Quantum mechanical calculations describe the electronic behavior of atoms and molecules and this is what make it suitable for the site of failure of Molecular Mechanical calculations which is the calculation of bond formation and dissociation energies, but QM methods is expensive from a computational perspective.
On the other hand Molecular Mechanical methods dominate in the molecular docking literature accounting for ligand-receptor bonding by (so to speak) fully empirical values that produce satisfactory results.
QM/MM hybrid methods are most of time used in reactivity and mechanism elucidation, rarely in docking.
Actually I'm biased to the QM methods because I'm convinced that it treats the real factor in the chemical reaction of receptor-ligand complex which is the electron, but I don't deny that I'm naive in this field and for sure there are reasons behind the use of pure MM methods in many research papers.
Are there benchmarks for judging which method is suitable and which is time- wasting or inaccurate ?
QM is more accurate than MM for docking if the equilibrium ensemble is considered.
First, remember that current implementations of QM are also MODELS (accurate, but still models) that try to approximate as good as possible experimental findings. Models are to be used, not believed. So you can be a purist believeing that QM describes the "real" factors in the close-to-real way and that's fine, but it will very much limit the scope of your research.
Also, let's say that you have a ligand binding to a protein pocket and can use a) autodock that will give you very good estimation of free enthalpy of binding in 10 minutes or b) QM DFT calculations, that will take 2-4 weeks (if you're lucky and can reduce the protein to the binding pocket alone) to come up with more less the same number. Would you reject the first method only because you're biases towards QM?
The obvious reason to use MM is the exponential scaling of QM methods, which makes the QM calculations for systems over 150/200 atoms prohibitive from the computational point of view. That problem won't be likely solved in the near future so if you want to study larger, non-periodic systems your only hope is MM.
There are tons of benchmarks for every method commonly used, look for papers by methods authors and reviews.
Thank you Dr. Philippe ..
Thank you Dr. Bartosz ..
Of course it is not wise to waste such a long period of time on QM methods while MM can give a good result in such substantially shorter period, but there are papers that show significant improvements on scoring due to the incorporation of QM methods accounting for electronic properties (e.g. charge distribution).
http://www.ncbi.nlm.nih.gov/pubmed/15841474
http://www.schrodinger.com/QM-Polarized-Ligand-Docking/
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Quantum Physics
- Quantum Mechanics