Re: [gmx-users] CG-MD simulation of protein, always crash with protein
The representation of extended regions (b-sheets) is more stable using elastic bonds than a dihedral angle because the flexibility of the force field makes possible the occurrence of situations where the two bonds get aligned and therefore the plan used to defined the dihedral can not be calculated and the program crashes. But I am not sure that was your problem. You mentioned that the program was stopping due to too many lincs warnings. Cycles (trp, phe, his) seemed to be the cause. Do you still have lincs warnings, may be less? The other thing is that is your protein deforms as much as before, there is another way to simulate you protein with Martini: http://md.chem.rug.nl/cgmartini/index.php/tools/118-end XAvier. On Apr 20, 2010, at 7:16 AM, Trang wrote: Elastic bonds did help the system survive. I'm not really understand why do we need "more robust" bonds in this system. Does it mean that there exists large forces that cause my system to blow up? Would you mind giving an explanation? Thanks indeed for your taking time to help. Trang On Mon, Apr 19, 2010 at 4:26 PM, Martti Louhivuori > wrote: On 19 Apr 2010, at 06:49, Trang wrote: The protein topo is too large, so I put it here. http://pastie.org/926617 The protein topology is fine, but you could try also with elastic bonds instead of dihedrals in the extended regions (--elastic option). Elastic bonds are more robust than dihedrals, especially if running with a larger timestep (30fs). Here is the system topology. #include "../martini_v2.1.itp" #include "../martini_v2.1_aminoacids.itp" The last line is unnecessary. Use it only when simulating single aminoacids instead of a polypeptide. Minimized structure showed no close contact (with distance cutoff 1.6, even to 1.9). Production run stopped at That's only 0.16-0.19nm, so overlap is still possible. step 175552 (5266.56 ps) with Segmentation fault, preceeded with a lot of Lincs warnings (I set ring_bonds = "constraints"). This is the mdp file, this file goes along with the system in vaccuum that crashed. I'm not sure if What's different this time around? Last time you said you couldn't run it even in vacuum. Sounds like you are just hitting statistical limits with your combination of (extended) dihedrals + 30fs timestep. Use a smaller timestep (20fs for example) or switch to local elastic bonds for the beta-strands... and solvate the system (without overlap) so the simulation makes some sense! -martti- -- Post-doctoral research fellow Moleculaire Dynamica University of Groningen Nijenborgh 4, 9747AG Groningen, the Netherlands tel. +(31) 50 363 4339 | fax. +(31) 50 363 4398 -- gmx-users mailing listgmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use thewww interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/mailing_lists/users.php -- gmx-users mailing listgmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/mailing_lists/users.php -- gmx-users mailing listgmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/mailing_lists/users.php
Re: [gmx-users] CG-MD simulation of protein, always crash with protein
Elastic bonds did help the system survive. I'm not really understand why do we need "more robust" bonds in this system. Does it mean that there exists large forces that cause my system to blow up? Would you mind giving an explanation? Thanks indeed for your taking time to help. Trang On Mon, Apr 19, 2010 at 4:26 PM, Martti Louhivuori wrote: > On 19 Apr 2010, at 06:49, Trang wrote: > >> The protein topo is too large, so I put it here. http://pastie.org/926617 >> > > The protein topology is fine, but you could try also with elastic bonds > instead of dihedrals in the extended regions (--elastic option). Elastic > bonds are more robust than dihedrals, especially if running with a larger > timestep (30fs). > > > Here is the system topology. >> >> >> #include "../martini_v2.1.itp" >> #include "../martini_v2.1_aminoacids.itp" >> > > The last line is unnecessary. Use it only when simulating single aminoacids > instead of a polypeptide. > > > Minimized structure showed no close contact (with distance cutoff 