Converting simulation files in CHAMBER to run CHARMM-based force field simulations in Amber (notes from Jun):
I encourage everyone to use CHAMBER v14 to convert CHARMM input psf, pdb into AMBER format prmtop and inpcrd. CHAMBER v14 is more flexible than v12. I have run energy calculation test with v14 which I will explain in the following and the result is exact the same as v12.
Here is the example to use v14 on ironman:
/usr/local/src/amber14/AmberTools/bin/chamber -psf chaps_solv3.psf -crd equi2.pdb -top top_all36_prot.rtf -param par_all36_prot.prm -str top_all36_lipid.rtf par_all36_lipid.prm top_all36_cgenff.rtf par_all36_cgenff.prm toppar_water_ions_jpc2013.str CHAPS.prm -cmap -box 116.444 116.444 116.444 -p prmtop.v14 -inpcrd inpcrd.v14 -verbose > chamber.log.v14
v12 and v14 comparison:
v12 requires combination of all rtf files into one file and all prm files into one file. This requires tedious work to combine files and check energy whenever new rtf/prm is added.
v14 can read multiple rtf,prm and str files. Please note that -top and -param only takes one file and you put the rest of your rtf, prm and str files after -str. v14 also claims to read xplore format psf which I have not tested.
Here is how to run one step energy calculation to validate the use of CHAMBER in general. I give this procedure in case anyone in the future needs to validate one force field running in the other type of soft ware. (eg. CHARMM in AMBER)
1. using charmm to calculate one step energy of an example system (eg. protein, lipid, ligand, water, ion), use ewald and cutoff=999 to run two separate calculations. CHARMM reads all the c36 force parameter files neccesary.
2. create rtf and prm which combines all individual files. This is a tedious and error prone step. So far, I have combined c36 protein, lipid, cgenff, water, ions, palm, retinal, CHAPS, DDM
3. using CHARMM again to read the combinant rtf and psf files and run one step energy calculation of the same system. Exactly the same energy means the combinant rtf and prm files is correct. If not, go check step 2.
4. Run CHAMBER to convert files
5. Run one step calculation using AMBER sander compare the energy. Make sure the nonbond calculation set up is the same as in CHARMM. All the energy terms except for slight different electrostatic energy should match with CHARMM.
6. Compare CHAMBER v12 and v14: convert the same psf/pdb and run energy calculation. This time we can use pmemd.cuda
Why taking extra steps to double check: We previously found CHAMBERv12 did not reproduce the same energy as in CHARMM and reported the bug. So it is always important to check by yourself. You are responsible for your research, not anybody else.
Input for running membrane simulations in Amber (from Jun)
# NPT dynamics with 10ps snapshot frequency
&cntrl
irest=1, ! restrt
ntx=5, ! coordinates, velocities, box size,will be read from a formatted file.)
imin=0, ! (no minimization)
dt=0.002,
ntf=2, ! (Force evaluation,complete interaction is calculated)
ntc=2, ! (SHAKE,bonds involving hydrogen are constrained)
tol=0.0000001, ! (shake tolerance)
iwrap=1, ! coordinates written to the restart and trajectory files will be "wrapped" into a primary box
cut=8.0, ! (default nonbonded cutoff)
ntt=3, ! (Langevin dynamics for temperature control)
ig=-1, ! random seed
gamma_ln=5.0, ! (collision frequency)
temp0=300.0, ! (reference temperature)
barostat=1, !
csurften=3, ! Constant surface tension with interfaces in the xy plane
ntp=3, ! ( pressure control, semi-anisotropic (xy-,z-) pressure scaling)
ntb=2, ! (constant pressure (default when ntp > 0))
gamma_ten=0.0, !Surface tension value in units of dyne/cm. Default value is 0.0 dyne/cm.
ninterface=2, !Two interfaces is appropriate for a lipid bilayer system and is the default value
taup=8.0, ! (Pressure relaxation time)
ntpr=5000, ! (print energy output interval)
ntwr=5000, ! (frequency to write restart file)
ntwe=0, ! (write energies and temperature to mden file)
ntwx=5000, ! (write coordinates to the mdcrd file)
nstlim=50000000, ! (Number of MD-steps 100ns)
/
I encourage everyone to use CHAMBER v14 to convert CHARMM input psf, pdb into AMBER format prmtop and inpcrd. CHAMBER v14 is more flexible than v12. I have run energy calculation test with v14 which I will explain in the following and the result is exact the same as v12.
