Re: [gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center
So lets say that I delete the first frame from the trajectory in which some atoms might have been outside the box. Everything should be within the box once the simulation starts (from the second frame onwards)? So the procedure should work if the reference structure is the second frame? I have tried that, and it fails as well. On Thu, Nov 5, 2009 at 4:40 PM, XAvier Periole x.peri...@rug.nl wrote: The nojump option will not apply the pbc when an atom is crossing the box boundaries ... in your case your bilayer should definitely be in the center of your box and all the atoms in If not ot course it does not work! On Nov 5, 2009, at 4:33 PM, maria goranovic wrote: my starting structure looks quite all right to me. everything is in the box (except the tails of some lipids) .. wonder whats wrong. thank you verymuch for helping On Thu, Nov 5, 2009 at 4:04 PM, XAvier Periole x.peri...@rug.nl wrote: On Nov 5, 2009, at 4:00 PM, maria goranovic wrote: Hi Xavier, Thanks for the clear instructions. The bilayer is not in one piece in the z direction after the -pbc nojump for some reason. the problem might be from your starting structure, everything should be in the box! Or you may be facing strange/funny/incomprehensible behavior ... after the third step, the water is in the right place, but the bilayer has expanded to periodic boxes in the xy plane. so the center of mass of the lipid molecules is not really being centered in the box ? On Thu, Nov 5, 2009 at 3:34 PM, XAvier Periole x.peri...@rug.nl wrote: you need to do: 1- trjconv -pbc nojump; this keeps your bilayer in one piece on the z direction 2- trjconv -center; using the bilayer to center and the system as output; this will translate your bilayer on the z axis and normally not modify it on the xy plan. 3- trjconv -pbc mol; will put your lipids in one piece in the box; I believe this step cn be coupled to the previous quite safely. On Nov 5, 2009, at 3:23 PM, maria goranovic wrote: One more note about -pbc nojump. I typically use -pbc mol. Using pbc nojump succeeds in keeping the center of the bilayer at 0 0 0, but the atoms have moved way out of the simulation box resulting in a dilute system On Thu, Nov 5, 2009 at 2:31 PM, maria goranovic mariagorano...@gmail.com wrote: Centering on one atom has a problem that the lipid diffuses in the plane of the membrane, and as a result, the entire system starts to center around the lipid resulting in a simulation box which translates a lot in the bilayer plane. The splitting is not a problem, yes. But during the simulation period when the bilayer is not split, it diffuses quite a bit along the bilayer normal (after use of -pbc mol, and centering around the lipid center of mass). a plot of the lipid center of mass shows the bilayer diffusing along z, when its not split. On Thu, Nov 5, 2009 at 2:25 PM, XAvier Periole x.peri...@rug.nlwrote: On Nov 5, 2009, at 2:02 PM, Justin A. Lemkul wrote: maria goranovic wrote: I did use -pbc nojump, but that does not help What about entering on a central lipid tail atom, I suggested some time ago? The bilayer probably just splits across periodic boundaries, so this is not really a problem; just a visualization artefact. The splitting is not a problem and I think that centering using one lipid (tail) won't change the problem if the bilayer is cut half! Or the -pbc mol should be applied ... -Justin The drift is about 1 nm per 10 microseconds . (this is a martini simulation) On Thu, Nov 5, 2009 at 1:02 PM, XAvier Periole x.peri...@rug.nlmailto: x.peri...@rug.nl wrote: On Nov 5, 2009, at 12:09 PM, maria goranovic wrote: Hello All (and especially Berk) This is an update of the problem that I was facing earlier. I used to tau_p of 3.0 ps, and the problem does not go away, the bilayers still drifts in the simulation box. So this is probably a bug then? How much is the drift (nm/ns)? Did you use removal of center of mass of the entire system of bilayer/solvent separately? I still cannot understand how to put the bilayer back into the center of the simulation box. As suggested by Justin, I tried to use just one tail atom of a lipid for centering, but that did not work either. I noticed that my bilayer, which is initially at the center of the simulation box, separates into two leaflets at the box edges from the very first step of the simulation itself, but i am not able to correct that using the -center and -boxcenter zero options. Can someone please make a suggestion and help? You have to do use -pbc nojump first and then center ... Thank you so much -Maria -- Maria G. Technical University of Denmark Copenhagen ___ gmx-users mailing listgmx-users@gromacs.org
Re: [gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center
Lets make it clear. I can not tell you if the atoms in the second frame are in the box or not! You have to visualize it! Honestly it is not in the first frame I can not see how it would in the second! You have to build a reference structure that has the bilayer in one piece, then the nojump option can actually to the job you want. On Nov 6, 2009, at 10:37 AM, maria goranovic wrote: So lets say that I delete the first frame from the trajectory in which some atoms might have been outside the box. Everything should be within the box once the simulation starts (from the second frame onwards)? So the procedure should work if the reference structure is the second frame? I have tried that, and it fails as well. On Thu, Nov 5, 2009 at 4:40 PM, XAvier Periole x.peri...@rug.nl wrote: The nojump option will not apply the pbc when an atom is crossing the box boundaries ... in your case your bilayer should definitely be in the center of your box and all the atoms in If not ot course it does not work! On Nov 5, 2009, at 4:33 PM, maria goranovic wrote: my starting structure looks quite all right to me. everything is in the box (except the tails of some lipids) .. wonder whats wrong. thank you verymuch for helping On Thu, Nov 5, 2009 at 4:04 PM, XAvier Periole x.peri...@rug.nl wrote: On Nov 5, 2009, at 4:00 PM, maria goranovic wrote: Hi Xavier, Thanks for the clear instructions. The bilayer is not in one piece in the z direction after the -pbc nojump for some reason. the problem might be from your starting structure, everything should be in the box! Or you may be facing strange/funny/incomprehensible behavior ... after the third step, the water is in the right place, but the bilayer has expanded to periodic boxes in the xy plane. so the center of mass of the lipid molecules is not really being centered in the box ? On Thu, Nov 5, 2009 at 3:34 PM, XAvier Periole x.peri...