[gmx-users] grompp does not find atomtype

[Justin Lemkul]

On 7/2/14, 4:40 AM, Dawid das wrote:
> I have managed to deal with this problem in following way:
>
> In topology file some bonding parameters were actually missing but some
> bonding parameters were not needed, e.g. U-B or dihedral parameters for
> atoms which are not connected, that is they belong to different parts of my
> new residue.
> I found parameters for the missing ones and hashed (;) them out. Is it what
> I can do? Like I say, theses bonding parameters that I hashed out were for
> atom types which are not connected directly.
>
> Then I performed minimization with steepest descent and in output file I
> got:
>
>   4655e+05 Fmax= 6.53501e+03, atom= 1003^MStep=  426, Dmax= 2.5e-06 nm,
> Epot= -4.54655e+05 Fmax= 9.45125e+03, atom= 1003^MStep=  427, Dmax= 1.2e-06
> nm, Epot= -4.54655e+05 Fmax= 1.25724e+04, atom= 1003^M
> Stepsize too small, or no change in energy.
> Converged to machine precision,
> but not to the requested precision Fmax < 1000
>
> Double precision normally gives you higher accuracy.
> You might need to increase your constraint accuracy, or turn
> off constraints alltogether (set constraints = none in mdp file)
>
> writing lowest energy coordinates.
>
> Steepest Descents converged to machine precision in 428 steps,
> but did not reach the requested Fmax < 1000.
> Potential Energy  = -4.5465475e+05
> Maximum force     =  1.7057922e+04 on atom 1003
> Norm of force     =  1.1677969e+02
>
> But I continued with NVT MD simulation. This is my nvt-md.mdp file:
>
> title               = NVT MD
> integrator          = md
> constraints         = all-bonds
> dt                  = 0.001
> nsteps              = 20000
> nstenergy           = 100
> nstlist             = 10
> nstxout             = 1000
> nstvout             = 1000
> nstfout             = 0
> nstlog              = 100
> nstxtcout           = 1000
> xtcprecision        = 500
> ns_type             = grid
> coulombtype         = PME
> rlist               = 1.0
> rcoulomb            = 1.0
> rvdw                = 1.0

These cutoffs are incorrect for using CHARMM force fields.  See previous posts 
on proper settings.

> tcoupl              = nose-hoover
> tc-grps             = Protein SOL NA
> tau_t               = 0.1 0.1 0.1 ;(3 numbers -> because 3 tc-grps)
> ref_t               = 100 100 100 ;(3 numbers -> because 3 tc-grps)

Coupling water and ions separately is not sensible.

http://www.gromacs.org/Documentation/Terminology/Thermostats#What_Not_To_Do

> Pcoupl              = No
> gen_vel             = yes
> gen_temp            = 100
> gen_seed            = 173529
> energygrps          = Protein  SOL NA
> constraint_algorithm = LINCS
> pbc                 = xyz
>
> And in my NVT output I got error for LINCS:
>
> relative constraint deviation after LINCS:
> rms 0.036755, max 1.509884 (between atoms 998 and 1000)
> bonds that rotated more than 30 degrees:
>   atom 1 atom 2  angle  previous, current, constraint length
>     1001   1003   62.1    0.1410   0.2032      0.1410
>     1001   1002   68.1    0.1240   0.1962      0.1240
>     1000   1001   88.1    0.1460   0.3276      0.1460
>      998   1000   90.0    0.1390   0.3489      0.1390
>      986   1000   90.7    0.1400   0.2237      0.1400
>      985    986   30.7    0.1300   0.1499      0.1300
> Wrote pdb files with previous and current coordinates
>
>
> These atoms are those for my new residue. For next step I get more
> constraint deviation.
> Now, is it possible that this is because of wrong parameters? Or maybe I
> should minimize my system further?
> Here are all the files:
>
> http://www.speedyshare.com/jjvgd/md-simul.tar.bz2
>

When sharing files, please only post what is actually necessary to reproduce the 
problem - a self-contained force field directory, coordinates, topology, .mdp, 
and index file if needed.  A dump of dozens of files with no context gets 
burdensome to go through.

Your minimized geometry is clearly junk.  Just visualize the structure and you 
will see that your mCherry residue is completely distorted.  The very large 
(10^4) force after EM should have been the first clue that something is very wrong.

The problems almost certainly stem from the manipulations you've made to the 
angles and dihedrals.  I don't understand why you've commented some of them out. 
  For instance, the first angle that you've removed is atoms 977-978-992, which 
clearly should have an angle interaction because 977-978 and 978-992 are bonded. 
  Simply removing "inconvenient" interactions (i.e. lacking parameters or 
whatever) leads to a total meltdown of the model.

-Justin

-- 
==================================================

Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 601
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441
http://mackerell.umaryland.edu/~jalemkul

==================================================