Re: Equivalent of gromacs' "rdd" option

From: Fotis Baltoumas (fbaltoumas_at_biol.uoa.gr)
Date: Mon Jul 25 2016 - 10:08:18 CDT

Hi,

Unfortunately, I don't know of any workarounds. Theoretically, you could
edit the part of the source that contains the rigidBonds function
(that's where the SHAKE/RATTLE constraints algorithm is) and add
LINCS-like code, but that would require more advanced knowledge of the
NAMD code than what I can offer.

You could try fooling around with the force magnitudes you use in the
elastic network or the colvars code, maybe that would fix things
somewhat. Also, if the simulation crashes occur after a while (for
example, after the first 10 ns) and not immediately, you could split
your big simulation to smaller ones and work with restart files.

I hope you find a solution

On 07/25/2016 10:56 AM, Marlon Sidore wrote:
> Thanks.
>
> I will fool around with margin, with hopes that it reduces the crashes.
>
> My simulations are already running (and crashing) at 10fs, probably
> since I am also using the colvars module and putting more forces on
> the protein bonds. There is no workaround to the absence of LINCS for
> all bonds then ? Because the main thing about using NAMD with MARTINI
> here is using the colvars.
>
> Attaching my configuration parameters below:
> #Run namd in the folder
> set pvmode "p" ;#p to activate pression control
> set temode "t" ;#t to activate temperature control
> set restart 1 ;#0 for new sim, 1 for continuation
>
> set inName ha8a-411
> set outName ha8a-412
>
> proc get_first_ts { xscfile } {
> set fd [open $xscfile r]
> gets $fd
> gets $fd
> gets $fd line
> set ts [lindex $line 0]
> close $fd
> return $ts
> }
>
> #input
> coordinates init_AqpZ.pdb
> bincoordinates $inName.restart.coor
> structure DimerAqpZ.psf
>
> paraTypeCharmm on
> parameters martini-par/martini-protein-bonds.par
> parameters martini-par/martini-protein-angles-cos.par
> parameters martini-par/martini-protein-dihedrals.par
> parameters martini-par/martini-all-nonb.par
> parameters martini-par/martini-lipids-bonds-angles-dihedrals.par
>
> set temp325.0
>
> if {$restart == 1} {
> binvelocities $inName.restart.vel
> extendedSystem $inName.restart.xsc
> set currenttimestep [get_first_ts $inName.restart.xsc]
> COMMotion yes
> } else {
> temperature $temp
> set currenttimestep 0
> COMmotion no
> }
>
> firsttimestep $currenttimestep
>
> #cutoff et martini
> cutoff 12.0
> switching on
> switchdist 9.0
> cosAngles on
> martiniSwitching on
> dielectric 15.0
> pairlistdist 24.0
> stepspercycle 20
> margin 2.0 ;#échantillonnage de l'espace
> splitpatch hydrogen
>
> #Boundary conditions
> cellBasisVector1 200.0 0.0 0.0
> cellBasisVector2 0.0 208.0 0.0
> cellBasisVector3 0.0 0.0 93.0
> cellOrigin 0.0 0.0 0.0
> wrapWater on
> wrapAll on
>
> #Output
> binaryoutput yes ;#binary plus précis que pdb : utiliser
> flipbinpdb pour utiliser ces fichiers en sortie
> binaryrestart yes
> outputname $outName
> restartname $outName.restart
> restartfreq 10000
>
> DCDfile $outName.dcd
> DCDfreq 10000
>
> outputTiming 10000 ;#CPU & wallclock times, energy and
> pressure output to stdout
> outputEnergies 10000
> outputPressure 10000
>
> XSTfile $outName.xst
> XSTFreq 1
>
> timestep 10.0
>
> #Constraints and restraints
> constraints on
> fixedAtoms off ;# switch to off after PR simulation
> extraBonds yes ;#For elastic network
> extraBondsFile ElastDimer.dat #For elastic network
> consref init_AqpZ.pdb
> conskfile harmonichelicesBAS.cnst
> conskcol O
> consexp 2
> constraintScaling 1
>
> #Colvars
> colvars on
> colvarsConfig ha8a.in <http://ha8a.in>
