Can your model actually predict the outcome of a GPCR experiment?

“Explain” is no longer enough. Can your model actually predict the outcome of a GPCR experiment?

At Uppsala University, Dr. Jens Carlsson and his team are redefining what computational modeling means in drug discovery. Their lab doesn’t just simulate receptor-ligand interactions after the fact; they aim to forecast receptor behavior before the first assay is run.

By integrating molecular docking, molecular dynamics, and machine learning, they design ligands with the goal to anticipate biological outcomes. This kind of predictive modeling challenges the traditional role of computation in pharmacology, where models have too often served as post hoc rationalizations.

But Carlsson’s lab stands out for another reason: knowing when not to predict. His team is candid about the limits of their models. If the resolution isn't good enough, or if the data is too uncertain, they’re not afraid to say, “We don’t know.”

That scientific humility (combined with deep collaborations with medicinal chemists and pharmacologists) is exactly what makes their predictions so useful.

This episode is essential listening for anyone thinking seriously about translational pharmacology and the future of GPCR drug discovery.

🎧 Explore how predictive modeling is reshaping GPCR science in this Dr. GPCR Podcast episode: model predict discover  

#DrGPCR #GPCR #MolecularModeling #PredictivePharmacology #DrugDiscovery