Dear all,
I am trying to (blind) dock a hydrocarbon to a protein using autodock4.2 (GA runs = 500). because of the flexibility the ligand is not showing any cluster at rmsd= 2.0. I want to know how to score the best cluster? Even if I guess the cluster (or the active site) by any other method, is there any reliable method to re-score for getting the correct pose?
PS: I have already used DINC (http://dinc.kavrakilab.org)
Dear Devashish,
I have been using Autodock for several years... My experience is that it performs reasonably well for drug-like small molecules. Up to 10 bonds is quite OK if you increase a bit the default settings (number of energy evaluations, of g.a. runs,...). Molecules with 10-15 bonds need large parameter increases if you want the program to provide a useful solution within a reasonable number of g.a. runs (I usually limit to 30). I am always using a well-defined binding site.
I'm really doubtful regarding results obtained with molecules containing 25 rotatable bonds or more. The reason for this is that the number of possible conformations of a molecule increases exponentially (and not linearly) with the number of rotatable bonds. Assuming 3 local energy minima for each rotatable bond (as with a linear alkane such as butane), your carbohydrate would be around 325 conformations. Even if not all of those are possible (steric clash,...) the number of possibilities remains huge and increasing your number of g.a. runs will probably not help...
If you really want to attempt this, then you can try to follow Ahmed's suggestion and reduce the number of rotatable bonds. Once you have a crude model, maybe you can try to refine it with energy minimization and molecular dynamics.
I have a feeling that if you try to dock something very long and flexible to a surface of a biomolecule you should take note of many poses. The minima you get will be closely spaced by energy. In some ways I would compare such large and flexible ligand to a "liquid". I personally managed to dock a 100 dihedral molecule (or rather, part of a polymer) onto a surface of DNA, and as a result got very interesting trends, which repeated from one pose to another, even though the poses themselves were very different RMSD-wise.
http://autodock.scripps.edu/faqs-help/how-to/how-to-dock-a-ligand-with-many-torsions
How-to dock a ligand with many torsions
Two types of special treatment for systems with many torsions are described which may enable the user to get reproducible results when docking molecules with many torsional degrees of freedom.
For AutoDock4 systems with more than 10-12 torsions require special treatment. AutoDock Vina is successful with systems up to about 20 or so torsions. Using AutoDock Vina to dock systems with more torsions may require special treatment, too.
Two approaches to try are:
- break the molecule to be docked into several pieces, and dock each one separately. Then, look for docked conformations that would be consistent with the entire molecule. This method was used many years ago to dock two proteins using AutoDock.
- dock a small portion from one end of the molecule. Then, extend the molecule, holding the end rigid in the conformation found in the first step, and allowing the new extension to move. Continue this, adding pieces to extend the molecule until the entire molecule is docked. This method was used successfully for a study of peptides bound to antibodies.
@VarunJaiswal Thanks for the answer.
We tried a different approach back then to solve it.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5651220/
I came across a ambiguity docking heterocyclic aromatic compound with no branching. How to dock a rigid ligand showing no rotatable bonds? Please suggest.
Parmita Chawley
Hello!
1. You can search first for all the PDB structures (assuming you are trying to dock the compound with a protein) with similar compounds.
2. Align with your protein to find a suitable active/binding site
3. Try to place the optimized ligand (using Gaussian or any QM software) at the active site.
4. After placing the ligand, you need to run MD simulation and score the structure by MM-PB/SA and MM-GB/SA.
5. Change the conformation/take another PDB and repeat from step 1.
6. Choose the structure with the lowest Double Delta G value as result.
Hope it works!
Firstly, if we speak of flexibility, we speak of the number of rotational bonds.
Flexibility poses a big problem in molecular docking calculations and for that there are three types of molecular docking:
Target-compound:
1-Rigid-Rigid (i.e. Target: Rigid, Compound: Rigid)
2-Semi-flexible (i.e. Target: Rigid, compound: flexible)
3-Flexible-Flexible (i.e. Target: flexible, compound: flexible)
The majority in this case uses the second type but even this type finds a problem of compound flexibility (number of rotational bonds).
If the number of rotational bonds is high in the compound we cannot find the best pose (target complex-most stable compound) because it will use a search algorithm to test all possible orientations of the compound in the active site of the target and it will take a lot of time as well as it will lead to an inaccurate result.
Note:
MOE software can do both types of molecular docking:
Rigid-Flexible
semi-Flexible (Induced fit option in MOE software).
Induced fit option : All active site residues of the target are considered flexible.
Best regards
