The Molecular Graphics Laboratory Forum

AutoDock, AutoLigand, MGLTools, Vina, PyRx and more.
It is currently Sun Sep 23, 2018 4:12 am

All times are UTC




Post new topic Reply to topic  [ 13 posts ]  Go to page 1, 2  Next
Author Message
PostPosted: Sun Feb 10, 2013 11:21 pm 
Offline
Millimolar User
Millimolar User
User avatar

Joined: Mon Feb 04, 2013 1:55 am
Posts: 9
Hello,

I have taken a crystal structure relevant to my PhD project (1WM1), and split the structure into receptor and ligand, and generated the corresponding PDBQT files. I have tried docking the ligand (and thousands of other structures) into the crystal structure sans ligand PDBQT file using Vina, however the correct binding site is not found - Vina finds an a region close to the binding site but all poses/conformations are far enough away from the known binding site to not be of use.
I have placed a large exhaustiveness (1000) bias on Vina but this has not seemed to assist, a similar result is obtained, which is far from the known binding site.
I have tried using a small grid box parameter to effectively force vina to find it, but it's still too far away within the box. I have also tried making a large box (including allosteric binding modes) with the same result, Vina always gets close, but not the correct site.

I have tested the same docking run using Schrodinger Maestro/Glide and it predicts the binding site perfectly. I understand they are different programs with different algorithms, and no docking program is perfect, but I'd really prefer to do this project with Vina. I was wondering if anyone know if there was any other ways to 'convince' Vina to find the correct binding site, or whether this is just a case of a particular combination of software and structure that just won't work.

Cheers
Sebastian


Top
 Profile  
 
PostPosted: Wed Feb 13, 2013 2:07 pm 
Offline
Micromolar User
Micromolar User
User avatar

Joined: Fri Sep 21, 2012 3:25 am
Posts: 40
Location: Mexico
Sebastian,

Well, I ran 1wm1 the way I usually do:

1.- Check the PDB with molprobity
2.- Minimized the structure energy
3.- Process it with MGLTools
4.- Run vina with exhaustiveness 5000, in a grid thats 40 by 40 by 40.

My results show the ligand like this:

Attachment:
1wm1.png
1wm1.png [ 197.36 KiB | Viewed 12373 times ]


Vina's result is purple while yellow is the original crystal. I think is pretty much like the original. Is this not what you get? On a second review, I think that there are two water molecules in the binding site. Maybe you should add those to your files. In my opinion, their absence might be biasing your results. Try that out and let us know what happened.


Best regards.


Top
 Profile  
 
PostPosted: Thu Feb 14, 2013 5:17 am 
Offline
Millimolar User
Millimolar User
User avatar

Joined: Mon Feb 04, 2013 1:55 am
Posts: 9
Hi LK,

Thankyou for looking into my case, this has given me a lot of hope!

I checked/downloaded 1WM1 from molprobidity after check was okay. I loaded the PDB into ADT, split the ligand and receptor. All waters were left in place in the receptor file. No additional processing was done during importation other than the automatic/default features ADT uses. Both saved as PDBQT's.

I could not perform a minimsation as I do not think this feature is a part of the ADT/MGL/Vina suite of tools. I would appreciate if you could provide me with more verbose details on the minimisation process?

Vina was run with x=24.39 y=-2.392 z=-7.082 sizes x,y,z 40 respectively (covering the vast majority of the receptor), exhaustiveness 5000 random seed 17148986.

You can see in the attached image, in pink/red, the lonesome ligand is the superimposed one from the crystal structure against the cluster of vina's results. The situation is pretty much the same as I have always experienced. I am thinking maybe this minimisation process you are using (by logical deduction) must be the key here, unless I am missing some very obvious step during importation/pre-processing.

Kind regards,
Sebastian


Attachments:
vinadockproblems.jpg
vinadockproblems.jpg [ 109.04 KiB | Viewed 12365 times ]
Top
 Profile  
 
PostPosted: Thu Feb 14, 2013 5:43 am 
Offline
Micromolar User
Micromolar User
User avatar

Joined: Fri Sep 21, 2012 3:25 am
Posts: 40
Location: Mexico
Hi Sebastian,

I think you are right. I looked around and I don't have a tutorial for the energy minimization. However, the software I use, UCSF Chimera, does. Look it up here: http://www.cgl.ucsf.edu/chimera/docs/Co ... imize.html

BTW, keeping the water molecules did not make any different in my docking setup.