@rug.nl wrote: you need to do: 1- trjconv -pbc nojump; this keeps your bilayer in one piece on the z direction 2- trjconv -center; using the bilayer to center and the system as output; this will translate your bilayer on the z axis and normally not modify it on the xy plan. 3- trjconv -pbc mol; will put your lipids in one piece in the box; I believe this step cn be coupled to the previous quite safely. On Nov 5, 2009, at 3:23 PM, maria goranovic wrote: One more note about -pbc nojump. I typically use -pbc mol. Using pbc nojump succeeds in keeping the center of the bilayer at 0 0 0, but the atoms have moved way out of the simulation box resulting in a dilute system On Thu, Nov 5, 2009 at 2:31 PM, maria goranovic mariagorano...@gmail.com wrote: Centering on one atom has a problem that the lipid diffuses in the plane of the membrane, and as a result, the entire system starts to center around the lipid resulting in a simulation box which translates a lot in the bilayer plane. The splitting is not a problem, yes. But during the simulation period when the bilayer is not split, it diffuses quite a bit along the bilayer normal (after use of -pbc mol, and centering around the lipid center of mass). a plot of the lipid center of mass shows the bilayer diffusing along z, when its not split. On Thu, Nov 5, 2009 at 2:25 PM, XAvier Periole x.peri...@rug.nl wrote: On Nov 5, 2009, at 2:02 PM, Justin A. Lemkul wrote: maria goranovic wrote: I did use -pbc nojump, but that does not help What about entering on a central lipid tail atom, I suggested some time ago? The bilayer probably just splits across periodic boundaries, so this is not really a problem; just a visualization artefact. The splitting is not a problem and I think that centering using one lipid (tail) won't change the problem if the bilayer is cut half! Or the -pbc mol should be applied ... -Justin The drift is about 1 nm per 10 microseconds . (this is a martini simulation) On Thu, Nov 5, 2009 at 1:02 PM, XAvier Periole x.peri...@rug.nl mailto:x.peri...@rug.nl wrote: On Nov 5, 2009, at 12:09 PM, maria goranovic wrote: Hello All (and especially Berk) This is an update of the problem that I was facing earlier. I used to tau_p of 3.0 ps, and the problem does not go away, the bilayers still drifts in the simulation box. So this is probably a bug then? How much is the drift (nm/ns)? Did you use removal of center of mass of the entire system of bilayer/solvent separately? I still cannot understand how to put the bilayer back into the center of the simulation box. As suggested by Justin, I tried to use just one tail atom of a lipid for centering, but that did not work either. I noticed that my bilayer, which is initially at the center of the simulation box, separates into two leaflets at the box edges from the very first step of the simulation itself, but i am not able to
Re: [gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center
For quite a long time I had the feeling that trjconv doesn't resolve all situations. Following the very recent discussion between Roland Schutz and Tsjerk, I'm not sure there is an immediate solution. Ad hoc approaches such as preparation of tpr files from intermediate snapshots were useful for me in some cases, so you can try these. For calculations of distances you can sometimes calculate the shift and apply an a posteriori fix, but that won't work for visualisation and isn't a robust solution. I don't think it has anything to do with the MARTINI FF. Ran. maria goranovic wrote: So lets say that I delete the first frame from the trajectory in which some atoms might have been outside the box. Everything should be within the box once the simulation starts (from the second frame onwards)? So the procedure should work if the reference structure is the second frame? I have tried that, and it fails as well. On Thu, Nov 5, 2009 at 4:40 PM, XAvier Periole x.peri...@rug.nl mailto:x.peri...@rug.nl wrote: The nojump option will not apply the pbc when an atom is crossing the box boundaries ... in your case your bilayer should definitely be in the center of your box and all the atoms in If not ot course it does not work! On Nov 5, 2009, at 4:33 PM, maria goranovic wrote: my starting structure looks quite all right to me. everything is in the box (except the tails of some lipids) .. wonder whats wrong. thank you verymuch for helping On Thu, Nov 5, 2009 at 4:04 PM, XAvier Periole x.peri...@rug.nl mailto:x.peri...@rug.nl wrote: On Nov 5, 2009, at 4:00 PM, maria goranovic wrote: Hi Xavier, Thanks for the clear instructions. The bilayer is not in one piece in the z direction after the -pbc nojump for some reason. the problem might be from your starting structure, everything should be in the box! Or you may be facing strange/funny/incomprehensible behavior ... after the third step, the water is in the right place, but the bilayer has expanded to periodic boxes in the xy plane. so the center of mass of the lipid molecules is not really being centered in the box ? On Thu, Nov 5, 2009 at 3:34 PM, XAvier Periole x.peri...@rug.nl mailto:x.peri...@rug.nl wrote: you need to do: 1- trjconv -pbc nojump; this keeps your bilayer in one piece on the z direction 2- trjconv -center; using the bilayer to center and the system as output; this will translate your bilayer on the z axis and normally not modify it on the xy plan. 3- trjconv -pbc mol; will put your lipids in one piece in the box; I believe this step cn be coupled to the previous quite safely. On Nov 5, 2009, at 3:23 PM, maria goranovic wrote: One more note about -pbc nojump. I typically use -pbc mol. Using pbc nojump succeeds in keeping the center of the bilayer at 0 0 0, but the atoms have moved way out of the simulation box resulting in a dilute system On Thu, Nov 5, 2009 at 2:31 PM, maria goranovic mariagorano...@gmail.com mailto:mariagorano...@gmail.com wrote: Centering on one atom has a problem that the lipid diffuses in the plane of the membrane, and as a result, the entire system starts to center around the lipid resulting in a simulation box which translates a lot in the bilayer plane. The splitting is not a problem, yes. But during the simulation period when the bilayer is not split, it diffuses quite a bit along the bilayer normal (after use of -pbc mol, and centering around the lipid center of mass). a plot of the lipid center of mass shows the bilayer diffusing along z, when its not split. On Thu, Nov 5, 2009 at 2:25 PM, XAvier Periole x.peri...@rug.nl mailto:x.peri...@rug.nl wrote: On Nov 5, 2009, at 2:02 PM, Justin A. Lemkul wrote: maria goranovic wrote: I did use -pbc nojump, but that does not help What about entering on a central lipid tail atom, I suggested some time ago? The bilayer probably just splits across periodic boundaries, so this is not really a problem; just a visualization artefact. The splitting is not a problem and I think that centering using one lipid (tail) won't change
