For virtual screening, what area of the receptor should be selected as the search box for the binding site, if the area around the bound ligand in the pdb is far away from the catalytic triad in the receptor? I ran virtual screening in GOLD of pdb 3U1I, for which the bound ligand is the peptide of (BEZ)(NLE)KR(OAR), against a PPI library, and I had chosen my active site to be all the atoms within 12A of a particular oxygen atom in SER 135, since the catalytic triad in this protein is SER135, HIS51, and ASP75, even though the site the peptide ligand was bound to doesn't completely overlap with the active site area I selected. I also re-docked the peptide ligand during the virtual screening as well to make sure it still binded similarly, but I noticed that the predicted binding poses of the re-docked peptide ligand were all kind of different from the original peptide ligand in the pdb file.
I'm a little worried about the validity of my docking results and unsure if the highest ranked docked compounds from the PPI library from the VS I ran and their predicted binding poses are likely since I don't really understand if I choose a good active site. Should I re-run the VS and increase the search box to 15A or 20A around the catalytic triad so that the binding pose of the re-docked peptide ligand will match the actual pose of the peptide ligand in the pdb? But wouldn't that give more false positives in my docking results?
I thought it was important to keep the search box as small as possible, so I was actually thinking maybe my results would be more accurate if I made the active site to be 10A around the catalytic triad, but I'm not sure if my reasoning makes sense?
I think I read somewhere that it also depends on what types of ligands are involved, right? Because I think the Protein-Protein Interaction library has mostly small molecules, so I guess that's very different from the peptide ligand in the pdb 3U1I, right? Is that because peptide ligands are bigger than small molecules or are there other differences to be aware of how peptides bind compared to how small molecules bind to a receptor? What other types of ligands are there that I would need to consider in virtual screening, and how would it change how I set up my VS run? I've been reading so many papers on virtual screening, etc, but none of the papers I've read have discussed this kind of situation, so I'm really confused on if what I'm doing will lead to likely potential compounds that I could test in the wet lab for binding.
I'm very new to virtual screening and am still learning about chemistry and biochemistry, so I was just hoping someone could explain how to decide what to choose as the active site during VS in general, especially when the ligand from the pdb isn't bound near the expected catalytic triad area?
Hi,
If you do not have structural information on the binding site, you can define the grid around the active site by searching in the literature for ligands that have an affinity for the active site with information on important residues in molecular recognition. The above will allow you to do your docking calculations and validate your docking method.
This is totally up to you. If you would like your docking-based hits to have the same effect as the co-crystallized ligand (e.g., allosteric inhibitor), then use it as a reference for the binding site. If you wish docking hits to be substrate-competitive inhibitors, then use the enzyme's catalytic site to generate the docking grid.
Run with both the distances of 10, and 15A, AND use the suggestions provided by Professors Dmitri and Bello. The docking and orientation may not be similar and may require other modeling input for you to proceed with wet chemistry screening.
- Badges
- Science topic