Hello
I am trying to find out the exact binding sites of a number of peptides against a specific protein target. I did docking using many different servers and obtained many divergent results. How can I assess the validity of these results and choose the best among them?
If the protein structure was retrieved from the PDB and it is a has a ligand bound to it, you can align the ligands of the co-crystallized complex and the redocked complex (the peptides). You can then validate each docking tool to identify the best performing one based on the ligand alignment results. Remember the accepted RMSD is 2 angstrom or less.
If the structure has no ligand or it was modelled, predict the binding pocket using servers like CASTp, PDBsum, etc. After that do exhaustive search in literature do identify previous works done on the protein and the binding sites the used to support the predicted binding pocket.
Lastly, you can do a surround the protein with the grid box and Dock again to see the binding sites most of the ligands dock in to.
I hope this will be of help.
Michael above gave a good answet on how you can convince someone (e.g. the reviewers) that you did computationally evertyhing that could be done.
The true validity can only be assessed by experimental methods :)
- Badges
- Science method