| 8:16:15 AM PDT - Fri, Jun 27th 2014 |
|
Hi,
I'm having some trouble running a QM/MM optimization on a system of a large protein in a box of solvent. I found that when I define two regions, qm and mm, the QM region optimizes reasonably, but the other atoms (MM) barely move at all. In the first cycle of optimization, the MM atoms undergo only about 40 steps of minimization. On the other hand, when I define two regions qm and solvent (i.e., not including the MM atoms of the protein), the solvent undergoes about 1400 steps of optimization during the first cycle, and the atoms move as would be expected during a minimization task. This is strange, since solvent is a subset of mm, so it's surprising that using mm for the minimization protocol would have a barely distinguishable effect, but solvent would look normal.
The system is one that had already been minimized and had undergone an MD run (for equilibration) in AMBER, with the entire system being under MM control (light restraints on the future QM region). Also, I am using TIP3P water for the solvent, but I don't think that has caused any problems (I wrote a T3E fragment and segment file).
Here is the input file, run.nw:
start fgfr
permanent_dir ./output
scratch_dir ./scratch
charge -4
prepare
source density-out.pdb
new_top new_seq
solvent name HOH model T3E
new_rst
amber
modify atom 169:_CB quantum
#### 39 more atoms modified as quantum here
modify atom 320:_HD2 quantum
update lists
ignore
write fgfr_ref.rst
write fgfr_ref.pdb
end
task prepare
md
mwm 32000
msa 32000
system fgfr_ref
noshake solute
print topol solvent
record coord 20
end
basis
* library "6-31G"
end
dft
xc b3lyp
maxiter 2000
end
qmmm
link_atoms hydrogen
###### HERE IS THE ONLY DIFFERENCE BETWEEN FILES:#######################
region qm mm ##### NOTE: in the other file, this says "region qm solvent"
########################################################################
maxiter 10 3000
ncycles 5
density espfit
xyz qregion
end
task qmmm dft optimize
I would really appreciate any help or pointers anyone can provide. I am happy to provide any input/output/processing files that would help.
Thank you for your time.
Sincerely,
Jerry Karp
Laboratory of Dr. David Cowburn
Albert Einstein College of Medicine
|