[Dock-fans] Unbiased docking

[Scott Brozell]

Hi,

On Sun, 21 Oct 2007, Francesco Pietra wrote:

> Based on the suggestions below, I have followed the layout and settings of the
> Dock6.1-web-page tutorials for my model pore protein (3400 heavy atoms, noH, no
> solvent added) and 150-atoms neutral, lipidic ligand (bearing all hydrogens),
> using the minimum energy conformation of the latter from Amber9 simulated
> annealing in vacuum.
> 
> sphgen failed, while sphgen_cpp built re.ms and rec.sph in a few minutes,
> 118807 surface points in 91 clusters. I was surprised to see roughly the same
> sphere distribution for clusters # 1, 2, and 91. Hope that sounds correct. I
> did not examine other clusters, and all was done very roughly at the open eye.
> 
> Grid OK in ca 1h and half.
> 
> Rigid docking completed in 5 min wall time, showing the elongated ligand (made
> of annealed rings) inside the cone, with the head out. Grid score -52.3.
> 
> Flex docking (cpu time ca 3h; anchors 1; orientations 500, conformations 115,
> grid score -69.2, vdw -67.2, es -2). The arrangement of the ligand within the
> pore was roughly the same as for rigid  docking, through the head of the ligand
> was docked differently.
> 
> I did not carry out any docking with higher-energy conformers of the ligand.


I do not notice any red flags in the above, but I do not have any
experience with pore proteins.

> This long story because it is the first time I am at docking outside tutorials,
> to hopefully get suggestions as to go on. I could either carry out amber score,
> and the relevant tutorial I have found outside the dock web page is clear
> enough. Is that what Scott had in mind when he kindly suggested "go to Amber
> after docking, and with the MD results back to Dock" or should I rather skip
> amber scoring and go directly to Amber 9 to carry out a regular MD, may be with
> the pore model inside POPC? If so, which is the shortest way to use my above
> results from DOCK to set up the Amber 9 MD? (I mean, how to save the pore
> protein plus the ligand so that Amber 9 can work; I am familiar with prmtop and
> inprc file, while I read that Amber 9 can't import mol2 files with more than
> one residue; if I save protein+ligand as pdb, how to get parameters for the
> ligand? Or, if I save the protein and ligand separately do they conserve their
> respective positions?.


My recollection is that I suggested DOCK for docking and Amber for MD.
There are a number of ways to go from docking results to MD.
If you rescore with Amber Score in DOCK then the Amber score preparation
and the DOCK outputs contain Amber readable prmtop and inpcrd files.
See
http://dock.compbio.ucsf.edu/DOCK_6/dock6_manual.htm#AMBERScore
for a description of the output files.  Also running DOCK with the
verbose flag, -v, may help to see what's happening under the hood.
(Currently the verbose output follows the normal DOCK output
when -o is used; so first try -v without -o.)

Scott

> --- Scott Brozell <sbrozell at scripps.edu> wrote:
> 
> > Hi,
> > 
> > On Fri, 28 Sep 2007, Francesco Pietra wrote:
> > 
> > > Although not yet ready with DOCK6.1 installation, inspection of tutorials
> > > prompts a couple of questions that may help planning.
> > >
> > > What I want to do at the moment is verifying if there is docking at all
> > between
> > > a family of natural products and a complicated protein. Not in a bilayer
> > for
> > > the moment.
> > >
> > > I have already carried out a conformational search for my natural products
> > by
> > > simulated annealing with an algorithm that calls Amber9. Parameter and
> > > coordinate files for Amber9 were built with Antechamber/Divcon (ie AM1).
> > >
> > > The tutorial on DOCK web site starts from a protein-ligand complex, ie from
> > a
> > > definite receptor site in the protein. Is any graphical or alphanumeric
> > > tutorial for the unbiased docking I am aimed at?
> > 
> > Not that I am aware of.
> > There is a lot of flexibility in binding sites selection;
> > see step 3 of
> >
> http://dock.compbio.ucsf.edu/DOCK_6/tutorials/sphere_generation/generating_spheres.htm
> > If your complicated protein is big then that may put a practical limit
> > on the size of your grid; one approach would be to divide and conquer
> > the big protein binding sites into subsets that would be processed
> > separately.
> > Perhaps others have comments ?
> > 
> > > DOCK6.1 is being compiled on a parallel amd64 machine (Debian Linux amd64
> > with
> > > 4GB RAM per cpu, ie a machine better suited for ab initio calculations than
> > > classical MD), where also Amber9 is installed. Although the X system exists
> > (in
> > > fact, xleap requires it) and even Gnome could be called, the machine is
> > > normally used with X killed. I manage anything graphical for Amber9 (to the
> > > exception of xleap) on a single-Athlon desktop (Debian Linux i386 etch).
> > The
> > > two machines are scp linked, which is the way I exchange files between the
> > two
> > > machines.
> > >
> > > At 32bit I (1) prepare pdb files for "small molecules" with a MM package;
> > (2)
> > > examine everything graphical with VMD.
> > >
> > > On these basis, where to install Chimera for DOCK? I noticed that Chimera
> > calls
> > > both Dock Prep and Antechamber.
> > 
> > 
> > Installing Chimera on the same machine as VMD seems reasonable.
> > Any textual input and output files can be communicated between the machines.
> > 
> > > Similarly, where to install dms for DOCK?
> > 
> > dms is small.  I install it in my dock directory.
> > 
> > 
> > > I understand that these are generic question, normally not for mailing
> > lists.
> > > Though, a couple of suggestions would greatly help me (an organic chemist,
> > just
> > > now beginning to touch the area of the biochemist) in the high complexity
> > of
> > > the docking affairs.
> > 
> > 
> > This is the best place for all questions.
> > 
> > Scott