A Quantum Accelerator for Drug Design ⚡ экспресс
In drug discovery, accurately estimating the binding energy between a protein and a candidate molecule is crucial, but precise methods like FEP (free energy perturbation) demand massive computational resources. The study presents a hybrid quantum-classical approach that combines quantum-mechanical charge calculations, hybrid QM/MM modeling, and the quantum algorithm VQE (variational quantum eigensolver) for energy correction. Across 543 molecules and 23 proteins, an average error of about 1.10 kcal/mol was achieved, with accuracy comparable to the best classical protocols, yet computing one molecule takes roughly 25 minutes — a 20-fold speedup.
Drug design is like picking a lock: the drug molecule must fit precisely into the target protein. But the lock is submerged in water, and the fluid’s friction alters the fit. Classical calculations account for this by modeling the key dissolving into the solvent for hours per attempt.
The hybrid approach speeds things up 20-fold: a classical computer does the heavy lifting, while a quantum one refines how carbon atoms in the key exchange electrons with the lock. This dance of charges was predicted by Schrödinger and Heisenberg, and modern spectroscopy has made it measurable. Now, one calculation takes 25 minutes instead of 8 hours.
A quantum computer sees not just the key’s shape but its fuzzy electron cloud, which gently deforms inside the lock. This allows it to more accurately account for entropy — the measure of disorder that always disrupts a perfect fit. Feynman dreamed of such simulations, and this method is a step toward his vision. Paired with AI, it will enable rapid screening of thousands of candidates.
🎯 A quantum computer considers not only the positions of atoms, but also how their electron clouds are smeared out — as if we knew not just the shape of the key, but also how it softly deforms inside the lock.