Re: [gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center
Hi Berk, Shorter tau_p? I thought you suggested 5.0 or 10.0 ps ? On Wed, Sep 2, 2009 at 5:28 PM, Berk Hess g...@hotmail.com wrote: Hi, I also just recalled that we have a bug report open since two years already about drift of the COM: http://bugzilla.gromacs.org/show_bug.cgi?id=165 But in that case double precision did not change anything, so that does not seem to be a precision issue. Here tau_p was 1 ps, but up till now we did not manage to find the source of this problem. Thus it would be useful to see if a shorter tau_p fixes it in your case. Berk -- Date: Wed, 2 Sep 2009 17:21:46 +0200 Subject: Re: [gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center From: mariagorano...@gmail.com To: gmx-users@gromacs.org I will change the tau_p values, and report back. This might take more than a week though. maria On Wed, Sep 2, 2009 at 5:16 PM, Berk Hess g...@hotmail.com wrote: It might actually affect the center of mass motion removal, because you would be scaling your system with 1 +- 1 bit at every step. This could produce consistent rounding in one direction in single precision, causing the system to move in one direction. This is something we should check in general. Often people are using too small tau_p values, like 0.5 or 1 ps, so I advise them to use 5 or 10 ps. But if larger values cause problems in single precision we should be aware of this. Could you report back if changing tau_p solves the drifting problem? Berk -- Date: Wed, 2 Sep 2009 17:06:50 +0200 Subject: Re: [gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center From: mariagorano...@gmail.com To: gmx-users@gromacs.org Oh dear. That is not good. the missing decimal point in tau_p it is a typo all right. but it seems i have used it in the simulations too. thank you for noticing, Xavier. that forces redoing a lot of simulations. that said, it should still not impact the center of mass removal anyway? -maria On Wed, Sep 2, 2009 at 4:50 PM, XAvier Periole x.peri...@rug.nl wrote: your second value for tau_p is missing the . is this a typo? On Sep 2, 2009, at 4:45 PM, maria goranovic wrote: Here are the mdp parameters: title= POPC cpp = /usr/bin/cpp integrator = md tinit= 0.0 dt = 0.030 nsteps = 300 nstcomm = 1 comm-grps = Lipid W ; OUTPUT CONTROL OPTIONS = ; Output frequency for coords (x), velocities (v) and forces (f) = nstxout = 3 nstvout = 3 nstfout = 0 nstlog = 3 nstenergy= 3 ns_type = grid nstlist = 10 pbc = xyz rlist= 1.2 ; Method for doing electrostatics = coulombtype = Shift rcoulomb_switch = 0.0 rcoulomb = 1.2 epsilon_r= 15 vdw_type = Shift ; cut-off lengths= rvdw_switch = 0.9 rvdw = 1.2 DispCorr = No ; Temperature coupling = tcoupl = Berendsen tc-grps = Lipid W tau_t= 0.3 0.3 ref_t= 323 323 ; Pressure coupling = Pcoupl = berendsen Pcoupltype = semiisotropic tau_p = 3.030 compressibility = 3e-53e-5 ref_p = 1.01.0 constraints = none constraint_algorithm = Lincs unconstrained_start = no lincs_order = 4 lincs_warnangle = 30 On Wed, Sep 2, 2009 at 4:33 PM, Berk Hess g...@hotmail.com wrote: Hi, I am 99.99% sure that there is no problem with COM motion removal in Gromacs. Could you post your mdp parameters? Berk From: x.peri...@rug.nl To: gmx-users@gromacs.org Subject: Re: [gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center Date: Wed, 2 Sep 2009 16:04:39 +0200 I am not sure how to fix the trajectory that has drifted ... But if your bilayer drifts even if you use a removal of the COM for the water and bilayer separately that means there is problem in the code! And this should be fixed. XAvier. On Sep 2, 2009, at 3:36 PM, maria goranovic wrote: Dear Experts I had posted this earlier, but the problem was not solved by earlier suggestions. So am posting again. I am simulating a POPC bilayer using MARTINI. The simulation ran fine, but the bilayer drifted towards the edge of the box along the bilayer normal, and eventually some of the atoms crossed the box boundaries. In some cases, entire lipid molecules
RE: [gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center
I meant 3 instead of 30 ps. I would say 1 ps is too short for systems with a phase with large molecules___ 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] Martini simulation problem in recentering trajectory so that the bilayer is at the center
Dear Experts I had posted this earlier, but the problem was not solved by earlier suggestions. So am posting again. I am simulating a POPC bilayer using MARTINI. The simulation ran fine, but the bilayer drifted towards the edge of the box along the bilayer normal, and eventually some of the atoms crossed the box boundaries. In some cases, entire lipid molecules crossed the box boundaries. I tried to recenter the trajectory, so that the lipid bilayer would be at the center of the box at all times. But for some reason, this does not seem to work? I have tried simulations using a single comm_group for the entire system, as well as separate ones for the lipid and water, but the same problem appears in either case. Typically, for all-atom bilayers, the following set of commands works to correct the drift: first convert original trajectory to a temp. xtc ### echo 3 0 | trjconv -s *tpr -f original.xtc -o temp.xtc -center -boxcenter zero -pbc mol -n popc.ndx then convert temp.xtc to the final trajecory ### echo 3 0 | trjconv -s k*tpr -f temp.xtc -o final.xtc -center -boxcenter zero -pbc mol -n popc.ndx where groups 3 and 0 are the lipid and the whole system respectively, and final.xtc is my final