[Dock-fans] Dock AMBER "bad" mols

Scott Brozell sbrozell at scripps.edu
Fri Aug 22 14:54:17 PDT 2008


Hi,

Please use dock-fans, and do not send email directly to the developers.

On Mon, 18 Aug 2008 xhan at ucsd.edu wrote:

> using DOCK6.2 for research purposes. I am using ZINC7's "drug-like" and
> "lead-like" databases. While executing prepare_amber.pl, some compounds in
> some slices would fail to prepare correctly. I was wondering if you know
> about this issue and possibly a solution? I've compiled all the output
> files of these compounds and a list of their ZINC IDs in 2 .tar.gz files.

Yes, a usually small fraction of almost all collections of ligands
fails to prepare correctly because Amber score requires a complete
force field.  In particular, Amber's GAFF is used for ligands. 
The prepare_amber scripts call three amberization scripts.
The amberization scripts call several Amber programs - LEaP and
antechamber.  When things go wrong, detailed knowledge of all these
pieces may be necessary.  Preparation for Amber score is preparation
for an Amber molecular dynamics calculation - a much higher level of
preparation than for other DOCK scores.

In general, the two solutions are
1) to ignore the errors from failing ligands which means
ignoring the ligands in the Amber rescoring; set dock6 parameter
amber_score_abort_on_unprepped_ligand to no, and
2) to investigate, correct, and re-prepare for each ligand failure:
Determine the failure by studying amberize_ligand.*.out - 
the most common failures are
z] charge issues, such as mopac convergence failures, wrong total charge,...;
y] Missing force field parameters;
x] bond typing issues;
w] atom typing issues;
with cases w and x usually being much more troublesome than z and y;
Determine a correction -
z] sometimes the geometry is obviously bad and can be fixed, sometimes
with detailed knowledge of mopac convergence can be attained and then
detailed knowledge of antechamber is needed to use that mopac output,
the simplest approach is to use the existing charges instead of
generating new AM1-BCC charges (in 6.3 we shall have this automated);
y] there is a recovery procedure described in the error message, manual,
and here: http://dock.compbio.ucsf.edu/DOCK_6/new_in_6.2.txt;
x] sometimes the geometry is obviously bad and can be fixed,
sometimes antechamber's -j 5 for atom and part bond type instead of the
default of -j 4 for atom and full bond type will work - detailed
knowledge of antechamber and the amberization process is needed to 
employ this correction;
w] usually the molecule is weird or the geometry is obviously bad...;
Re-prepare - sometimes prepare_amber.pl can be rerun to implement
corrections; if not then detailed knowledge of amberization is needed:
read the dock manual and open up the amberize scripts.

The antechamber home page
http://amber.scripps.edu/antechamber/antechamber.html
and the Amber mailing list
http://structbio.vanderbilt.edu/archives/amber-archive/
http://amber.scripps.edu/#reflector
should be searched since most antechamber issues have been aired already.
The dock fans archives should also be searched:
http://dock.compbio.ucsf.edu/DOCK_6/index.htm#searchdockfans


All your failures look troublesome.
Typical examples from your failures:

amberize_ligand.1018.out 
----
Using ligand file prefix: 22_top.1018
Using antechamber options: -nc -1 -c bcc
Executing antechamber to generate the mol2 file with GAFF and charges.

Info: Bond types are assigned for valence state 1 with penalty of 1

Total number of electrons: 127; net charge: -1
Number of electrons is odd: 127
Please check the total charge and your -nc flag
Running: /home/xhan/dock6/bin/bondtype -i ANTECHAMBER_BOND_TYPE.AC0 -o ANTECHAMBER_BOND_TYPE.AC -f ac -j full

Running: /home/xhan/dock6/bin/atomtype -i ANTECHAMBER_AC.AC0 -o ANTECHAMBER_AC.AC -p gaff
Error running antechamber!
----

This may look like case z, but is the more serious case w.
Check the structure, geometry and total charge - an odd number of electrons
indicates a non-closed shell molecule which is highly suspicious.
If you give this one a serious attempt and fail then send the molecule
and details to the Amber reflector; sometimes it's antechamber.


amberize_ligand.237.out
----
Using ligand file prefix: 36_top.237
Using antechamber options: -nc -2 -c bcc
Executing antechamber to generate the mol2 file with GAFF and charges.

Warning: the assigned bond types may be wrong, please :
(1) double check the structure (the connectivity) and/or
(2) adjust atom valence penalty parameters in APS.DAT, and/or
(3) increase MAXVASTATE in define.h and recompile bondtype.C
(4) increase PSCUTOFF in define.h and recompile bondtype.C
    Be cautious, use a large value of PSCUTOFF (>10) will significantly increase the computer time
Error: cannot run "/home/xhan/dock6/bin/bondtype -i ANTECHAMBER_BOND_TYPE.AC0 -o ANTECHAMBER_BOND_TYPE.AC -f ac -j full" in judgebondtype() of antechamber.c properly, exit
Running: /home/xhan/dock6/bin/bondtype -i ANTECHAMBER_BOND_TYPE.AC0 -o ANTECHAMBER_BOND_TYPE.AC -f ac -j full
Executing parmchk to generate the force field parameters.
Cannot open file 36_top.237.gaff.mol2, exit
grep: 36_top.237.frcmod: No such file or directory
Executing LEaP to generate the Amber input files.
----

This starts out as case x, follow the advice after the warning;
search the web pages above for help;
try -j 5 - to rerun antechamber manually:
antechamber -s 2 -i "input_mol2" -fi mol2 -o "output_mol2" -fo mol2 -nc -2 -c bcc -j 5

good luck,
Scott




More information about the Dock-fans mailing list