[Dock-fans] docking on a linux cluster

Terry Lang terry at cgl.ucsf.edu
Wed Aug 10 17:44:47 PDT 2005

Dear Jose,

1)  There is no difference in the code base itself between the operating 
systems, only in the precompiled binaries.  However, that being said, we 
have had several issues with how the compilers on different systems 
interpret the code.  This issue has become more of a problem for Mac OS 
X lately because we have an increased number of Mac users who are trying 
out an increasing variety of functions.  We do try to address these 
issues and fix them as soon as possible with bug fixes on the web site.

2)  In the 5.2.0 release, we now recommend using energy score for both 
primary and secondary scoring.  In DOCK 4, we had recommended using the 
contact score first, but that was primarily for length of calculation 
issues.  Now that computers are faster, we recommend as above.

3)  There were some people having the same problem with large 
databases.  It turned out that they were using sphere files that 
contained multiple clusters, which through off the sphere reader 
internally but did not warn the user.  This particular issue can be 
fixed by using a sphere file with only cluster or by applying the 
appropriate bug fix from the web site.  If you are still having problems 
with your database, please send me the tarred files and I will take a 
look (NOTE:  Please send these files to me only so as not to crash the 
discussion board).


Jose A wrote:

>Hello everybody,
>I have some questions...
>1) has DOCK been developed on other platforms (windows, linux, mac) to
>the same extent as for UNIX? would you say DOCK runs for Mac as
>smoothly as it does for UNIX? or is the UNIX version constantly
>"ahead" of the others?
>2) I noticed that in the demo scripts DOCK is set up with energy
>scoring both for primary AND secondary scoring. is this the recomended
>combo? I used to score with contact score primary and energy as
>secondary. is this better or worse? or does it vary from target to
>target? After reading the manual a couple of times, it is still not
>clear to me how setting up one scoring function as primary or
>otherwise makes any difference. If anybody cares to educate me I would
>be grateful...
>3) It seems there is at least another person having a similar problem
>to mine when trying to use DOCK on a linux cluster. I can run the demo
>fine. I can change some keywords to my liking without problems (such
>as turning on the bump filter and switching automated matching
>off)...but as soon as I switch to a database
>with a few hundred molecules DOCK scores a few and then gets stuck and
>kills itself (without warning). Rerunning repeats the process with
>DOCK failing at the same molecule. Removing the "bad" molecule allows
>DOCK to go farther down the list, but it gets stuck again at some
>other molecule later. The mol2 file seems to be ok (zinc#267). How
>could I begin to troubleshoot this???? can I assume grid and cluster
>files are "OK" since a single molecule runs fine?  finally, I have to
>say so far I hadn't had this problem on an SGI workstation...should I
>run DOCK with dock5.linux on the cluster instead of dock5.mpi? or that
>wouldn't make a difference?
>Thanks for giving this any thought,
>PS: here is a copy of the PBS script I am using to run dock. If you
>think it is missing something let me know...
>#!/bin/csh -f
>#PBS -N dock_only
>#PBS -o dock_only.out
>#PBS -e dock_only.err
>#PBS -m abe
>#PBS -M my email address goes here
>#PBS -q submit
>#PBS -l nodes=2:ppn=1
>set path = ($path /opt/lam/icc/bin)
>set MPIRUN = mpirun
>setenv LD_LIBRARY_PATH /opt/lam/icc/lib:/opt/intel/cc81/lib
>echo Running dock_only...
># Regenerate ligand:receptor complex with flexible ligand
>set EXE = /apps/dock/dock52/bin/dock5
>cd /home/jahp/demo/4_dock
>mpirun -np 2 $EXE -i dock.in -o dock.out
>Dock-fans mailing list
>Dock-fans at docking.org

P. Therese Lang
Kuntz and James Labs
UCSF--Chemistry and Chemical Biology
Phone: (415)476-3986


More information about the Dock-fans mailing list