You are here

Problems with Ligand Docking

9 posts / 0 new
Last post
Problems with Ligand Docking
#1

Hi everyone,
I am attempting to run a blind dock of a ligand (chain X) into a protein complex (chains ABCDEF). However, try as I might, I seem to always encounter problems. Currently, despite the fact that the flags I'm using should perturb the ligand by a significant amount, the resulting structures are virtually identical (for all chains) to the starting structure, except all chains have been rotated.
Here are my current command line flags:

-in:path:database /ifs/scratch/saxen/rosetta_database
-in:file:s struct_rpk_C2E.pdb
-in:file:extra_res_fa /ifs/scratch/saxen/struct_ligdock/C2E.fa.params
-in:file:extra_res_cen /ifs/scratch/saxen/struct_ligdock/C2E.cen.params
-packing:unboundrot /ifs/scratch/saxen/struct_ligdock/struct_rpk_C2E.pdb
-packing:ex1
-packing:ex1aro
-packing:ex2aro
-packing:extrachi_cutoff 1
-docking:partners X_ABCDEF
-docking:spin
-docking:uniform_trans 20 # angstroms
-docking:randomize1
-docking:ligand:random_conformer
-docking:ligand:minimize_ligand
-docking:ligand:harmonic_torsions 10
-docking:ligand:minimize_backbone
-docking:ligand:harmonic_Calphas 0.3
-docking:ligand:soft_rep
-docking:ligand:old_estat
-docking:ligand:protocol abbrev2
-out:pdb
-out:mute protocols.geometry.RB_geometry core.scoring.dunbrack.SingleLigandRotamerLibrary core.scoring.rms_util
-out:nstruct 200

I've added and removed a number of these trying to get a different result and so far have not been successful. I'd really appreciate any insight anyone can offer on this!

Post Situation: 
Fri, 2011-08-19 00:19
saxen

A) Which protocol are you using? Ligand docking or regular docking? It is my impression that the ligand docking standalone executable is deprecated, and it is mostly done via the scripter now. I have brought this to the attention of the relevant developer.

B) Half of your flags are for ligand docking and half are for protein-protein docking. To my knowledge, neither standalone executable does both, you HAVE to rely on the scripter to get both motions.

C) If you want to use the ligand docking executable, I think you should remove most of the docking flags as extraneous. If you've been using the protein docking executable, that would explain your results so far...

Fri, 2011-08-19 09:10
smlewis

Thanks for the quick response!
I'm using ligand docking, following the instructions for Rosetta 3.2.1 (my installed version) at http://www.rosettacommons.org/manuals/archive/rosetta3.2.1_user_guide/ap.... This uses the standalone executable. I also based my flags file off of the one recommended here. The author of that page uses -docking and -docking:ligand flags in his ligand docking run.
Could you direct me to an updated page that would recommend how to set up a ligand docking run?

Fri, 2011-08-19 09:51
saxen

While the ligand docking application does re-use some of the -docking flags, it's a separate application from protein-protein docking, and doesn't use all of them. For example, -docking:partners is not used for ligand docking. Unless there's indications otherwise, it's best to assume that protein-protein-docking issues/options don't apply to ligand docking (except, of course, for general Rosetta issues).

The best documentation on how to set up a ligand docking run with the ligand_dock application is the page you linked. The Meiler Lab (Jens Meiler did the original Rosetta Ligand docking) also has some tutorials on Rosetta in general, which includes information on ligand docking (See http://www.meilerlab.org/index.php/jobs/resources).

For the simplest setup, my suggestion is to use the arls.py script, as mentioned in the manual page. If you don't have a Condor cluster, there's a little effort to convert the Condor submission script into a usable form, but the Condor submission script is rather straightforward, so it's relatively easy to combine the executable with the command line arguments to run as, say, a bash script.

Fri, 2011-08-19 10:57
rmoretti

Thanks! I don't have a Condor cluster here, but I'll see if I can tweak the script to work for me.

Fri, 2011-08-19 12:41
saxen

The ligand_dock executable doesn't recognize -docking:partners or -docking:spin. The rest of your options look good. Is the ligand (chain X) the LAST thing in your PDB file. This is required by the ligand_dock executable.

Fri, 2011-08-19 12:32
glemmon

Also uniform_trans 20 means you are sampling within a sphere of radius 20 Angstroms. That is a really big area to sample. You could instead use multiple "start_from" points and a smaller uniform_trans (smaller spheres of sampling around each start_from point)

Fri, 2011-08-19 12:40
glemmon

Yes, chain X is the last item in the PDB file. There is a TER between chains ABCDEF and chain X, but not between any of the other chains. When I remove the -docking:partners and -docking:spin flags and run the ligand docking, the ligand and chains B-F remain stationary while chain A is perturbed about 20 Angstroms from the start position. Any ideas why this might be?

Yes, I'm planning on tweaking that uniform_trans, once I can even get this to run correctly. Would you recommend for a blind dock of a large ligand starting the ligand inside the complex in various locations or picking a large number of points around the complex to start from?

Fri, 2011-08-19 13:48
saxen

It's hard for me to diagnose without the input files, but I always prepare my input files by making sure the ligand atom lines start with HETATM, and the residues and atoms of the ligand chain are numbered separately starting with "1". If you still have problems you can attach your input file.

Fri, 2011-08-19 14:11
glemmon