Dear Dave and all,
regarding the second issue of Dave's previous email:
The changes we made in the ICOSA code removed bugs in the ICOSA calculation and decomposition.
More specifically:
- The first commit on Feb 3 14:41:25 2012 corrected a wrong initialization of fortran variables in the icosasurf.F90 subroutine icosa_init(). Without this change the variables angle, axis(1), axis(2), and axis(3) are not set to the default values if a second force() call is conducted, as it is done in line 651 of runmin.F90.
- The second commit on Feb 3 14:41:52 2012 corrected a bug in the ICOSA decomposition analysis, which additionally caused example idecomp4 to fail. The problem can be seen in the decomp.out.save reference file provided currently for the example idecomp4. The total ESURF listed under FINAL RESULTS devided by surften (default = 0.005) is not equal to the surface area ESURF listed under TOTAL ENERGIES, which is calculated by summing up the decomposed SAS contributions. We corrected the calculation procedure in decomp.F90 subroutine decsasa_icosa(), so that now the two energies match.
Nadine provided new decomp.out.save files for both cases which were pushed into origin/master. The idecomp1 and idecomp4 tests pass now.
I still encounter a problem with sander_pbsa_decres and sander_pbsa_decpw. This seems to be related to the PB polar part, though. I will thus leave it to Ray and Qin to look into this.
Best regards
Holger
On 04.03.12, case wrote:
> There are two issues in this email: one regarding recent changes in the PB
> codes, the second regarding changes to icosasurf and decomp in sander.
> Qin and Ray should look at the first one, Holger and Nadine at the second.
>
> ----------------------------------------------------------------------------
> On Sun, Mar 04, 2012, Holger Gohlke wrote:
>
> > I forward an email from Nadine regarding the pbsa / decomp issue. Would
> > this be a starting point for further searching?
> >
> >
> > we re-checked the differences in the polar solvation energy. We
> > observed differences in all mm_pbsa/Examples test runs, in which the
> > solvation free energy is calculated with the PBSA module. For the
> > hybrid model test and the decomposition analyses the differences in PB
> > energy were especially large. According to our investigations these
> > differences were introduced by changes made between 27.02.2012, 5:35
> > pm (8ad8c24ef9016553bcc3e9830bb3aac040877873) and 03.03.2012, 1:37 pm
> > (23df75b9a36f3d1fceedc94be921f94776c96ffa).
>
> Thanks for the information. My current guess is that these were made by
> changes in the default PB parameters. If so, it is important to find out
> if there is a way to get Amber12 to give the same PB output as Amber11 gave.
>
> [Also, if Amber12 is giving significantly different results by default than
> Amber11, do we understand why? Are the new results supposed to be "better"
> than the old ones? It is rather disconcerting that we are seeing changes in
> PB behavior so very, very late in the development cycle. As a reminder,
> distributed code is supposed to be code that groups are actually using for
> research. It is not supposed to represent the latest ideas that might seem
> good but haven't been tested much.]
>
> ----------------------------------------------------------------------------
>
> In addition to the changes in mm_pbsa_Examples, there have been
> changes in the sander "decomposition" tests. In particular, the "idecomp1"
> test uses GB (not PB), so problems there are (probably) unrelated to changes
> in the PB code. These test cases look like they started to fail near the
> beginning of Feb. Specifically, the following two commits appear to have
> changed the idecomp1 test output:
>
> commit 8d206b3848de52d17549055b948848b1c1e558a9
> Author: Holger Gohlke <gohlke.hhu.de>
> Date: Fri Feb 3 14:41:25 2012 +0100
>
> Bugfix of initialization in icosa_init
>
> commit 6ac8353095b0adc12119a23193ffc7d29382d28d
> Author: Holger Gohlke <gohlke.hhu.de>
> Date: Fri Feb 3 14:41:52 2012 +0100
>
> Bugfix to ensure backward compatibility for ICOSA surface calculation to
> Amb
>
> [I cherry-picked these into the olderDusseldorf branch: before, the tests
> pass; afterwards they failed. -- note: just "make sander" in that branch,
> then "mv sander ../../bin" in order to run the tests.] So, Holger and Nadine
> should carefully revisit the above commits.
>
> ----------------------------------------------------------------------------
>
> ...thanks....dave
>
>
>
> _______________________________________________
> AMBER-Developers mailing list
> AMBER-Developers.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber-developers
--
++++++++++++++++++++++++++++++++++++++++++++++++++
Dr. Holger Gohlke
Professor fuer Pharmazeutische/Medizinische Chemie
Heinrich-Heine-Universitaet Duesseldorf
Institut fuer Pharm. und Med. Chemie
Universitaetsstr. 1
40225 Duesseldorf
Germany
Tel.: (+49) 211-81-13662; Fax: (+49) 211-81-13847
Email: gohlke.uni-duesseldorf.de
URL: http://cpclab.uni-duesseldorf.de
++++++++++++++++++++++++++++++++++++++++++++++++++
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Mon Mar 05 2012 - 09:30:02 PST
