If I don't have any prior structural studies on my protein-ligand complex, how can I ensure that my dockings are fool-proof?
Hello Shalinee,
If the co-crystallized structure of protein is available in protein data bank (PDB), then you can extract the ligand and then performed the docking studies to validate your docking protocol. Once the docking is done you can select the best pose based on binding energy, ligand-receptor interactions and the active site residues. Then simply align both docked pose with that of co-crystallized structure and calculate the RMSD; lower value signifies that you are on the right track.
It doesn't matter whether prior structural studies are available or not. If you validate your docking protocol properly then there are few chances of distraction. Further you can performed docking studies of your interested substrates or inhibitors.
If anything more need please let me know.
Best luck..
The problem is that co-crystallized structures are available for only the catalytic domain of the protein, whereas I am carrying out docking studies with a full-length one. We are moving on with the assumption that our ligand of choice binds to some allosteric site. I am getting somewhat expected results through docking but the point is how can I be absolutely sure of the results. Is there any Molecular Dynamics studies that I can do to back up my docking data.
If you thins that the ligand bind to the allosteric site rather than binding to the active site and you got some expected results from docking studies then to prove this it it is always better to go for Molecular Dynamics to know which amino acid residue is contributing more for stabilization of your ligand. You will come to know whether the residue is located in active site or allosteric site.
I believe MD results will help you to prove your results.
Dear Shalinee,
RMS fluctuations result from MD simulations would give you the idea about the residues which are contributing in stabilizing the protein-ligand complex. You can also do fluorescence studies on your protein and ligand(if protein and the ligand are available) to get a particular binding energy and affinity and then you can compare them with your docking results.
Best Regards,
Dear Shalinee,
Unfortunately there is not much you can do other than to do experiments. Don't run, cause it will take a long time to run and it will be very difficult to convince reviewers that the MD results are correct. MD force-fields are particularly bad at handling small molecules. Computationally, you can do the following:
1. First redo the docking only with the catalytic domain and check how well the method you are using recuperates the information of the crystal structure. For example whats the RMSD of the ligand, is the docking pose same, accuracy and coverage of predicted contacts etc. This will serve as a benchmark for your docking method of choice. If this fails, its obvious you cannot trust the results of the docking method you are using.
2. Compare your docked full-length structure with the co-crystallized structure. The catalytic domain should not be affected too much by the rest of protein. Compare in terms of RMSD, docking pose, and accuracy and coverage of contacts. The values should be close to what you get from 1.
3. Use methods like X-score http://sw16.im.med.umich.edu/software/xtool/ to calculate binding affinity of your docked structure and compare with experimental binding affinity. You may consider using multiple tools like X-score and use consensus/average etc.
Thank you all..let me try whatever you all have have suggested. I will get back to you if I have any problem.
Docking without any experimental tests is not foolproof. You solutions may contain the correct answer but you will certainly get false positives – this is a major problem with computational methods in general. It is important to do some experimental tests. Mutatgenesis of the implicated binding site and examine the impact on binding affinity with controls that bind another site would be one test.
while MD simulation is the best model for confirming the computational prediction ( which still needs experimental validation), you can dock other compounds known to be capable of binding to such binding site (Actives) and another set of inactives (decoys) to determine if the software is able to differentiate between actives and inactives.
you could use this site to download the sets of active and in actives
http://dude.docking.org/
Get about 6 inhibitors of the target you used and generate their decoys at Database of Useful Decoys, Enhanced( DUD:E). Afterwards dock the inhibitors and their decoys and plot an ROC curve from the results. If you get an AUC greater than or equal to 7, it means the docking protocol you used is good as it can distinguish between inhibitors and decoys.
You may also try LigAlign.
- Badges
- Science topic
- Similar topics
- Correction