The day in conversation for 2026-11-18

discussion

Transcript

The discussion: Yuki: The STORM model’s structure-free binding affinity prediction seems promising, but I wonder what specific limitations are being addressed regarding conformational sampling when you bypass explicit structural information entirely. Ines: What interests me is how this moves beyond traditional docking scores; if it truly captures the underlying physical constraints without relying on a template, it changes the baseline for what we consider computationally tractable in this space. Have you looked at how the error rate scales when moving from simple protein-ligand pairs to larger multi-domain systems? Marcus: I'm more focused on that scaling issue. While structure-free approaches are elegant, they often struggle with the complexity of solvent effects or induced fit mechanisms that aren't purely energetic. Are we assuming a purely enthalpic or entropic balance in the prediction, and how robust is that assumption across different classes of ligands? Yuki: That brings up my concern about those non-globular proteins we discussed last week; if the underlying energy landscape itself is highly anisotropic

More episodes

← Home