Weird protein split in flexible docking

[farmaceut]

Hello,

I want to perform flexible docking. However, when I run vina, the job doesn't proceed. No error is raised. I checked the rigid and flexible parts of the receptor, and it seems like something is wrong. Some residues are missing.

Here is the receptor 4lxd.pdb, and these are the outputs of prepare_flexreceptor.py:

4lxd_flex.pdbqt
4lxd_rigid.pdbqt

For the flexible part, I specified amino acids using -s ALA97_ASP100_PHE101_SER102_TYR105_ASP108_PHE109_MET112_VAL130_GLU133_LEU134_ASN140_TRP141_GLY142_ARG143_VAL145_ALA146_GLU149_PHE150_VAL153_PHE195_TYR199. They were chosen based on a distance of up to 5Å from the known ligand.

What could be wrong?

[diogomart]

For people to be able to help you, you need to provide more detailed information:

1. exact vina command and all input files to reproduce the problem. You posted only part of what's needed.
2. "the job doesn't proceed" can mean a number of things. Does it print something?
3. "some residues are missing". What residues are missing?

[farmaceut]

@diogomart
Command lines:

prepare_flexreceptor.py -r 4lxd.pdbqt -s ALA97_ASP100_PHE101_SER102_TYR105_ASP108_PHE109_MET112_VAL130_GLU133_LEU134_ASN140_TRP141_GLY142_ARG143_VAL145_ALA146_GLU149_PHE150_VAL153_PHE195_TYR199
prepare_gpf.py -l H4Fst.pdbqt -r 4lxd_rigid.pdbqt
autogrid4 -p 4lxd_rigid.gpf -l 4lxd_rigid.glg
vina --flex 4lxd_flex.pdbqt --ligand H4Fst.pdbqt --maps 4lxd_rigid --scoring ad4 --exhaustiveness 6 --out output.pdbqt

The output:

#################################################################
# If you used AutoDock Vina in your work, please cite:          #
#                                                               #
# O. Trott, A. J. Olson,                                        #
# AutoDock Vina: improving the speed and accuracy of docking    #
# with a new scoring function, efficient optimization and       #
# multithreading, Journal of Computational Chemistry 31 (2010)  #
# 455-461                                                       #
#                                                               #
# DOI 10.1002/jcc.21334                                         #
#                                                               #
# Please see http://vina.scripps.edu for more information.      #
#################################################################

Scoring function : ad4
Flex receptor: 4lxd_flex.pdbqt
Ligand: H4Fst.pdbqt
Exhaustiveness: 6
CPU: 0
Verbosity: 1

Reading AD4.2 maps ... done.
Performing docking (random seed: 165862369) ... 
0%   10   20   30   40   50   60   70   80   90   100%
|----|----|----|----|----|----|----|----|----|----|
*

It's just stuck at *. It doesn't change even though I waited an hour.

The files are in archive:
archive.zip

I use Chimera. Opening simulatenously 4lxd_flex.pdbqt with 4lxd_rigid.pdbqt you can see there are gaps between residues which are filled by AAs in case of original 4lxd.pdbqt file.

[diogomart]

Then it's running. But with that many sidechains, it just takes extremely long.

Molecular viewers show gaps because the protein is split between different files, but all atoms should be there. Each atom should be in either _flex or _rigid, not both.

[rwxayheee]

Hi @farmaceut

The inputs look correct. I was able to run the calculation, too; The job does proceed very slowly as @diogomart pointed out. If you're not running the job on a cluster, I recommend starting with fewer flexible residues :>

also because your ligand is much smaller than the known ligand in 4lxd. It can only occupy a subsite of this binding site, so having fewer flexible residues might be worth a try. Maybe you could run some jobs with different selection of flex residues and then compare.. or summarize them?

About the missing residues: Gly, Pro and Ala cannot be made flexible, because their sidechains do not have rotatable torsions (for Ala, the nonpolar hydrogens are excluded from PDBQT)

[farmaceut]

@diogomart @rwxayheee
Oh well. I though, it was me doing something wrong. At least I am aware my old PC is not enough :( Wonder if running on M3 would make it notably faster?

[diogomart]

Possibly. But not as fast as having fewer sidechains.

[rwxayheee]

Hi @farmaceut Forgive me for interjecting again, one more point to consider with your selection of flexible residues is the size of the box. If the box does not cover all the mobile atoms in the flexible residues (for example, Asp100, Glu133, Tyr199), you will get enormously large INTRA energies in the docking output (and those residues will appear to be strained, in the output. So please make sure to check if the size of the box is sufficient to contain residues' sidechains, preferably with some extra room for alternate sidechain conformations.