trajectory. However, this does not work for the MARTINI systems. Looking at the final trajectory in VMD, the bilayer is either at the center of the box, or it is split at the box edges, with each monomer being in different leaflets. If I plot the center of mass motion of the entire system in the original trajectory .. the system seems to drift by ~ 2-3 angstroms in one direction. As a result, water center of mass drifts in the opposite direction (because of PBC). Are there any suggestions to sort this out? One option is to write the entire trajectory to .gro files, recenter all of them (depending upon whether the bilayer is in the center or is split at the box edge), and concatenate the gro files again.but this is tedious, even if scripted. Please let me know if i can provide any additional info ? -- Maria G. Technical University of Denmark Copenhagen ___ 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] Martini simulation problem in recentering trajectory so that the bilayer is at the center
maria goranovic wrote: Dear Experts I had posted this earlier, but the problem was not solved by earlier suggestions. So am posting again. I am simulating a POPC bilayer using MARTINI. The simulation ran fine, but the bilayer drifted towards the edge of the box along the bilayer normal, and eventually some of the atoms crossed the box boundaries. In some cases, entire lipid molecules crossed the box boundaries. I tried to recenter the trajectory, so that the lipid bilayer would be at the center of the box at all times. But for some reason, this does not seem to work? I have tried simulations using a single comm_group for the entire system, as well as separate ones for the lipid and water, but the same problem appears in either case. Typically, for all-atom bilayers, the following set of commands works to correct the drift: first convert original trajectory to a temp. xtc ### echo 3 0 | trjconv -s *tpr -f original.xtc -o temp.xtc -center -boxcenter zero -pbc mol -n popc.ndx then convert temp.xtc to the final trajecory ### echo 3 0 | trjconv -s k*tpr -f temp.xtc -o final.xtc -center -boxcenter zero -pbc mol -n popc.ndx where groups 3 and 0 are the lipid and the whole system respectively, and final.xtc is my final trajectory. However, this does not work for the MARTINI systems. Looking at the final trajectory in VMD, the bilayer is either at the center of the box, or it is split at the box edges, with each monomer being in different leaflets. The -center option will place the center of mass of the group in the center of the box, so having to the two leaflets at the top and bottom of the box still satisfies this criterion. The better approach is to choose a single lipid, or even a tail atom of one lipid, as the group to be centered. If I plot the center of mass motion of the entire system in the original trajectory .. the system seems to drift by ~ 2-3 angstroms in one direction. As a result, water center of mass drifts in the opposite direction (because of PBC). How did you specify COM motion removal in your .mdp file? -Justin Are there any suggestions to sort this out? One option is to write the entire trajectory to .gro files, recenter all of them (depending upon whether the bilayer is in the center or is split at the box edge), and concatenate the gro files again.but this is tedious, even if scripted. Please let me know if i can provide any additional info ? -- Maria G. Technical University of Denmark Copenhagen ___ 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 -- Justin A. Lemkul Ph.D. Candidate ICTAS Doctoral Scholar Department of Biochemistry Virginia Tech Blacksburg, VA jalemkul[at]vt.edu | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin ___ 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] Martini simulation problem in recentering trajectory so that the bilayer is at the center
I am not sure how to fix the trajectory that has drifted ... But if your bilayer drifts even if you use a removal of the COM for the water and bilayer separately that means there is problem in the code! And this should be fixed. XAvier. On Sep 2, 2009, at 3:36 PM, maria goranovic wrote: Dear Experts I had posted this earlier, but the problem was not solved by earlier suggestions. So am posting again. I am simulating a POPC bilayer using MARTINI. The simulation ran fine, but the bilayer drifted towards the edge of the box along the bilayer normal, and eventually some of the atoms crossed the box boundaries. In some cases, entire lipid molecules crossed the box boundaries. I tried to recenter the trajectory, so that the lipid bilayer would be at the center of the box at all times. But for some reason, this does not seem to work? I have tried simulations using a single comm_group for the entire system, as well as separate ones for the lipid and water, but the same problem appears in either case. Typically, for all-atom bilayers, the following set of commands works to correct the drift: first convert original trajectory to a temp. xtc ### echo 3 0 | trjconv -s *tpr -f original.xtc -o temp.xtc -center - boxcenter zero -pbc mol -n popc.ndx then convert temp.xtc to the final trajecory ### echo 3 0 | trjconv -s k*tpr -f temp.xtc -o final.xtc -center - boxcenter zero -pbc mol -n popc.ndx where groups 3 and 0 are the lipid and the whole system respectively, and final.xtc is my final trajectory. However, this does not work for the MARTINI systems. Looking at the final trajectory in VMD, the bilayer is either at the center of the box, or it is split at the box edges, with each monomer being in different leaflets. If I plot the center of mass motion of the entire system in the original trajectory .. the system seems to drift by ~ 2-3 angstroms in one direction. As a result, water center of mass drifts in the opposite direction (because of PBC). Are there any suggestions to sort this out? One option is to write the entire trajectory to .gro files, recenter all of them (depending upon whether the bilayer is in the center or is split at the box edge), and concatenate the gro files again.but this is tedious, even if scripted. Please let me know if i can provide any additional info ? -- Maria G. Technical University of Denmark Copenhagen ___ 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] Martini simulation problem in recentering trajectory so that the bilayer is at the center