1.6, >> even to 1.9). Production run stopped at >> > > That's only 0.16-0.19nm, so overlap is still possible. > > > step 175552 (5266.56 ps) with Segmentation fault, preceeded with a lot of >> Lincs warnings (I set ring_bonds = "constraints"). This is the mdp file, >> this file goes along with the system in vaccuum that crashed. I'm not sure >> if >> > > What's different this time around? Last time you said you couldn't run it > even in vacuum. > > Sounds like you are just hitting statistical limits with your combination > of (extended) dihedrals + 30fs timestep. Use a smaller timestep (20fs for > example) or switch to local elastic bonds for the beta-strands... and > solvate the system (without overlap) so the simulation makes some sense! > > > -martti- > -- > Post-doctoral research fellow > Moleculaire Dynamica > University of Groningen > Nijenborgh 4, 9747AG Groningen, the Netherlands > tel. +(31) 50 363 4339 | fax. +(31) 50 363 4398 > > -- > gmx-users mailing listgmx-users@gromacs.org > http://lists.gromacs.org/mailman/listinfo/gmx-users > Please search the archive at http://www.gromacs.org/search before posting! > Please don't post (un)subscribe requests to the list. Use thewww interface > or send it to gmx-users-requ...@gromacs.org. > Can't post? Read http://www.gromacs.org/mailing_lists/users.php > -- gmx-users mailing listgmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/mailing_lists/users.php
Re: [gmx-users] CG-MD simulation of protein, always crash with protein
On Mon, Apr 19, 2010 at 4:08 PM, XAvier Periole wrote: > > Hi, > > the rmsd you show is for the protein in vacuum? It goes up to 0.7 nm! That > is a very > large deformation, is it expected? Isn't your protein freezing and all the > kinetic energy > going into translational motion. > > there is a few things you could try. > 1- First you should increase the rlilst to at least 1.3 but 1.4 would be > better. > 2- And/or reduce the nstlist to 5. > > You should modify at least one of the two, that will increases the cost > of your simulation > but should make increase energy conservation. > > Then you should try a smaller time step 0.020 ps might just remove all the > lincs warnings. > > XAvier. > > I made all the modifications suggested, but the system still crashed (with segmentation fault). However, 'elastic bonds' representation that Martti suggested works and I may need these modifications later. Thanks for being so patient with my problem. > > > On Fri, Apr 16, 2010 at 5:17 PM, Martti Louhivuori > wrote: > >> On 15 Apr 2010, at 18:06, Trang wrote: >> >>> My target system is a protein with lipid molecules added randomly (using >>> GENBOX). Running MD, I expect to >>> >> >> I hope you're using a larger van der Waals distance (0.24nm or so) when >> inserting the lipids. > > > I tried 0.3 - 0.5. It looked fine, but didn't work. > >> >> >> broke down the problem, that is, to run md simulation for the protein >>> molecule only, and in vacuum. Still no improvement. Although all the >>> distances in the minimized structure are visually proper, the system >>> exploded. >>> >> >> If you can't run the protein even in vacuum, then the problem is either in >> your MD parameters, starting co-ordinates or some simple mistake somewhere. >> Since the .mdp files you posted seem ok, my bet is on the starting >> co-ordinates, just as Xavier proposed. >> >> Could you post the system topology (.top) and the protein topology (.itp), >> so we can rule out any mistakes in those? > > > The protein topo is too large, so I put it here. http://pastie.org/926617 > Here is the system topology. > > > #include "../martini_v2.1.itp" > #include "../martini_v2.1_aminoacids.itp" > #include "11BHSD1.itp" > > ; Include Position restraint file > #ifdef POSRES > #include "posre.itp" > #endif > > > [ system ] > 11BHSD1 > > [ molecules ] > 11BHSD1 1 > > > > >> You should also double check whether you have close contacts between some >> atoms in the protein; e.g. in VMD this is easily done using the "dynamic >> bonds" representation. > > >> -martti- >> > > Minimized structure showed no close contact (with distance cutoff 1.6, even > to 1.9). Production