Here is the example to use v14 on ironman:
/usr/local/src/amber14/AmberTools/bin/chamber -psf chaps_solv3.psf -crd equi2.pdb -top top_all36_prot.rtf -param par_all36_prot.prm -str top_all36_lipid.rtf par_all36_lipid.prm top_all36_cgenff.rtf par_all36_cgenff.prm toppar_water_ions_jpc2013.str CHAPS.prm -cmap -box 116.444 116.444 116.444 -p prmtop.v14 -inpcrd inpcrd.v14 -verbose > chamber.log.v14
v12 and v14 comparison:
v12 requires combination of all rtf files into one file and all prm files into one file. This requires tedious work to combine files and check energy whenever new rtf/prm is added.
v14 can read multiple rtf,prm and str files. Please note that -top and -param only takes one file and you put the rest of your rtf, prm and str files after -str. v14 also claims to read xplore format psf which I have not tested.
Here is how to run one step energy calculation to validate the use of CHAMBER in general. I give this procedure in case anyone in the future needs to validate one force field running in the other type of soft ware. (eg. CHARMM in AMBER)
1. using charmm to calculate one step energy of an example system (eg. protein, lipid, ligand, water, ion), use ewald and cutoff=999 to run two separate calculations. CHARMM reads all the c36 force parameter files neccesary.
2. create rtf and prm which combines all individual files. This is a tedious and error prone step. So far, I have combined c36 protein, lipid, cgenff, water, ions, palm, retinal, CHAPS, DDM
3. using CHARMM again to read the combinant rtf and psf files and run one step energy calculation of the same system. Exactly the same energy means the combinant rtf and prm files is correct. If not, go check step 2.
4. Run CHAMBER to convert files
5. Run one step calculation using AMBER sander compare the energy. Make sure the nonbond calculation set up is the same as in CHARMM. All the energy terms except for slight different electrostatic energy should match with CHARMM.
6. Compare CHAMBER v12 and v14: convert the same psf/pdb and run energy calculation. This time we can use pmemd.cuda
Why taking extra steps to double check: We previously found CHAMBERv12 did not reproduce the same energy as in CHARMM and reported the bug. So it is always important to check by yourself. You are responsible for your research, not anybody else.
Input for running membrane simulations in Amber (from Jun)
# NPT dynamics with 10ps snapshot frequency
&cntrl
irest=1, ! restrt
ntx=5, ! coordinates, velocities, box size,will be read from a formatted file.)
imin=0, ! (no minimization)
dt=0.002,
ntf=2, ! (Force evaluation,complete interaction is calculated)
ntc=2, ! (SHAKE,bonds involving hydrogen are constrained)
tol=0.0000001, ! (shake tolerance)
iwrap=1, ! coordinates written to the restart and trajectory files will be "wrapped" into a primary box
cut=8.0, ! (default nonbonded cutoff)
ntt=3, ! (Langevin dynamics for temperature control)
ig=-1, ! random seed
gamma_ln=5.0, ! (collision frequency)
temp0=300.0, ! (reference temperature)
barostat=1, !
csurften=3, ! Constant surface tension with interfaces in the xy plane
ntp=3, ! ( pressure control, semi-anisotropic (xy-,z-) pressure scaling)
ntb=2, ! (constant pressure (default when ntp > 0))
gamma_ten=0.0, !Surface tension value in units of dyne/cm. Default value is 0.0 dyne/cm.
ninterface=2, !Two interfaces is appropriate for a lipid bilayer system and is the default value
taup=8.0, ! (Pressure relaxation time)
ntpr=5000, ! (print energy output interval)
ntwr=5000, ! (frequency to write restart file)
ntwe=0, ! (write energies and temperature to mden file)
ntwx=5000, ! (write coordinates to the mdcrd file)
nstlim=50000000, ! (Number of MD-steps 100ns)
/