I used VMD to find the interaction energy between Major Histocompatibility complex-II (MHC-II) and a small peptide epitope. I found the interaction energy between these molecules are three order of magnitude lower i.e. - 1500 Kj/mol than docking energy i.e. -65 kj/mol. And it is continuously going down in case of more complex such as MHC-II with IL17A or CBS or CGL or so on. Is this results valid?
I used following formula to calculate Linear interaction energy:
ΔG(bind) = {β*(elec bound – elec unbound) – α*(vdw bound – vdw unbound)}, α, β values are developed based on training set molecule. I consider it as R2 = 1 linear.
You don't provide too many details, bu my guess is that you're calculating different things.
The interaction energy, defined as E(int) = E(complex) - (E(protein)+E(peptide)) can have very low values, as its counting multiple, favorable interactions between residues of the petide and the protein. Particularly the charged residues (LYS and ARG interacting with GLU and ASP) can contribute something like -50 kcal/mol to the iteraction energy.
The docking energy usually tries to estimate the difference in free energy of binding, which additionally takes into the account solvation and entropy. -65 kJ/mol seems like a decent estimate of a good interaction, but it can't be directly compared to interaction energy calculated by NAMD.
There are protocols to estimate binding deltaG from MD runs: in the case of NAMD see e.g. http://www.ks.uiuc.edu/Training/Tutorials/freenergy-index.html.
Dear Shamsuddin,
Docking methods use a simplified energy model, so they must be followed by a more accurate estimation, LIE (Molecular Dinamycs) can be a good choice. You will find differences due to both methods don’t calculate the same and they take into account different variables (flexibility, entropy, solvent model…). I don’t know the exact molecular model but that difference could be correct. In my point of view, please, check the LIE calculation and, perhaps, use an second method to calculate the binding energy MM/PB(GB)SA.
A good review was published by Gilson and Zhou “Calculation of Protein-Ligand Binding”. Attached you can find the link.
Best!!
Miguel
http://www.annualreviews.org/doi/abs/10.1146/annurev.biophys.36.040306.132550
Dear Bartosz,
Could you provide list of such energies in a table if available i.e. Docking and NAMD energies?
Not really, since I'm only guessing what you might have done. Also, many docking programs have their own scoring functions, which are again different from either interaction energy or delta G. They may also estimate deltaG using different equations and while they should give similar estimates of deltaG, they often differ a lot between each implementation.
For the interaction energy: as I said, a charged-charged (plus-minus) interactions contributes around -40-60 kcal/mol, while pi-pi stacking, pi-cation interactions, hydrogen bonds and hydrophobic (van der Waals) interactions contribute around -2-10 kcal/mol. The interacton energy is just the sum of such interactions of fragments and, obviously, a larger interaction surface usualyl means a better interaction energy. That's why it's not a good measure of strength of binding, due to the negligence of solvations and changes in entropy.
For the deltaG of binding: usually a strong interaction between ligand and protein means around -10 kcal/mol or less in deltaG.
Dear Miguel,
Could you give me idea to do MMPBSA using VMD/NAMD? Thank you.
- Badges
- Science topic