I will try centering with one of the lipid tail atoms .. that could solve the problem. This the way I have specified the comm_groups: nstcomm = 1 comm-grps = Lipid W OR nstcom = 1 comm-grps = system -Maria On Wed, Sep 2, 2009 at 4:04 PM, XAvier Periole x.peri...@rug.nl wrote: I am not sure how to fix the trajectory that has drifted ... But if your bilayer drifts even if you use a removal of the COM for the water and bilayer separately that means there is problem in the code! And this should be fixed. XAvier. On Sep 2, 2009, at 3:36 PM, maria goranovic wrote: Dear Experts I had posted this earlier, but the problem was not solved by earlier suggestions. So am posting again. I am simulating a POPC bilayer using MARTINI. The simulation ran fine, but the bilayer drifted towards the edge of the box along the bilayer normal, and eventually some of the atoms crossed the box boundaries. In some cases, entire lipid molecules crossed the box boundaries. I tried to recenter the trajectory, so that the lipid bilayer would be at the center of the box at all times. But for some reason, this does not seem to work? I have tried simulations using a single comm_group for the entire system, as well as separate ones for the lipid and water, but the same problem appears in either case. Typically, for all-atom bilayers, the following set of commands works to correct the drift: first convert original trajectory to a temp. xtc ### echo 3 0 | trjconv -s *tpr -f original.xtc -o temp.xtc -center -boxcenter zero -pbc mol -n popc.ndx then convert temp.xtc to the final trajecory ### echo 3 0 | trjconv -s k*tpr -f temp.xtc -o final.xtc -center -boxcenter zero -pbc mol -n popc.ndx where groups 3 and 0 are the lipid and the whole system respectively, and final.xtc is my final trajectory. However, this does not work for the MARTINI systems. Looking at the final trajectory in VMD, the bilayer is either at the center of the box, or it is split at the box edges, with each monomer being in different leaflets. If I plot the center of mass motion of the entire system in the original trajectory .. the system seems to drift by ~ 2-3 angstroms in one direction. As a result, water center of mass drifts in the opposite direction (because of PBC). Are there any suggestions to sort this out? One option is to write the entire trajectory to .gro files, recenter all of them (depending upon whether the bilayer is in the center or is split at the box edge), and concatenate the gro files again.but this is tedious, even if scripted. Please let me know if i can provide any additional info ? -- Maria G. Technical University of Denmark Copenhagen ___ 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 thewww interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/mailing_lists/users.php -- Maria G. Technical University of Denmark Copenhagen ___ 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] Martini simulation problem in recentering trajectory so that the bilayer is at the center
Hi, I am 99.99% sure that there is no problem with COM motion removal in Gromacs. Could you post your mdp parameters? Berk From: x.peri...@rug.nl To: gmx-users@gromacs.org Subject: Re: [gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center Date: Wed, 2 Sep 2009 16:04:39 +0200 I am not sure how to fix the trajectory that has drifted ... But if your bilayer drifts even if you use a removal of the COM for the water and bilayer separately that means there is problem in the code! And this should be fixed. XAvier. On Sep 2, 2009, at 3:36 PM, maria goranovic wrote: Dear Experts I had posted this earlier, but the problem was not solved by earlier suggestions. So am posting again. I am simulating a POPC bilayer using MARTINI. The simulation ran fine, but the bilayer drifted towards the edge of the box along the bilayer normal, and eventually some of the atoms crossed the box boundaries. In some cases, entire lipid molecules crossed the box boundaries. I tried to recenter the trajectory, so that the lipid bilayer would be at the center of the box at all times. But for some reason, this does not seem to work? I have tried simulations using a single comm_group for the entire system, as well as separate ones for the lipid and water, but the same problem appears in either case. Typically, for all-atom bilayers, the following set of commands works to correct the drift: first convert original trajectory to a temp. xtc ### echo 3 0 | trjconv -s *tpr -f original.xtc -o temp.xtc -center - boxcenter zero -pbc mol -n popc.ndx then convert temp.xtc to the final trajecory ### echo 3 0 | trjconv -s k*tpr -f temp.xtc -o final.xtc -center - boxcenter zero -pbc mol -n popc.ndx where groups 3 and 0 are the lipid and the whole system respectively, and final.xtc is my final trajectory. However, this does not work for the MARTINI systems. Looking at the final trajectory in VMD, the bilayer is either at the center of the box, or it is split at the box edges, with each monomer being in different leaflets. If I plot the center of mass motion of the entire system in the original trajectory .. the system seems to drift by ~ 2-3 angstroms in one direction. As a result, water center of mass drifts in the opposite direction (because of PBC). Are there any suggestions to sort this out? One option is to write the entire trajectory to .gro files, recenter all of them (depending upon whether the bilayer is in the center or is split at the box edge), and concatenate the gro files again.but this is tedious, even if scripted. Please let me know if i can provide any additional info ? -- Maria G. Technical University of Denmark Copenhagen ___ 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 _ Express yourself instantly with MSN Messenger! Download today it's FREE! http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/___ 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] Martini simulation problem in recentering trajectory so that the bilayer is at the center