> colvarsInput $inName.restart.colvars.state
>
> # CONSTANT-T
> if {$temode == "t"} {
> langevin yes ;# do langevin dynamics
> langevinDamping 1 ;# damping coefficient(gamma)5/ps
> langevinTemp $temp
> langevinHydrogen off ;# don't couple langevin bath to hydrogens
> }
>
> # Constant Temperature Control ONLY DURING EQUILB
> #reassignFreq 500; # reassignFreq: use this to
> reassign velocity every 500 steps
>
> # CONSTANT-P
> useGroupPressure no
> useFlexibleCell yes
> useConstantRatio yes
>
> if {$pvmode == "p"} {
> langevinPiston on
> langevinPistonTarget 1.01325 ;# in bar -> 1 atm
> #one may need to bump up the pressure constants at first
> langevinPistonPeriod 2000 ;#usually 2000
> langevinPistonDecay 1000 ;#usually 1000
> langevinPistonTemp $temp
> }
>
> #1-4's
> exclude 1-2
>
> #Run
>
> run 60000000
>
>
> Marlon Sidore
>
>
> PhD Student
> Laboratoire d'Ingénierie des Systèmes Macromoléculaire (LISM)
> CNRS - UMR7255
> 31, Chemin Joseph Aiguier
> 13402 cedex 20 Marseille
> France
>
> 2016-07-22 18:05 GMT+02:00 Fotis Baltoumas <fbaltoumas_at_biol.uoa.gr
> <mailto:fbaltoumas_at_biol.uoa.gr>>:
>
> Hello,
>
> The quick answer is no, there isn't any equivalent of -rdd in
> Namd. But you could try replicating its effect by fooling around
> with parameters that determine the system partition size, such as
> "margin" (manually set its value in the configuration file, start
> with low values and increase until the crashes stop). Although,
> until you give us your configuration parameters and the actual
> error you get, I cannot be of more help.
>
> Having said that, it should be pointed out that the NAMD version
> of protein topology/parameters of MARTINI is problematic at best.
> The main issue is that, since NAMD doesn't implement any
> equivalent for LINCS (constraints for all bonds, not just those of
> hydrogens), all bond constraints were modeled as regular bonds but
> with very high force constant values, which leads to a lot of
> crashes when timesteps of 20 fs or above are used. If you're
> using such a value, try lowering the timestep to 10 fs and see if
> that solves your issue. Most crashes I have had when using
> MARTINI with NAMD to simulate proteins were solved by doing that
> (the lipid parameter implementation was more successful since no
> constraints needed to be defined).
>
>
> Hope that helps
>
>
>
> On 07/22/2016 06:12 PM, Marlon Sidore wrote:
>
> Hello,
>
> I am simulating membrane proteins with the martini forcefield
> in NAMD but
> have been struggling with random crashes from the beginning of
> this journey
>
>
> --
> *******************************************
> Fotis A. Baltoumas
> Phd Candidate, Bioinformatics Postgraduate Programme
> Department of Cell Biology and Biophysics
> Faculty of Biology, University of Athens
> Panepistimiopolis, Athens 157 01, GREECE
> --------------------------------------
>
> email : fbaltoumas_at_biol.uoa.gr <mailto:fbaltoumas_at_biol.uoa.gr>
> http://biophysics.biol.uoa.gr
> http://bioinformatics.biol.uoa.gr
> *******************************************
>
>

-- 
*******************************************
Fotis A. Baltoumas
Phd Candidate, Bioinformatics Postgraduate Programme
Department of Cell Biology and Biophysics
Faculty of Biology, University of Athens
Panepistimiopolis, Athens 157 01, GREECE
   --------------------------------------
email : fbaltoumas_at_biol.uoa.gr
http://biophysics.biol.uoa.gr
http://bioinformatics.biol.uoa.gr
*******************************************

This archive was generated by hypermail 2.1.6 : Tue Dec 27 2016 - 23:22:20 CST