I am working on a range of molecules for my PhD project, and I have done all the docking and MM-GBSA calculations using the Schrodinger suite of applications (Glide and Prime). But now I'm stuck - I don't understand which one of the sets of values I should be looking at in order to predict which compound is better than another. Should I be using the top docking scores, or the top MM-GBSA scores, which do not always correspond to the top Glide scores? I do not yet have any biological data for any of the compounds.(The ligands are basically congeneric, and the same protein and settings were used for all of the calculations)
Hi Cathy,
See comparing the scores between docking score adn MM-GBSA scores will be very difficult. To solve this problem you can find the natural ligand for the receptor and perform the Docking as well as MM-GBSA studies. The scores for the natural ligands are the standard scores from which you compare your two different results. By this way you will ensure the standard for the two methods. Then you can apply what Elvis has suggested to find out the best poses hence the best structures.
Cheers!!!
Dear Cathy,
I would correct the docking for indirect solvent interactions using MMGBSA and look at the scores. Typically, a convergence of Glide and MMGBSA scores means that you may look more closely at the compounds on the top. Typically, I would drop anything above -50 in MMGBSA. Also, ask them for Desmond (free for academia) and try to do MD on the compounds. If they stay on after 20ns, I would consider them as potential hits.
The other part you should look at is the RMSD of Ca backbones (protein, ligand and protein around ligand), number of contacts and the major contacts. The more contacts you have, the better for you. If your IT guys can arrange for a good server, you may try 50-100ns MD with Desmond. At this stage, I would seriously think about using full Schroedinger Suite to build an advanced model based on energy, correct it and dock your poses into it using GlideXP.
It is my sad experience that people think about the software as a plugin to get them a quick fix. For your sake, spend more time learning the software, watch videos, get to know people using it and then try doing it. If you do it correctly, a simple HTVS may get you low uM inhibitors. Be prepared to have 1 in 50 hit ratio when testing it experimentally.
Good luck in your research,
Wes
- Badges
- Science topic