I've been using Conda ambertools 21.11.
I'm already using a small system (NALA-ALA-CALA) and, to Ray, in my first
email I gave all the steps for anyone to reproduce the issue, so yes, I'm
absolutely sure I'm using the same files.
With sander I do use ntmin=2 (steep descent) and maxcyc=0 so I want the
energies at step 0 of my systems.
I did suspect that python was doing 1 or more md steps so I played with md
in mdin for sander (igb=6, imin=0, maxcyc=0, ntmin=2, nstlim=X), where I
increased X one by one but I didn't reproduce the energies seen from
pysander.
Also, it's not SHAKE since pysander inp.ntc = inp.ntf = 1 which are the
same as default in sander.
So, I dive into the code.
I'm using gitpod from
https://github.com/dacase/msander
Installed `sudo apt install gfortran` there.
Then did:
./configure --no-rism --no-netcdf --no-boost
make clean && make install
Alas, it failed to build with:
...
[MSANDER] FC extra_pts.F90
[MSANDER] FC ../rism/amber_rism_interface.F90
../rism/amber_rism_interface.F90:16:6:
16 | use safemem
| 1
Fatal Error: Cannot open module file ‘safemem.mod’ for reading at (1): No
such file or directory
compilation terminated.
make[2]: *** [Makefile:172: amber_rism_interface.API.o] Error 1
make[1]: *** [Makefile:10: serial] Error 2
make[1]: Leaving directory '/workspace/msander/src'
make: *** [Makefile:4: install] Error 2
gitpod /workspace/msander $
My reasoning here was to build and compare all the original tools from
source and in a clean, fresh, isolated machine.
On Thu, 13 Jan 2022 at 17:37, David A Case <david.case.rutgers.edu> wrote:
> On Thu, Jan 13, 2022, Alan wrote:
> >In details:
> > Python x Sander
> >error angle : 0.48950 % 4.29874 x 4.27770
> >error bond : 2.04820 % 0.75717 x 0.77300
>
> If your bond and angle energies don't agree, then debugging should be much
> easier.
>
> Check things like doing a single-step minimiation instead of MD. But it
> probably comes down to getting into the code and seeing what the python
> gb_input is doing (in complete detail); I don't know that off the top of my
> head. It's possible that the python interface is doing somthing like
> setting SHAKE by default, which you are not doing in your sander
> calculation.
>
> You might try a very small system to see if that helps find the
> differences.
>
> ...dac
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
Alan Silva 🚲🏊♂🏃♂🇧🇷🇫🇷🇬🇧
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Jan 14 2022 - 04:00:02 PST
