[Dock-fans] Failed Docking even after relaxing params

#AMR ALI MOKHTAR ALHOSSARY# AMRA0001 at e.ntu.edu.sg
Tue Nov 4 04:52:07 PST 2014


Thank you both Sudipto and Trent!
Cancelling the bump filter alone did not improve it, but after changing the 2 threshold parameters I could get suitable poses for the four complexes.

There was no need for splitting the pocket into subpockets.

Thank you so much.

Amr



From: Sudipto Mukherjee [mailto:sudmukh at yahoo.com] 
Sent: Tuesday, November 04, 2014 1:08 AM
To: Trent E. Balius; #AMR ALI MOKHTAR ALHOSSARY#; dock-fans at docking.org
Subject: Re: [Dock-fans] Failed Docking even after relaxing params

Hi Amr,

To add to what Trent said:

It sounds like you are docking to multiple potential binding sites on the same protein. DOCK isn't suited for blind docking or pocket search out of the box. The anchor and grow algorithm is best suited for predicting the most favorable pose in a single site. 

However, if you wish to go ahead, I would recommend running a separate DOCK calculation for each potential pocket. Then use your own script to compile the results and pick the "best" score pose over all the runs. As a starting point, you could use the sphere clusters predicted by the sphgen program. However, for blind docking, you may get better results by using one of the many machine learning oriented ligand binding site prediction tools available.

Sudipto

On Monday, November 3, 2014 11:50 AM, Trent E. Balius <tbalius at aol.com> wrote:

Hi Amr,

To add to Sudipto's suggestions, you might also consider increasing the following parameters: 

from 
pruning_clustering_cutoff                                    25
pruning_conformer_score_cutoff                               25.0

to 

pruning_clustering_cutoff                                    100
pruning_conformer_score_cutoff                               100.0

This will increase sampling but might slow down your calculations.


http://mailman.docking.org/pipermail/dock-fans/2012-April/002650.html

Since you are docking to meny sites you may need to increase pruning_clustering_cutoff even more. 
See the manual:
http://dock.compbio.ucsf.edu/DOCK_6/dock6_manual.htm#PruningtheConformationSearchTree
http://dock.compbio.ucsf.edu/DOCK_6/dock6_manual.htm#LigandFlexibility


Typically, if you want to dock to multiple sites we would suggest that you perform multiple docking runs for each site.  

DOCK is not well suited to docking to multiple sites in the same run.  

I hope that this helps,

Trent
======================================
Trent E. Balius, PhD
Postdoc, Shoichet Lab, 
Dept. Pharm. Chem., UCSF
Phone: 415-514-4289
URL: http://docking.org/~tbalius
======================================


-----Original Message-----
From: #AMR ALI MOKHTAR ALHOSSARY# <AMRA0001 at e.ntu.edu.sg>
To: Sudipto Mukherjee <sudmukh at yahoo.com>; dock-fans <dock-fans at docking.org>
Sent: Mon, Nov 3, 2014 7:28 am
Subject: Re: [Dock-fans] Failed Docking even after relaxing params
Hi sudipto,
 
Thank you for your reply.
Yes, there is a reason for that. I am comparing the performance of Dock to the performance of another tool that I am developing within the same search space without any sort of preference (i.e. even if it end up docking in another pocket within the same search space), for the whole test set.
 
I will try re-dock without the bump filter and feed you back.
 
Thanks
 
Amr
 
From: Sudipto Mukherjee [mailto:sudmukh at yahoo.com] 
Sent: Monday, November 03, 2014 11:20 PM
To: #AMR ALI MOKHTAR ALHOSSARY#; dock-fans at docking.org
Subject: Re: [Dock-fans] Failed Docking even after relaxing params
 
Hi Amr,
 
Try redocking those four complexes without the bump filter. From the verbose dock output, it looks like a lot of your partially grown conformers were pruned by the bump filter. 
 
We typically use < 100 spheres in the binding site. Is there a reason why you are using so many spheres? Despite starting with so many spheres and consequently a very large number of cliques during anchor orienting, the flexible docking calculation stage starts with only 23,10 and 9 anchor orients for each anchor after clustering. 
 
Regards
Sudipto Mukherjee, PhD
Dept. of Chemistry, Temple University
1901 N 13th St, Philadelphia, PA 19122
 
On Monday, November 3, 2014 5:26 AM, #AMR ALI MOKHTAR ALHOSSARY# <AMRA0001 at e.ntu.edu.sg> wrote:
 
Hi
While docking PDBbind 2014 core collection, four of the receptor-ligand complexes failed to find a suitable conformation, using the same parameter value of the application tutorial.
 
Even after relaxing the parameters: min_anchor_size to 6, max_orientations to 50000, pruning_max_orients to 1000, pruning_clustering_cutoff to 25; four complexes fail to find any suitable conformations. They are namely 3PWW, 1NVQ, 2PQ9, 2YGE.
 
May I have any recommendations?
I am attaching the output of one of them for your reference.
 
Amr
 
 
Best Regards, 
Amr AL-HOSSARY 
 

Amr Ali AL-HOSSARY |   Ph.D. Student | International PhD program in Computational Biology and Bioinformatics | School of Computer Engineering |   Nanyang Technological University | Singapore (GMT+8) |  Email:  aalhossary at pmail.ntu.edu.sg |  Mobile: +65-94572816 
 
 
 

_______________________________________________
Dock-fans mailing list
Dock-fans at docking.org
http://mailman.docking.org/mailman/listinfo/dock-fans
_______________________________________________
Dock-fans mailing list
Dock-fans at docking.org
http://mailman.docking.org/mailman/listinfo/dock-fans



More information about the Dock-fans mailing list