Best regards.

Addendum, I tried getting the ligand from another source as a mol2 file and got the same results even with exhaustivenes 50. So, energy minimization it is. I guess I will prepare a video tutorial to cover this.


Top
 Profile  
 
PostPosted: Thu Feb 14, 2013 4:41 pm 
Offline
Micromolar User
Micromolar User
User avatar

Joined: Fri Sep 21, 2012 3:25 am
Posts: 40
Location: Mexico
Hi Sebastian,

I ran another (more extended) energy minimization and that increased the number of results finding the "right" site, even with exhaustiveness 50. It also increased the "apparent" affinity (-7.9 kcal/mol). (Green is the XRD ligand, pink and blue docking)

Attachment:
last-ligand.png
last-ligand.png [ 103.91 KiB | Viewed 12360 times ]


So, my guess is that the structure has too many clashes to allow the ligand in. Those clashes are relieved by the energy minimization. (see attached molprobity table for before)

Attachment:
Screen Shot 2013-02-14 at 11.00.28 AM.png
Screen Shot 2013-02-14 at 11.00.28 AM.png [ 70.57 KiB | Viewed 12356 times ]


And, after minimization
[attachment=0]Screen Shot 2013-02-14 at 11.04.06 AM.png[/attachme
nt]

Also, the protonation state of the nearby histidine is crucial: if charged, binding dropped to nil. Uncharged, ligand binds. I hope this behavior mimics biochemistry.

Best regards and good luck.


Attachments:
Screen Shot 2013-02-14 at 11.04.06 AM.png
Screen Shot 2013-02-14 at 11.04.06 AM.png [ 69.45 KiB | Viewed 12356 times ]
Top
 Profile  
 
PostPosted: Mon Feb 18, 2013 5:58 am 
Offline
Millimolar User
Millimolar User
User avatar

Joined: Mon Feb 04, 2013 1:55 am
Posts: 9
Hello again,

Just did a long minimisation in Chimera and got even better score in Molprobidity, but unfortunately, again, I still seem to be getting close but not close enough. You said the protonation state of the nearby HIS residue mattered and that score dropped off to zero - does this look similar to that issue? I am getting apparent affinities of around -6.9 for this binding mode... Running out of ideas.

Kind regards,
Sebastian


Attachments:
results.jpg
results.jpg [ 168.49 KiB | Viewed 12288 times ]
molprobidity_500steps_minimised_results.png
molprobidity_500steps_minimised_results.png [ 43.86 KiB | Viewed 12288 times ]
Top
 Profile  
 
PostPosted: Mon Feb 18, 2013 12:47 pm 
Offline
Micromolar User
Micromolar User
User avatar

Joined: Fri Sep 21, 2012 3:25 am
Posts: 40
Location: Mexico
Sebastian,

How many step to your minimization? I usually run 1000 steps only (steepest descent). That should be enough.


Top
 Profile  
 
PostPosted: Mon Feb 18, 2013 11:08 pm 
Offline
Millimolar User
Millimolar User
User avatar

Joined: Mon Feb 04, 2013 1:55 am
Posts: 9
LK,

I did a 500 step minimisation, for steepest decent and conjugate gradient. I've got a spare machine here and not a lot of work today so I will run one with 2000 steps and get back to you.


Top
 Profile  
 
PostPosted: Mon Feb 18, 2013 11:43 pm 
Offline
Micromolar User
Micromolar User
User avatar

Joined: Fri Sep 21, 2012 3:25 am
Posts: 40
Location: Mexico
In my experience, 1000 steepest descent is optimal. I think that using conjugate gradients is taking it too far. My two cents.


Top
 Profile  
 
PostPosted: Tue Feb 19, 2013 1:20 am 
Offline
Millimolar User
Millimolar User
User avatar

Joined: Mon Feb 04, 2013 1:55 am
Posts: 9
Ok, I will keep that in mind. Is that strictly a waste in terms of computational time, or minimising the structure too much might cause the same issues as what I'm experiencing - perhaps lowering the energy of the structure so much that ligands might have a hard time fitting (unrealistically low energy)?


Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 13 posts ]  Go to page 1, 2  Next

All times are UTC


Who is online

Users browsing this forum: No registered users and 5 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
POWERED_BY
Translated by Maël Soucaze © 2009 phpBB.fr