[Dock-fans] Fw: Re: Fw: Re: Fw: Re: Fwd: Re: Ligand Docking with DOCK6
sbrozell at scripps.edu
Wed Apr 23 13:31:48 PDT 2008
On Wed, 23 Apr 2008, Francesco Pietra wrote:
> --- On Tue, 4/22/08, Scott Brozell <sbrozell at scripps.edu> wrote:
> > There are several issues here. First a clarification:
> > receptor_site_file
> > and
> > grid_score_grid_prefix
> > are distinct.
> > In all docking calculations (i.e., orient_ligand == yes)
> > receptor_site_file defines the active site(s).
> > grid_score_grid_prefix represents a special region of the
> > receptor;
> > special in that contact or energy scores can be computed
> > very quickly
> > for ligands that dock such that the whole ligand is in the
> > region.
> > Grids are not used with continuous score;
> > grid_score_grid_prefix is not a Continuous Energy Score
> > Parameter;
> > with continuous score cont_score_rec_filename is the
> > receptor.
> OK. Sorry for having posted mixed continuous/grid scoring. However, in the computations, the two were treated separately, and in accordance with what you write above.
> I have now carried out continuous docking serial runs (same *.in file as before) for the tree, same-family compounds already dealt with on present thread, and for which I have carried out biochemical experiments as to their interaction with the protein. Computations for two situations of
> sphere_select -i rec.sph -c protein_noH.pdb -a 3421 -r radius_in_A
> where 3421 is the strategic atom number where the sphere was centered. I use two values of the radius, 23A and 25A. In both cases the spheres (568 for 23A and 685 for 25A) represent satisfactorily the protein. The 568 spheres situation only leaves out protein loops that may safely be considered irrelevant to the docking.
> For the 5-rings and 8-rings compounds all (separately carried out before) grid docking simulations OK, followed by several ns MD with Amber9 after that the complex was embedded in a lipidic membrane. For the 13-rings compound we are here trying to solve the remaining issues.
> RUNS WITH 568 SPHERES:
> 5-rings-compound 5.7%mem usage
> 8-rings-compound 10.7%mem usage
> 13-rings-compound 27.9%mem usage
> RUNS WITH 685 SPHERES
> 5-rings compound 8.3%mem usage
> 8-rings compound 15%mem usage
> 13-rings compound what () std.::bad_alloc
> %mem usage refers to the available 23GB for the computation (NUMA-type MACHINE with 8 logical processors). Although no one of these continuous-docking runs was brought to completion, I extracted the above data when %mem usage was no more increasing.
So 5 and 8 rings ligands ran with grid score.
13 rings ligands with 568 spheres are running with continuous score.
13 rings ligands with 685 spheres won't run with continuous score ?
> It seems to me (correct, please, if I am wrong) % memory usage should be considered for "size" of sphere_select.sph AND "size" of the ligand.
Yes. I assumed that the spheres would dominant, but 13 rings
makes for a huge (and atypical) ligand.
> > Fourth maybe your ligand(s) do not DOCK to your receptor.
> > Adding rings may produce a ligand that cannot fit into a
> > pocket
> > that was a previous active site. This is research ...
> > DOCK is only parallelized over ligands. With one ligand in
> > ligand_atom_file running in parallel is the same as running
> > serially.
> > Start with serial DOCK on a small representative set of
> > ligands.
> > Once you have useful serial
> > calculation then run a library of ligands in parallel.
> > http://dock.compbio.ucsf.edu/DOCK_6/dock6_manual.htm#ParallelProcessing
> OK, I have badly overlooked the point. I expected that there was parallelization with respect to the receptor, too. I was mislead by amber (re)scoring, where all CPUs are at work (at least as observed with the 4-CPUs previous machine, not yet tried with the 8-CPUs machine). I understand that parallelization is the hardest codification in these programs.
I have added some clarifications and updated the parallel processing
section of the manual.
More information about the Dock-fans