run stopped at step 175552 (5266.56 ps) with > Segmentation fault, preceeded with a lot of Lincs warnings (I set ring_bonds > = "constraints"). This is the mdp file, this file goes along with the > system in vaccuum that crashed. I'm not sure if too long simulation time can > be the cause. The strange thing is that the system seemed to be relatively > stable at the time of crash. rmds.xvg is below, in case you want to see the > graph first-hand. > > -MD--- > integrator = md > tinit= 0.0 > dt = 0.030 > nsteps = 90 > nstcomm = 1 > comm-grps= > nstxout = 5000 > nstvout = 5000 > nstfout = 0 > nstlog = 2000 > nstenergy= 2000 > nstxtcout= 1000 > xtc_precision= 100 > xtc-grps = > energygrps = > nstlist = 10 > ns_type = grid > pbc = xyz > rlist= 1.2 > coulombtype = Shift > rcoulomb_switch = 0.0 > rcoulomb = 1.2 > epsilon_r= 15 > vdw_type = Shift > rvdw_switch = 0.9 > rvdw = 1.2 > DispCorr = No > tcoupl = Berendsen > tc-grps = Protein > tau_t= 0.3 > ref_t= 323 > Pcoupl = berendsen > Pcoupltype = isotropic > tau_p= 3.0 > compressibility = 3e-5 > ref_p= 1.0 > gen_vel = no > gen_temp = 323 > gen_seed = 666 > constraints = none > constraint_algorithm = Lincs > unconstrained_start = no > lincs_order = 4 > lincs_warnangle = 30 > -- > > -rmsd.xvg--- > @title "RMSD" > @xaxis label "Time (ps)" > @yaxis label "RMSD (nm)" > @TYPE xy > @ subtitle "Protein after lsq fit to Protein" >0.0000.0001400 > 15
Re: [gmx-users] CG-MD simulation of protein, always crash with protein
On 19 Apr 2010, at 06:49, Trang wrote: The protein topo is too large, so I put it here. http://pastie.org/926617 The protein topology is fine, but you could try also with elastic bonds instead of dihedrals in the extended regions (--elastic option). Elastic bonds are more robust than dihedrals, especially if running with a larger timestep (30fs). Here is the system topology. #include "../martini_v2.1.itp" #include "../martini_v2.1_aminoacids.itp" The last line is unnecessary. Use it only when simulating single aminoacids instead of a polypeptide. Minimized structure showed no close contact (with distance cutoff 1.6, even to 1.9). Production run stopped at That's only 0.16-0.19nm, so overlap is still possible. step 175552 (5266.56 ps) with Segmentation fault, preceeded with a lot of Lincs warnings (I set ring_bonds = "constraints"). This is the mdp file, this file goes along with the system in vaccuum that crashed. I'm not sure if What's different this time around? Last time you said you couldn't run it even in vacuum. Sounds like you are just hitting statistical limits with your combination of (extended) dihedrals + 30fs timestep. Use a smaller timestep (20fs for example) or switch to local elastic bonds for the beta-strands... and solvate the system (without overlap) so the simulation makes some sense! -martti- -- Post-doctoral research fellow Moleculaire Dynamica University of Groningen Nijenborgh 4, 9747AG Groningen, the Netherlands tel. +(31) 50 363 4339 | fax. +(31) 50 363 4398 -- gmx-users mailing listgmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/mailing_lists/users.php
Re: [gmx-users] CG-MD simulation of protein, always crash with protein
Hi, the rmsd you show is for the protein in vacuum? It goes up to 0.7 nm! That is a very large deformation, is it expected? Isn't your protein freezing and all the kinetic energy going into translational motion. there is a few things you could try. 1- First you should increase the rlilst to at least 1.3 but 1.4 would be better. 2- And/or reduce the nstlist to 5. You should modify at least one of the two, that will increases the cost of your simulation but should make increase energy conservation. Then you should try a smaller time step 0.020 ps might just remove all the lincs warnings. XAvier. On Fri, Apr 16, 2010 at 5:17 PM, Martti Louhivuori > wrote: On 15 Apr 2010, at 18:06, Trang wrote: My target system is a protein with lipid molecules added randomly (using GENBOX). Running MD, I expect to I hope you're using a larger van der Waals distance (0.24nm or so) when inserting the lipids. I tried 0.3 - 0.5. It looked fine, but didn't work. broke down the problem, that is, to run md