bio, Brooklyn, New York, USA.
  • Nimbus Therapeutics, Cambridge, Massachusetts, USA.

ORCID-Profile: Daniel J Price

Zitieren

APA

Rich, A. S., Dodda, L. S., & Price, D. J. (2026). Evaluating Machine Learning Affinity Prediction Methods on Novel Lead Optimization Data: A 15‐PGDH Case Study. ChemMedChem, 21(20). https://doi.org/10.1002/cmdc.70531

Vancouver

Rich AS, Dodda LS, Price DJ. Evaluating Machine Learning Affinity Prediction Methods on Novel Lead Optimization Data: A 15‐PGDH Case Study. ChemMedChem. 2026;21(20). doi:10.1002/cmdc.70531

BibTeX
@article{Rich2026,
  author = {Rich, Alexander S and Dodda, Leela S and Price, Daniel J},
  title = {Evaluating Machine Learning Affinity Prediction Methods on Novel Lead Optimization Data: A 15‐PGDH Case Study},
  journal = {ChemMedChem},
  year = {2026},
  volume = {21},
  number = {20},
  doi = {10.1002/cmdc.70531}
}
Kostenlos starten →