Here are the mdp parameters: title= POPC cpp = /usr/bin/cpp integrator = md tinit= 0.0 dt = 0.030 nsteps = 300 nstcomm = 1 comm-grps = Lipid W ; OUTPUT CONTROL OPTIONS = ; Output frequency for coords (x), velocities (v) and forces (f) = nstxout = 3 nstvout = 3 nstfout = 0 nstlog = 3 nstenergy= 3 ns_type = grid nstlist = 10 pbc = xyz rlist= 1.2 ; Method for doing electrostatics = coulombtype = Shift rcoulomb_switch = 0.0 rcoulomb = 1.2 epsilon_r= 15 vdw_type = Shift ; cut-off lengths= rvdw_switch = 0.9 rvdw = 1.2 DispCorr = No ; Temperature coupling = tcoupl = Berendsen tc-grps = Lipid W tau_t= 0.3 0.3 ref_t= 323 323 ; Pressure coupling = Pcoupl = berendsen Pcoupltype = semiisotropic tau_p = 3.030 compressibility = 3e-53e-5 ref_p = 1.01.0 constraints = none constraint_algorithm = Lincs unconstrained_start = no lincs_order = 4 lincs_warnangle = 30 On Wed, Sep 2, 2009 at 4:33 PM, Berk Hess g...@hotmail.com wrote: Hi, I am 99.99% sure that there is no problem with COM motion removal in Gromacs. Could you post your mdp parameters? Berk From: x.peri...@rug.nl To: gmx-users@gromacs.org Subject: Re: [gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center Date: Wed, 2 Sep 2009 16:04:39 +0200 I am not sure how to fix the trajectory that has drifted ... But if your bilayer drifts even if you use a removal of the COM for the water and bilayer separately that means there is problem in the code! And this should be fixed. XAvier. On Sep 2, 2009, at 3:36 PM, maria goranovic wrote: Dear Experts I had posted this earlier, but the problem was not solved by earlier suggestions. So am posting again. I am simulating a POPC bilayer using MARTINI. The simulation ran fine, but the bilayer drifted towards the edge of the box along the bilayer normal, and eventually some of the atoms crossed the box boundaries. In some cases, entire lipid molecules crossed the box boundaries. I tried to recenter the trajectory, so that the lipid bilayer would be at the center of the box at all times. But for some reason, this does not seem to work? I have tried simulations using a single comm_group for the entire system, as well as separate ones for the lipid and water, but the same problem appears in either case. Typically, for all-atom bilayers, the following set of commands works to correct the drift: first convert original trajectory to a temp. xtc ### echo 3 0 | trjconv -s *tpr -f original.xtc -o temp.xtc -center - boxcenter zero -pbc mol -n popc.ndx then convert temp.xtc to the final trajecory ### echo 3 0 | trjconv -s k*tpr -f temp.xtc -o final.xtc -center - boxcenter zero -pbc mol -n popc.ndx where groups 3 and 0 are the lipid and the whole system respectively, and final.xtc is my final trajectory. However, this does not work for the MARTINI systems. Looking at the final trajectory in VMD, the bilayer is either at the center of the box, or it is split at the box edges, with each monomer being in different leaflets. If I plot the center of mass motion of the entire system in the original trajectory .. the system seems to drift by ~ 2-3 angstroms in one direction. As a result, water center of mass drifts in the opposite direction (because of PBC). Are there any suggestions to sort this out? One option is to write the entire trajectory to .gro files, recenter all of them (depending upon whether the bilayer is in the center or is split at the box edge), and concatenate the gro files again.but this is tedious, even if scripted. Please let me know if i can provide any additional info ? -- Maria G. Technical University of Denmark Copenhagen ___ gmx-users mailing list gmx-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 list
Re: [gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center
your second value for tau_p is missing the . is this a typo? On Sep 2, 2009, at 4:45 PM, maria goranovic wrote: Here are the mdp parameters: title= POPC cpp = /usr/bin/cpp integrator = md tinit= 0.0 dt = 0.030 nsteps = 300 nstcomm = 1 comm-grps = Lipid W ; OUTPUT CONTROL OPTIONS = ; Output frequency for coords (x), velocities (v) and forces (f) = nstxout = 3 nstvout = 3 nstfout = 0 nstlog = 3 nstenergy= 3 ns_type = grid nstlist = 10 pbc = xyz rlist= 1.2 ; Method for doing electrostatics = coulombtype = Shift rcoulomb_switch = 0.0 rcoulomb = 1.2 epsilon_r= 15 vdw_type = Shift ; cut-off lengths= rvdw_switch = 0.9 rvdw = 1.2 DispCorr = No ; Temperature coupling = tcoupl = Berendsen tc-grps = Lipid W tau_t= 0.3 0.3 ref_t= 323 323 ; Pressure coupling = Pcoupl = berendsen Pcoupltype = semiisotropic tau_p = 3.030 compressibility = 3e-53e-5 ref_p = 1.01.0 constraints = none constraint_algorithm = Lincs unconstrained_start = no lincs_order = 4 lincs_warnangle = 30 On Wed, Sep 2, 2009 at 4:33 PM, Berk Hess g...@hotmail.com wrote: Hi, I am 99.99% sure that there is no problem with COM motion removal in Gromacs. Could you post your mdp parameters? Berk From: x.peri...@rug.nl To: gmx-users@gromacs.org Subject: Re: [gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center Date: Wed, 2 Sep 2009 16:04:39 +0200 I am not sure how to fix the trajectory that has drifted ... But if your bilayer drifts even if you use a removal of the COM for the water and bilayer separately that means there is problem in the code! And this should be fixed. XAvier. On Sep 2, 2009, at 3:36 PM, maria goranovic wrote: Dear Experts I had posted this earlier, but the problem was not solved by earlier suggestions. So am posting again. I am simulating a POPC bilayer using MARTINI. The simulation ran fine, but the bilayer drifted towards the edge of the box along the bilayer normal, and eventually some of the atoms crossed the box boundaries. In some cases, entire lipid molecules crossed the box boundaries. I tried to recenter the trajectory, so that the lipid bilayer would be at the center of the box at all times. But for some reason, this does not seem to work? I have tried simulations using a single comm_group for the entire system, as well as separate ones for the lipid and water, but the same problem appears in either case. Typically, for all-atom bilayers, the following set of commands works to correct the drift: first convert original trajectory to a temp. xtc ### echo 3 0 | trjconv -s *tpr -f original.xtc -o temp.xtc -center - boxcenter zero -pbc mol -n popc.ndx then convert temp.xtc to the final trajecory ### echo 3 0 | trjconv -s k*tpr -f temp.xtc -o final.xtc -center - boxcenter zero -pbc mol -n popc.ndx where groups 3 and 0 are the lipid and the whole system respectively, and final.xtc is my final trajectory. However, this does not work for the MARTINI systems. Looking at the final trajectory in VMD, the bilayer is either at the center of the box, or it is split at the box edges, with each monomer being in different leaflets. If I plot the center of mass motion of the entire system in the original trajectory .. the system seems to drift by ~ 2-3 angstroms in one direction. As a result, water center of mass drifts in the opposite direction (because of PBC). Are there any suggestions to sort this out? One option is to write the entire trajectory to .gro files, recenter all of them (depending upon whether the bilayer is in the center or is split at the box edge), and concatenate the gro files again.but this is tedious, even if scripted. Please let me know if i can provide any additional info ? -- Maria G. Technical University of Denmark Copenhagen ___ gmx-users mailing list gmx-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] Martini simulation problem in recentering trajectory so that the bilayer is at the center