simulation for the protein molecule only, and in vacuum. Still no improvement. Although all the distances in the minimized structure are visually proper, the system exploded. If you can't run the protein even in vacuum, then the problem is either in your MD parameters, starting co-ordinates or some simple mistake somewhere. Since the .mdp files you posted seem ok, my bet is on the starting co-ordinates, just as Xavier proposed. Could you post the system topology (.top) and the protein topology (.itp), so we can rule out any mistakes in those? The protein topo is too large, so I put it here. http://pastie.org/926617 Here is the system topology. #include "../martini_v2.1.itp" #include "../martini_v2.1_aminoacids.itp" #include "11BHSD1.itp" ; Include Position restraint file #ifdef POSRES #include "posre.itp" #endif [ system ] 11BHSD1 [ molecules ] 11BHSD1 1 You should also double check whether you have close contacts between some atoms in the protein; e.g. in VMD this is easily done using the "dynamic bonds" representation. -martti- Minimized structure showed no close contact (with distance cutoff 1.6, even to 1.9). Production run stopped at step 175552 (5266.56 ps) with Segmentation fault, preceeded with a lot of Lincs warnings (I set ring_bonds = "constraints"). This is the mdp file, this file goes along with the system in vaccuum that crashed. I'm not sure if too long simulation time can be the cause. The strange thing is that the system seemed to be relatively stable at the time of crash. rmds.xvg is below, in case you want to see the graph first-hand. -MD--- integrator = md tinit= 0.0 dt = 0.030 nsteps = 90 nstcomm = 1 comm-grps= nstxout = 5000 nstvout = 5000 nstfout = 0 nstlog = 2000 nstenergy= 2000 nstxtcout= 1000 xtc_precision= 100 xtc-grps = energygrps = nstlist = 10 ns_type = grid pbc = xyz rlist= 1.2 coulombtype = Shift rcoulomb_switch = 0.0 rcoulomb = 1.2 epsilon_r= 15 vdw_type = Shift rvdw_switch = 0.9 rvdw = 1.2 DispCorr = No tcoupl = Berendsen tc-grps = Protein tau_t= 0.3 ref_t= 323 Pcoupl = berendsen Pcoupltype = isotropic tau_p= 3.0 compressibility = 3e-5 ref_p= 1.0 gen_vel = no gen_temp = 323 gen_seed = 666 constraints = none constraint_algorithm = Lincs unconstrained_start = no lincs_order = 4 lincs_warnangle = 30 -- -rmsd.xvg--- @title "RMSD" @xaxis label "Time (ps)" @yaxis label "RMSD (nm)" @TYPE xy @ subtitle "Protein after lsq fit to Protein" 0.0000.0001400 150.0000.6150032 300.0000.6162945 450.0000.6451609 600.0000.6317724 750.0000.6501579 900.0000.6642945 1050.0000.6742513 1200.0000.6651280 1350.0000.6965100 1500.0000.6856629 1650.0000.7271055 1800.0000.7174531 1950.0000.7088170 2100.0000.7492073 2250.0000.7352809 2400.0000.7196461 2550.0000.7162255 2700.0000.7277457 2850.0000.7121301 3000.0000.7084662 3150.0000.7153199 3300.0000.72
Re: [gmx-users] CG-MD simulation of protein, always crash with protein
On Fri, Apr 16, 2010 at 5:17 PM, Martti Louhivuori wrote: > On 15 Apr 2010, at 18:06, Trang wrote: > >> My target system is a protein with lipid molecules added randomly (using >> GENBOX). Running MD, I expect to >> > > I hope you're using a larger van der Waals distance (0.24nm or so) when > inserting the lipids. I tried 0.3 - 0.5. It looked fine, but didn't work. > > > broke down the problem, that is, to run md simulation for the protein >> molecule only, and in vacuum. Still no improvement. Although all the >> distances in the minimized structure are visually proper, the system >> exploded. >> > > If you can't run the protein even in vacuum, then the problem is either in > your MD parameters, starting co-ordinates or some simple mistake somewhere. > Since the .mdp files you posted seem ok, my bet is on the starting > co-ordinates, just as Xavier proposed. > > Could you post the system topology (.top) and the protein topology (.itp), > so we can rule out any mistakes in those? The protein topo is too large, so I put it here. http://pastie.org/926617 Here is the system topology. #include "../martini_v2.1.itp" #include "../martini_v2.1_aminoacids.itp" #include "11BHSD1.itp" ; Include Position restraint file #ifdef POSRES #include "posre.itp" #endif [ system ] 11BHSD1 [ molecules ] 11BHSD1 1 > You should also double check whether you have close contacts between some > atoms in the protein; e.g. in VMD this is easily done using the "dynamic > bonds" representation. > -martti- > Minimized structure showed no close contact (with distance cutoff 1.6, even to 1.9). Production run stopped at step 175552 (5266.56 ps) with Segmentation fault, preceeded with a lot of Lincs warnings (I set ring_bonds = "constraints"). This is the mdp file, this file goes along with the system in vaccuum that crashed. I'm not sure if too long simulation time can be the cause. The strange thing is that the system seemed to be relatively stable at the time of crash. rmds.xvg is below, in case you want to see the graph first-hand. -MD--- integrator = md tinit= 0.0 dt = 0.030 nsteps = 90 nstcomm = 1 comm-grps= nstxout = 5000 nstvout = 5000 nstfout = 0 nstlog = 2000 nstenergy= 2000 nstxtcout= 1000 xtc_precision= 100 xtc-grps = energygrps = nstlist = 10 ns_type = grid pbc = xyz rlist= 1.2 coulombtype = Shift rcoulomb_switch = 0.0 rcoulomb = 1.2 epsilon_r= 15 vdw_type = Shift rvdw_switch = 0.9 rvdw = 1.2 DispCorr = No tcoupl = Berendsen tc-grps = Protein tau_t= 0.3 ref_t= 323 Pcoupl = berendsen Pcoupltype = isotropic tau_p= 3.0 compressibility = 3e-5 ref_p= 1.0 gen_vel = no gen_temp = 323 gen_seed = 666 constraints = none constraint_algorithm = Lincs unconstrained_start = no lincs_order = 4 lincs_warnangle = 30 -- -rmsd.xvg--- @title "RMSD" @xaxis label "Time (ps)" @yaxis label "RMSD (nm)" @TYPE xy @ subtitle "Protein after lsq fit to Protein" 0.0000.0001400 150.0000.6150032 300.0000.6162945 450.0000.6451609 600.0000.6317724 750.0000.6501579 900.0000.6642945 1050.0000.6742513 1200.0000.6651280 1350.0000.6965100 1500.0000.6856629 1650.0000.7271055 1800.0000.7174531 1950.0000.7088170 2100.0000.7492073 2250.0000.7352809 2400.0000.7196461 2550.0000.7162255 2700.0000.7277457 2850.0000.7121301 3000.0000.7084662 3150.0000.7153199 3300.0000.7209989 3450.0000.7342823 3600.0000.7297466 3750.0000.7254455 3900.0000.7175814 4050.0000.7134616 4200.0000.7212958 4350.0000.7258847 4500.0000.7201833 4650.0000.7286560 4800.0000.7188574 4950.0000.715 5100.0000.7247230 5250.0000.7153631 --- -- gmx-users mailing listgmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)su
Re: [gmx-users] CG-MD simulation of protein, always crash with protein
On 15 Apr 2010, at 18:06, Trang wrote: My target system is a protein with lipid molecules added randomly (using GENBOX). Running MD, I expect to I hope you're using a larger van der Waals distance (0.24nm or so) when inserting the lipids. broke down the problem, that is, to run md simulation for the protein molecule only, and in vacuum. Still no improvement. Although all the distances in the minimized structure are visually proper, the system exploded. If you can't run the protein even in vacuum, then the problem is either in your MD parameters, starting co-ordinates or some simple mistake somewhere. Since the .mdp files you posted seem ok, my bet is on the starting co-ordinates, just as Xavier proposed. Could you post the system topology (.top) and the protein topology (.itp), so we can rule out any mistakes in those? You should also double check whether you have close contacts between some atoms in the protein; e.g. in VMD this is easily done using the "dynamic bonds" representation. -martti- -- Post-doctoral research fellow Moleculaire Dynamica University of Groningen Nijenborgh 4, 9747AG Groningen, the Netherlands tel. +(31) 50 363 4339 | fax. +(31) 50 363 4398 -- gmx-users mailing listgmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/mailing_lists/users.php
Re: [gmx-users] CG-MD simulation of protein, always crash with protein
My target system is a protein with lipid molecules added randomly (using
GENBOX). Running MD, I expect to have the protein "inserted" through
self-assembly. However, I encountered the crash every different trials. So I
broke down the problem, that is, to run md simulation for the protein
molecule only, and in vacuum. Still no improvement. Although all the
distances in the minimized structure are visually proper, the system
exploded.
