> You can run some simulations with the new prmtop just to be sure (i.e.,
> make sure you get the same properties). However, did you run these
> simulations with pmemd? If so, pmemd should have failed, since by default
> it does not allow intermolecular bonds, and should have detected this
> error...
>
> I'm a bit concerned if this was not caught...
The MD simulations with the original prmtop file seems to be reasonable
(density, pressure, etc. looks ok). But I cannot make a direct 1-to-1
comparison between the simulations done with each prmtop files (which I think
is because of the differences in the sequences).
I did a test, where I modified the restart equilibration file (which then I
used with the right prmtop file - prmtop2). I see some slight differences in
TEMP and PRESS in the first steps when I compare the MD simulations done with
prmtop1/mdeq.rst (original -bad- prmtop and mdeq.rst) and prmtop2/mdeq_mod.rst
(right prmtop with modified mdeq.rst files).
----- prmtop2/mdeq2_mod.rst ----------------------------
NSTEP = 1 TIME(PS) = 120.001 TEMP(K) = 301.90 PRESS = -54.6
--------------------------------------------------------
----- prmtop1/mdeq2.rst ----------------------------
NSTEP = 1 TIME(PS) = 120.001 TEMP(K) = 301.89 PRESS = -55.3
----------------------------------------------------
I think the differences in TEMP and PRESS is due to different ordering of the
atoms in the system.
By the way, I can run both prmtop files with PMEMD and SANDER. Why do you say
that pmemd should not work with this type of a system? The frcmod file includes
all the bond/angle/dihedral/vdw/etc. for the AG binding region (so it actually
looks like a one molecule ; protein+AG).
Ilyas,
PS: There is one thing that seems to be changed between AMBER9 and AMBER11. In
AMBER9, the AVERAGES printed at the end of output file is using all the nstlim
steps to calculate the averages, while in AMBER11 the AVERAGES calculated is
using the printed NSTEP values (a total of nstlim/ntpr values). For example,
let's say nstlim=5000000 and ntpr=1000. In AMBER9, the AVERAGES uses 5000000
values to get the averages while AMBER11 uses 5000000/1000=5000 values. I do
not know why this was changed or if AMBER12 does the same thing as AMBER11, but
the AVERAGES printed in AMBER9 is much more useful.
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Tue Jul 03 2012 - 13:00:03 PDT