Oh dear. That is not good. the missing decimal point in tau_p it is a typo all right. but it seems i have used it in the simulations too. thank you for noticing, Xavier. that forces redoing a lot of simulations. that said, it should still not impact the center of mass removal anyway? -maria On Wed, Sep 2, 2009 at 4:50 PM, XAvier Periole x.peri...@rug.nl wrote: your second value for tau_p is missing the . is this a typo? On Sep 2, 2009, at 4:45 PM, maria goranovic wrote: Here are the mdp parameters: title= POPC cpp = /usr/bin/cpp integrator = md tinit= 0.0 dt = 0.030 nsteps = 300 nstcomm = 1 comm-grps = Lipid W ; OUTPUT CONTROL OPTIONS = ; Output frequency for coords (x), velocities (v) and forces (f) = nstxout = 3 nstvout = 3 nstfout = 0 nstlog = 3 nstenergy= 3 ns_type = grid nstlist = 10 pbc = xyz rlist= 1.2 ; Method for doing electrostatics = coulombtype = Shift rcoulomb_switch = 0.0 rcoulomb = 1.2 epsilon_r= 15 vdw_type = Shift ; cut-off lengths= rvdw_switch = 0.9 rvdw = 1.2 DispCorr = No ; Temperature coupling = tcoupl = Berendsen tc-grps = Lipid W tau_t= 0.3 0.3 ref_t= 323 323 ; Pressure coupling = Pcoupl = berendsen Pcoupltype = semiisotropic tau_p = 3.030 compressibility = 3e-53e-5 ref_p = 1.01.0 constraints = none constraint_algorithm = Lincs unconstrained_start = no lincs_order = 4 lincs_warnangle = 30 On Wed, Sep 2, 2009 at 4:33 PM, Berk Hess g...@hotmail.com wrote: Hi, I am 99.99% sure that there is no problem with COM motion removal in Gromacs. Could you post your mdp parameters? Berk From: x.peri...@rug.nl To: gmx-users@gromacs.org Subject: Re: [gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center Date: Wed, 2 Sep 2009 16:04:39 +0200 I am not sure how to fix the trajectory that has drifted ... But if your bilayer drifts even if you use a removal of the COM for the water and bilayer separately that means there is problem in the code! And this should be fixed. XAvier. On Sep 2, 2009, at 3:36 PM, maria goranovic wrote: Dear Experts I had posted this earlier, but the problem was not solved by earlier suggestions. So am posting again. I am simulating a POPC bilayer using MARTINI. The simulation ran fine, but the bilayer drifted towards the edge of the box along the bilayer normal, and eventually some of the atoms crossed the box boundaries. In some cases, entire lipid molecules crossed the box boundaries. I tried to recenter the trajectory, so that the lipid bilayer would be at the center of the box at all times. But for some reason, this does not seem to work? I have tried simulations using a single comm_group for the entire system, as well as separate ones for the lipid and water, but the same problem appears in either case. Typically, for all-atom bilayers, the following set of commands works to correct the drift: first convert original trajectory to a temp. xtc ### echo 3 0 | trjconv -s *tpr -f original.xtc -o temp.xtc -center - boxcenter zero -pbc mol -n popc.ndx then convert temp.xtc to the final trajecory ### echo 3 0 | trjconv -s k*tpr -f temp.xtc -o final.xtc -center - boxcenter zero -pbc mol -n popc.ndx where groups 3 and 0 are the lipid and the whole system respectively, and final.xtc is my final trajectory. However, this does not work for the MARTINI systems. Looking at the final trajectory in VMD, the bilayer is either at the center of the box, or it is split at the box edges, with each monomer being in different leaflets. If I plot the center of mass motion of the entire system in the original trajectory .. the system seems to drift by ~ 2-3 angstroms in one direction. As a result, water center of mass drifts in the opposite direction (because of PBC). Are there any suggestions to sort this out? One option is to write the entire trajectory to .gro files, recenter all of them (depending upon whether the bilayer is in the center or is split at the box edge), and concatenate the gro files again.but this is tedious, even if scripted. Please let me know if i can provide any additional info ? -- Maria G. Technical University of Denmark
RE: [gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center
It might actually affect the center of mass motion removal, because you would be scaling your system with 1 +- 1 bit at every step. This could produce consistent rounding in one direction in single precision, causing the system to move in one direction. This is something we should check in general. Often people are using too small tau_p values, like 0.5 or 1 ps, so I advise them to use 5 or 10 ps. But if larger values cause problems in single precision we should be aware of this. Could you report back if changing tau_p solves the drifting problem? Berk Date: Wed, 2 Sep 2009 17:06:50 +0200 Subject: Re: [gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center From: mariagorano...@gmail.com To: gmx-users@gromacs.org Oh dear. That is not good. the missing decimal point in tau_p it is a typo all right. but it seems i have used it in the simulations too. thank you for noticing, Xavier. that forces redoing a lot of simulations. that said, it should still not impact the center of mass removal anyway? -maria On Wed, Sep 2, 2009 at 4:50 PM, XAvier Periole x.peri...