When I set RING_BONDS="bonds", the system crashed at the very first step.
With RING_BONDS="constraints", it exploded at about 5800 ps. It is strange
that RMSD differences at that time has been relatively constant. with such
RMSD, I think the system must be stable enough?!?!
I built the topology using the script seq2itp provided at Martini's site.
Already did some random check to see if the script works fine, special
notice was made at amino acids with ring. Should I make a more thorough
check on the itp file?
Trang
On Thu, Apr 15, 2010 at 6:32 PM, XAvier Periole wrote:
>
> It is mot likely that your preparation of the system is somehow corrupted.
> The insertion of a protein in a lipid bilayer might easily introduce strong
> forces in the system and thereby result in a crash of your simulation.
>
> It is also possible that your protein topology is not describing your
> system
> correctly. This no one can tell without knowing how you built your topology
> and for which kind of system.
>
> Note that cycle in the Martini FF (Tyr/Trp/His) use constrains and are
> therefore extremely sensible to strong forces in there environment.
>
> I would suggest you get back to the point where you insert your protein
> in the bilayer and check very carefully potential bad contact between the
> protein and the lipid. The minimization should go smooth and running
> should not be problematic.
>
> XAvier.
>
> On Apr 15, 2010, at 12:49 PM, Trang wrote:
>
> Dear gmx users,
> I'm trying to run some CG-MD with gromacs, using the available Martini FF.
> The first run with {lipid + water} is fine. The POPC lipid bilayer is
> successfully self-assembled. I then applied Martini FF for system involving
> protein.
> All the systems I've tried so far could not survive an md run for long
> enough. Most of them crashed at the very first step.
>
> I've encountered different types of errors (range checking error,
> segmentation fault...) and even hanging of the system, all of which converge
> at the same suggestion that the system hasn't been minimized/equilibrated
> well enough.
>
> I've tried some tracing back but still cannot figure out where the problem
> can be.
>
> * From screen output, I traced back to the atoms on which large force
> affect: most of the cases, these atoms belong to the ring of amino acid such
> as TRP, TYR. The worst case was that, 2 beads of 1 TRP totally overlap after
> minimization!!! The newest trial is related to atom 195 - 196 - 197, which
> are beads of the same TYR
> residue.
> What kind of problem can cause such a bad minimization?
>
> * From log file after crashed runs, I found the initial temperature is
> extremely high (e.g 6.67074e+12 K, etc). A colleague of mine suggest that it
> may be due to the high pressure, which in turn caused by unevenly
> distributed molecules in the input structure. This seemed to be reasonable,
> for after minimization
> (picture),
> the distribution of water is even worse than the initial state (
> picture).
> But, I get stuck and have no idea why minimization can make system so bad.
