[gmx-users] Drude-2013 FF, seg fault 11 on repeat minimization of beta-D-glucose monomer
Justin Lemkul
jalemkul at vt.edu
Wed Feb 15 20:11:26 CET 2017
On 2/15/17 1:46 PM, Dayhoff, Guy wrote:
> Hi,
>
> Thanks for taking the time to humor my inquiry.
>
> I’m hoping to find a fix for my issue or obtain a working beta-D-glucose monomer model that I can use to
> continue troubleshooting my issue. Details below… (This is my first post, forgive me for including too much/
> too little context)
>
> My Best,
> Guy Dayhoff
>
>
>
> I am attempting to employ the Drude-2013FF available below to use with some carbohydrates:
>
> http://mackerell.umaryland.edu/download.php?filename=CHARMM_ff_params_files/drude-2013b_16sep2016.tgz
>
> I have obtained and installed the DRUDE compatible GROMACS distro available via:
>
> git clone git://git.gromacs.org/gromacs.git
> cd gromacs
> git checkout drude
>
> I am using pre-equilibrated SWM4-NDP water available from:
>
> http://mackerell.umaryland.edu/download.php?filename=CHARMM_ff_params_files/swm4-ndp_1024.gro
>
> I have adjusted Drude parameters in my mdp files as demonstrated in the additional material from the publication
> regarding implementation of the Drude FF and extended Lagrangian dynamics in GROMACS.
>
> Publication: http://dx.doi.org/10.1002/jcc.23937
>
> ; DRUDE PARAMETERS
> drude = yes
> drude-mode = SCF
> drude-hyper = yes
> drude-khyp = 16736000.0
> drude-r = 0.02
> drude-pow = 4
>
> I prepared a Drude model of beta-D-glucose(BGLC) to run before using more complex models by:
>
> Downloading .CIF and .SDF data from RCSB-PDB @ https://www3.rcsb.org/ligand/BGC
>
> Feeding this data into the CHARMM-GUI’s Ligand Reader & Modeler to produce .PDB and .PSF files
>
> Feeding the .PDB and .PSF files into the CHARMM-GUI’s Drude Prepper producing bDgluc-Drude.pdb
>
> I generated a topology file with pdb2gmx without any warnings/notes/errors:
>
> gmx pdb2gmx -f bDgluc-Drude.pdb -o bDgluc-Drude.gro -p bDgluc-Drude.top
>
> I placed the BGLC model in the center of a 3.0 x 3.0 x 3.0 box and minimized successfully:
>
> gmx editconf -f bDgluc-Drude.gro -o bDgluc-Drude-CEN.gro -box 3.0 3.0 3.0 -center 1.5 1.5 1.5
>
> gmx grompp -v -f ../mdps/min.mdp -c bDgluc-Drude-CEN.gro -p bDgluc-Drude.top -o bDgluc-Drude-EMV.tpr
>
> gmx mdrun -v -deffnm bDgluc-Drude-EMV
>
> I then solvate the system and attempt to minimize it once more, this results in a seg fault 11:
>
> gmx solvate -cp bDgluc-Drude-EMV.gro -cs swm4-ndp.gro -p bDgluc-Drude.top -o bDgluc-Drude-SOL.gro
>
> gmx grompp -v -f ../mdps/min.mdp -c bDgluc-Drude-SOL.gro -p bDgluc-Drude.top -o bDgluc-Drude-EMS.tpr
>
> gmx mdrun -v -deffnm bDgluc-Drude-EMS
>
> Step= 0, Dmax= 1.0e-02 nm, Epot= nan Fmax= 3.77896e+06, atom= 16
> Segmentation fault: 11
>
> Likewise, if I simply take the minimized BGLC and attempt to minimize a second time I get the seg fault 11:
>
> gmx grompp -v -f ../mdps/min.mdp -c bDgluc-Drude-EMV.gro -p bDgluc-Drude.top -o bDgluc-Drude-EMV2.tpr
>
> gmx mdrun -v -deffnm bDgluc-Drude-EMV2
>
> Step= 0, Dmax= 1.0e-03 nm, Epot= nan Fmax= 1.95057e+04, atom= 11
> Segmentation fault: 11
>
> I have attempted to isolate the cause and have noted:
>
> minimization and subsequent runs of a box of pure swm4-ndp produces no such errors, indicating the issue
> is most likely related to the BGLC molecule in the system.
>
> when beginning with a beta-D-glucose model not prepared with the CHARMM-GUI Drude Prepper (i.e. without
> lone pairs and drudes included in the coordinate file) pdb2gmx properly adds all drudes, however fails to add
> LP5A and LP5B, LPX5 is added though (CHARMM-GUI’s Drude Prepper doesn’t appear to have this issue).
>
> using the -missing flag and proceeding with minimzation, solvation, NVT and NPT ensembles using the incomplete
> model does not cause any seg faults and completes without errors. In this case however, the molecule has a charge
> (which corresponds exactly to the missing LP5A and LP5B).
>
> with all things considered, I believe the issue lies with the handling of LP5A/LP5B in the BGLC residue in my model
> but am unsure of how to proceed at this point.
>
It appears that LP5A and LP5B are missing from the .rtp definition so indeed
they will not be built and the exclusions, etc. will be wrong. I didn't test
all the carbohydrates yet so I don't know why this occurred.
I'll look into it.
-Justin
> the seg fault: 11 occurs on the first step of the minimizations
>
> My complete min.mdp for reference:
>
> ; PREPROCESSING
> define =
>
> ; RUN CONTROL
> integrator = steep
> nsteps = 10000
> dt = 0.001
> comm-mode = Linear
> nstcomm = 1
> nstcalcenergy = 1
>
> ; NEIGHBORSEARCHING
> ns-type = grid
> rlist = 1.2
> pbc = xyz
> cutoff-scheme = Verlet
> periodic-molecules = no
>
> ; BONDS
> constraints = none
> continuation = no
>
> ; ELECTROSTATICS AND VDW
> coulombtype = PME
> rcoulomb = 1.2
> vdwtype = Cut-off
> rvdw = 1.2
> DispCorr = EnerPres
> vdw-modifier = potential-switch
> rvdw-switch = 1.0
>
> ; ENERGY MINIMIZATION OPTIONS
> emtol = 10.0
> emstep = 0.001
>
> ; EWALD
> fourierspacing = 0.1
> pme_order = 4
> ewald_rtol = 1e-5
> ewald_geometry = 3d
>
> ; DRUDE PARAMETERS
> drude = yes
> drude-mode = SCF
> drude-hyper = yes
> drude-khyp = 16736000.0
> drude-r = 0.02
> drude-pow = 4
>
>
--
==================================================
Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow
Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 629
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201
jalemkul at outerbanks.umaryland.edu | (410) 706-7441
http://mackerell.umaryland.edu/~jalemkul
==================================================
More information about the gromacs.org_gmx-users
mailing list