QM/MM methods
An enzyme has thousands of atoms, but the chemistry happens among a few dozen in its active site. Treating every atom quantum mechanically is unaffordable; treating every atom classically cannot describe bonds breaking. In 1976 Arieh Warshel and Michael Levitt proposed the obvious compromise: quantum mechanics where the chemistry is, classical mechanics everywhere else. It took decades to become routine, and it won the 2013 Nobel Prize in Chemistry, shared with Martin Karplus. QM/MM is also the frame into which any future quantum-computer subroutine for biochemistry will slot.
On the plate: choose how much of an enzyme to treat quantum mechanically, and see the cost.
Splitting the system
The energy is written
The MM region uses a force field: springs for bonds and angles, torsion terms, Lennard-Jones contacts and fixed partial charges, cheap enough for hundreds of thousands of atoms. In mechanical embedding the regions interact only through classical terms; in electrostatic embedding, the standard choice, the MM partial charges enter the QM Hamiltonian, so the protein’s electric field polarizes the active site, which is often the main way enzymes catalyse. Hans Martin Senn and Walter Thiel reviewed the methods in 2009. Established
How big a quantum region?
Choose a QM radius around the active site of 6.0 Å and a method: or . Proteins contain about 0.1 atoms per cubic ångström, so the QM atom count grows as the cube of the radius, and the cost as a high power of that. Results also converge slowly with QM size; studies typically need a few hundred QM atoms for converged energetics, which is why cheaper QM methods and smart embedding matter. Where the bonds cut the boundary, a hydrogen link atom caps the dangling bond of the QM region. Established
Free energies, and the quantum subroutine
Enzymes fluctuate, so a single structure is not enough. Reaction free energies need sampling: thousands of configurations along a reaction coordinate, each requiring a QM/MM energy and forces, with methods such as umbrella sampling or free-energy perturbation. That multiplies the cost again. In Part 5 this structure reappears: a quantum computer would compute the energy of a small, strongly correlated active region (an iron–sulfur cluster, say), embedded in DFT, embedded in molecular mechanics, sampled by classical dynamics. The quantum part only has to handle what nothing else can. Established for QM/MM; Speculative for the quantum subroutine.
What you now have
The QM/MM energy partition, force fields, mechanical and electrostatic embedding, link atoms, how QM region size drives cost, the need for free-energy sampling, and the place a quantum subroutine could occupy.
· · ·
Three questions to carry with you.
And one to work out. With CCSD(T) scaling as , by what factor does cost grow if the QM region doubles?
Sources: A. Warshel and M. Levitt, J. Mol. Biol. 103, 227 (1976). H. M. Senn and W. Thiel, Angew. Chem. Int. Ed. 48, 1198 (2009). Nobel Prize in Chemistry 2013 (Karplus, Levitt, Warshel). H. J. Kulik et al., J. Phys. Chem. B 120, 11381 (2016) (QM-region convergence).