> This is the mdp for the minimization step
>
> ---MINI---
> title= Martini
> cpp = /usr/bin/cpp
> integrator = steep
> tinit= 0.0
> dt = 0.025
> nsteps = 500
> nstcomm = 1
> comm-grps=
> nstxout = 5000
> nstvout = 5000
> nstfout = 0
> nstlog = 1000
> nstenergy= 1000
> nstxtcout= 1000
> xtc_precision= 100
> xtc-grps =
> energygrps =
> nstlist = 10
> ns_type = grid
> pbc = xyz
> rlist= 1.2
> coulombtype = Shift
> rcoulomb_switch = 0.0
> rcoulomb = 1.2
> epsilon_r= 15
> vdw_type = Shift
> rvdw_switch = 0.9
> rvdw = 1.2
> DispCorr = No
> tcoupl = no
> Pcoupl = no
> gen_vel = no
> gen_temp = 320
> gen_seed = 473529
Re: [gmx-users] CG-MD simulation of protein, always crash with protein
It is mot likely that your preparation of the system is somehow
corrupted.
The insertion of a protein in a lipid bilayer might easily introduce
strong
forces in the system and thereby result in a crash of your simulation.
It is also possible that your protein topology is not describing your
system
correctly. This no one can tell without knowing how you built your
topology
and for which kind of system.
Note that cycle in the Martini FF (Tyr/Trp/His) use constrains and are
therefore extremely sensible to strong forces in there environment.
I would suggest you get back to the point where you insert your protein
in the bilayer and check very carefully potential bad contact between
the
protein and the lipid. The minimization should go smooth and running
should not be problematic.
XAvier.
On Apr 15, 2010, at 12:49 PM, Trang wrote:
Dear gmx users,
I'm trying to run some CG-MD with gromacs, using the available
Martini FF.
The first run with {lipid + water} is fine. The POPC lipid bilayer
is successfully self-assembled. I then applied Martini FF for system
involving protein.
All the systems I've tried so far could not survive an md run for
long enough. Most of them crashed at the very first step.
I've encountered different types of errors (range checking error,
segmentation fault...) and even hanging of the system, all of which
converge at the same suggestion that the system hasn't been
minimized/equilibrated well enough.
I've tried some tracing back but still cannot figure out where the
problem can be.
* From screen output, I traced back to the atoms on which large
force affect: most of the cases, these atoms belong to the ring of
amino acid such as TRP, TYR. The worst case was that, 2 beads of 1
TRP totally overlap after minimization!!! The newest trial is
related to atom 195 - 196 - 197, which are beads of the same TYR
residue . What kind of problem can cause such a bad minimization?
* From log file after crashed runs, I found the initial temperature
is extremely high (e.g 6.67074e+12 K, etc). A colleague of mine
suggest that it may be due to the high pressure, which in turn
caused by unevenly distributed molecules in the input structure.
This seemed to be reasonable, for after minimization (picture ), the
distribution of water is even worse than the initial state
(picture). But, I get stuck and have no idea why minimization can
make system so bad. This is the mdp for the minimization step
---
MINI---
title= Martini
cpp = /usr/bin/cpp
integrator = steep
tinit= 0.0
dt = 0.025
nsteps = 500
nstcomm = 1
comm-grps=
nstxout = 5000
nstvout = 5000
nstfout = 0
nstlog = 1000
nstenergy= 1000
nstxtcout= 1000
xtc_precision= 100
xtc-grps =
energygrps =
nstlist = 10
ns_type = grid
pbc = xyz
rlist= 1.2
coulombtype = Shift
rcoulomb_switch = 0.0
rcoulomb = 1.2
epsilon_r= 15
vdw_type = Shift
rvdw_switch = 0.9
rvdw = 1.2
DispCorr = No
tcoupl = no
Pcoupl = no
gen_vel = no
gen_temp = 320
gen_seed = 473529
constraints = none
constraint_algorithm = Lincs
unconstrained_start = no
lincs_order = 4
lincs_warnangle = 30
and this is for the next step md
MD--
title= Martini
cpp = /usr/bin/cpp
integrator = md
tinit= 0.0
dt = 0.025
nsteps = 5
nstcomm = 1
comm-grps=
nstxout = 5000
nstvout = 5000
nstfout = 0
nstlog = 2000
nstenergy= 2000
nstxtcout= 1000
xtc_precision= 100
xtc-grps =
energygrps =
nstlist = 10
ns_type = grid
pbc = xyz
rlist= 1.2
coulombtype = Shift
rcoulomb_switch = 0.0
rcoulomb = 1.2
epsilon_r= 15
vdw_type = Shift
rvdw_switch = 0.9
rvdw = 1.2
DispCorr = No
tcoupl = Berendsen
tc-grps = Protein W
tau_t= 0.3 0.3
r
[gmx-users] CG-MD simulation of protein, always crash with protein
Dear gmx users,
I'm trying to run some CG-MD with gromacs, using the available Martini FF.
