I was trying redock a ligand to a protein, which was already present in the PDB crystal structure. I extracted the ligand and tried to dock it again using Autodock vina in that same site specified by a grid box centered on the original position of the ligand and of size 30x30x30. The original docking mode in the crystal structure had 4 H-bonds, but the vina docking result has only one, and that too in a different position, about 10 A from the original position. I am trying it with different gridbox dimensions, with different exhaustiveness values (8, 32), but every time it is getting docked at that wrong position. It should also be noted that the docking results themselves are very consistent among themselves. (Refer to the attached image. Blue: Original position with 4 H bonds, Pink: Vina docked position with only 1 H bond)
Why is this happening? How do I get the correct docked structure? Is this a problem with vina itself, that it is not being able to find the correct docked position? If so, then is there any better tool for docking?
Could you please furnish the PDB code for your protein-ligand complex? If so, I could try the docking to see if I can successfully redock the ligand. Thank you.
The PDB code is 6OAC, which has the ligand M1J co-crystallised with it. I was trying to dock that same ligand into it using Vina. I have also attached the pdb file with the missing residues fixed.
Arkajyoti Banerjee
Using YASARA-Structure 23.09.29 as the graphical front-end and AutoDock Vina 1.2.5 as the docking engine, I did a quick redocking of PDB ID 6OAC that I downloaded from the PDB. I did not repair missing loops because my simulation cell did not include the space that would have been occupied by the missing residues. However, I did use the YASARA "clean" function, which included adding all hydrogens and optimizing the hydrogen-bonding network to prepare the protein and ligand for redocking. I extracted the ligand and saved it as an SDF file. For redocking, I kept both the protein and the ligand rigid.
I have attached PNG image files showing just the ligand and the hydrogen-bonding residues from the protein. 6oac_xtal shows the ligand in the crystal structure, 6oac_dock shows the redocked ligand, and 6oac_super shows the crystal structure and redocked structure after a 3D alignment of the proteins using Sheba. Protein residues are gray, the docked ligand is green, and the ligand in the crystal structure is magenta. Only one hydrogen bond was detected in the crystal structure (with VAL851) and two hydrogen bonds were detected in the redocked structure (with ASP810 and VAL851). Note that the detection of hydrogen bonds depends on the the bond angles and distances used by the algorithms in different software programs to define hydrogen bonds. Moreover, hydrogen bonds are often not the dominant contributors to the overall docking score.
As you can see from the images, the redocked ligand was nearly congruent with the ligand in the crystal structure.
Although this rigid redocking appears to have been quite successful, the resolution of the X-ray crystal structure was only 3.15 A, which would be considered too poor to be used for docking. It is preferable to find structures with resolutions less than or equal to 2.5 A.
I hope that this has been helpful.
So what exactly were the parameters that you specified for the Vina docking? I mean the grid box size and coordinates, search space, exhaustiveness etc.?
Arkajyoti Banerjee
YASARA uses a customized version of Vina (and also of classical AutoDock), so it does not include some of the same user-selectable parameters that the parent version of Vina uses. To create my simulation cell, I selected the ligand in the crystal structure and set the dimensions of a cuboid to extend 5.0 A around the selected atoms and centered on the geometric center of the ligand. The dimensions (x,y,z) were 19.57, 20.14, and 20.81 A. The number of runs was set to 25. This resulted in 8 poses that ranged from -6.408 to -9.613 kcal/mol (apparent free energy of binding). The figures that I posted for you were for the lowest-energy pose (-9.613 kcal/mol). I have attached a PDB file of the crystal structure aligned with the top-scoring pose.
Alright. Will check it with this method. Thanks a lot. Is this YASARA open source?
Also, I did it now with Autodock4, and got the correct binding.
Arkajyoti Banerjee
My version of YASARA (YASARA-Structure) requires a paid license. However, the price is very reasonable compared to other comparable software such as Schrodinger or CCG MOE. In addition, the licensing policies are extremely liberal. For example, I can run as many simultaneous instances as I wish, on the same computer and/or on multiple computers with the same or different operating systems (it runs on Linux, Mac, Windows, and Android).
There is also a free "viewer" version of YASARA. I have not tried the free version, and I do not know its capabiliites, but I suspect that it would not include docking.
I am glad to know that you got good results with classical AutoDock 4. Sometimes, it works out that AutoDock 4 gives better results than Vina. It depends in part on the particular receptor-ligand system. Standard AutoDock 4 is slower than Vina, although now there is an accelerated version, AutoDock-GPU, than runs on OpenCL or CUDA to achieve very fast speeds.
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Atomic, Molecular and Optical Physics
- Atoms