The first run with {lipid + water} is fine. The POPC lipid bilayer is
successfully self-assembled. I then applied Martini FF for system involving
protein.
All the systems I've tried so far could not survive an md run for long
enough. Most of them crashed at the very first step.
I've encountered different types of errors (range checking error,
segmentation fault...) and even hanging of the system, all of which converge
at the same suggestion that the system hasn't been minimized/equilibrated
well enough.
I've tried some tracing back but still cannot figure out where the problem
can be.
* From screen output, I traced back to the atoms on which large force
affect: most of the cases, these atoms belong to the ring of amino acid such
as TRP, TYR. The worst case was that, 2 beads of 1 TRP totally overlap after
minimization!!! The newest trial is related to atom 195 - 196 - 197, which
are beads of the same TYR
residue.
What kind of problem can cause such a bad minimization?
* From log file after crashed runs, I found the initial temperature is
extremely high (e.g 6.67074e+12 K, etc). A colleague of mine suggest that it
may be due to the high pressure, which in turn caused by unevenly
distributed molecules in the input structure. This seemed to be reasonable,
for after minimization
(picture),
the distribution of water is even worse than the initial state (
picture).
But, I get stuck and have no idea why minimization can make system so bad.
This is the mdp for the minimization step
---MINI---
title= Martini
cpp = /usr/bin/cpp
integrator = steep
tinit= 0.0
dt = 0.025
nsteps = 500
nstcomm = 1
comm-grps=
nstxout = 5000
nstvout = 5000
nstfout = 0
nstlog = 1000
nstenergy= 1000
nstxtcout= 1000
xtc_precision= 100
xtc-grps =
energygrps =
nstlist = 10
ns_type = grid
pbc = xyz
rlist= 1.2
coulombtype = Shift
rcoulomb_switch = 0.0
rcoulomb = 1.2
epsilon_r= 15
vdw_type = Shift
rvdw_switch = 0.9
rvdw = 1.2
DispCorr = No
tcoupl = no
Pcoupl = no
gen_vel = no
gen_temp = 320
gen_seed = 473529
constraints = none
constraint_algorithm = Lincs
unconstrained_start = no
lincs_order = 4
lincs_warnangle = 30
and this is for the next step md
MD--
title= Martini
cpp = /usr/bin/cpp
integrator = md
tinit= 0.0
dt = 0.025
nsteps = 5
nstcomm = 1
comm-grps=
nstxout = 5000
nstvout = 5000
nstfout = 0
nstlog = 2000
nstenergy= 2000
nstxtcout= 1000
xtc_precision= 100
xtc-grps =
energygrps =
nstlist = 10
ns_type = grid
pbc = xyz
rlist= 1.2
coulombtype = Shift
rcoulomb_switch = 0.0
rcoulomb = 1.2
epsilon_r= 15
vdw_type = Shift
rvdw_switch = 0.9
rvdw = 1.2
DispCorr = No
tcoupl = Berendsen
tc-grps = Protein W
tau_t= 0.3 0.3
ref_t= 323 323
Pcoupl = berendsen
Pcoupltype = isotropic
tau_p= 3.0
compressibility = 3e-5
ref_p= 1.0
gen_vel = no
gen_temp = 323
gen_seed = 666
constraints = none
constraint_algorithm = Lincs
unconstrained_start = no
lincs_order = 4
lincs_warnangle = 30
I've struggled with this for about 2 weeks and still have no idea of where
to fix the problem. Am I missing something too basic for tutorials to