@rug.nl wrote: your second value for tau_p is missing the . is this a typo? On Sep 2, 2009, at 4:45 PM, maria goranovic wrote: Here are the mdp parameters: title= POPC cpp = /usr/bin/cpp integrator = md tinit= 0.0 dt = 0.030 nsteps = 300 nstcomm = 1 comm-grps = Lipid W ; OUTPUT CONTROL OPTIONS = ; Output frequency for coords (x), velocities (v) and forces (f) = nstxout = 3 nstvout = 3 nstfout = 0 nstlog = 3 nstenergy= 3 ns_type = grid nstlist = 10 pbc = xyz rlist= 1.2 ; Method for doing electrostatics = coulombtype = Shift rcoulomb_switch = 0.0 rcoulomb = 1.2 epsilon_r= 15 vdw_type = Shift ; cut-off lengths= rvdw_switch = 0.9 rvdw = 1.2 DispCorr = No ; Temperature coupling = tcoupl = Berendsen tc-grps = Lipid W tau_t= 0.3 0.3 ref_t= 323 323 ; Pressure coupling = Pcoupl = berendsen Pcoupltype = semiisotropic tau_p = 3.030 compressibility = 3e-53e-5 ref_p = 1.01.0 constraints = none constraint_algorithm = Lincs unconstrained_start = no lincs_order = 4 lincs_warnangle = 30 On Wed, Sep 2, 2009 at 4:33 PM, Berk Hess g...@hotmail.com wrote: Hi, I am 99.99% sure that there is no problem with COM motion removal in Gromacs. Could you post your mdp parameters? Berk From: x.peri...@rug.nl To: gmx-users@gromacs.org Subject: Re: [gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center Date: Wed, 2 Sep 2009 16:04:39 +0200 I am not sure how to fix the trajectory that has drifted ... But if your bilayer drifts even if you use a removal of the COM for the water and bilayer separately that means there is problem in the code! And this should be fixed. XAvier. On Sep 2, 2009, at 3:36 PM, maria goranovic wrote: Dear Experts I had posted this earlier, but the problem was not solved by earlier suggestions. So am posting again. I am simulating a POPC bilayer using MARTINI. The simulation ran fine, but the bilayer drifted towards the edge of the box along the bilayer normal, and eventually some of the atoms crossed the box boundaries. In some cases, entire lipid molecules crossed the box boundaries. I tried to recenter the trajectory, so that the lipid bilayer would be at the center of the box at all times. But for some reason, this does not seem to work? I have tried simulations using a single comm_group for the entire system, as well as separate ones for the lipid and water, but the same problem appears in either case. Typically, for all-atom bilayers, the following set of commands works to correct the drift: first convert original trajectory to a temp. xtc ### echo 3 0 | trjconv -s *tpr -f original.xtc -o temp.xtc -center - boxcenter zero -pbc mol -n popc.ndx then convert temp.xtc to the final trajecory ### echo 3 0 | trjconv -s k*tpr -f temp.xtc -o final.xtc -center - boxcenter zero -pbc mol -n popc.ndx where groups 3 and 0 are the lipid and the whole system respectively, and final.xtc is my final trajectory. However, this does not work for the MARTINI systems. Looking at the final trajectory in VMD, the bilayer is either at the center of
Re: [gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center
I will change the tau_p values, and report back. This might take more than a week though. maria On Wed, Sep 2, 2009 at 5:16 PM, Berk Hess g...@hotmail.com wrote: It might actually affect the center of mass motion removal, because you would be scaling your system with 1 +- 1 bit at every step. This could produce consistent rounding in one direction in single precision, causing the system to move in one direction. This is something we should check in general. Often people are using too small tau_p values, like 0.5 or 1 ps, so I advise them to use 5 or 10 ps. But if larger values cause problems in single precision we should be aware of this. Could you report back if changing tau_p solves the drifting problem? Berk -- Date: Wed, 2 Sep 2009 17:06:50 +0200 Subject: Re: [gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center From: mariagorano...@gmail.com To: gmx-users@gromacs.org Oh dear. That is not good. the missing decimal point in tau_p it is a typo all right. but it seems i have used it in the simulations too. thank you for noticing, Xavier. that forces redoing a lot of simulations. that said, it should still not impact the center of mass removal anyway? -maria On Wed, Sep 2, 2009 at 4:50 PM, XAvier Periole x.peri...@rug.nl wrote: your second value for tau_p is missing the . is this a typo? On Sep 2, 2009, at 4:45 PM, maria goranovic wrote: Here are the mdp parameters: title= POPC cpp = /usr/bin/cpp integrator = md tinit= 0.0 dt = 0.030 nsteps = 300 nstcomm = 1 comm-grps = Lipid W ; OUTPUT CONTROL OPTIONS = ; Output frequency for coords (x), velocities (v) and forces (f) = nstxout = 3 nstvout = 3 nstfout = 0 nstlog = 3 nstenergy= 3 ns_type = grid nstlist = 10 pbc = xyz rlist= 1.2 ; Method for doing electrostatics = coulombtype = Shift rcoulomb_switch = 0.0 rcoulomb = 1.2 epsilon_r= 15 vdw_type = Shift ; cut-off lengths= rvdw_switch = 0.9 rvdw = 1.2 DispCorr = No ; Temperature coupling = tcoupl = Berendsen tc-grps = Lipid W tau_t= 0.3 0.3 ref_t= 323 323 ; Pressure coupling = Pcoupl = berendsen Pcoupltype = semiisotropic tau_p = 3.030 compressibility = 3e-53e-5 ref_p = 1.01.0 constraints = none constraint_algorithm = Lincs unconstrained_start = no lincs_order = 4 lincs_warnangle = 30 On Wed, Sep 2, 2009 at 4:33 PM, Berk Hess g...@hotmail.com wrote: Hi, I am 99.99% sure that there is no problem with COM motion removal in Gromacs. Could you post your mdp parameters? Berk From: x.peri...@rug.nl To: gmx-users@gromacs.org Subject: Re: [gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center Date: Wed, 2 Sep 2009 16:04:39 +0200 I am not sure how to fix the trajectory that has drifted ... But if your bilayer drifts even if you use a removal of the COM for the water and bilayer separately that means there is problem in the code! And this should be fixed. XAvier. On Sep 2, 2009, at 3:36 PM, maria goranovic wrote: Dear Experts I had posted this earlier, but the problem was not solved by earlier suggestions. So am posting again. I am simulating a POPC bilayer using MARTINI. The simulation ran fine, but the bilayer drifted towards the edge of the box along the bilayer normal, and eventually some of the atoms crossed the box boundaries. In some cases, entire lipid molecules crossed the box boundaries. I tried to recenter the trajectory, so that the lipid bilayer would be at the center of the box at all times. But for some reason, this does not seem to work? I have tried simulations using a single comm_group for the entire system, as well as separate ones for the lipid and water, but the same problem appears in either case. Typically, for all-atom bilayers, the following set of commands works to correct the drift: first convert original trajectory to a temp. xtc ### echo 3 0 | trjconv -s *tpr -f original.xtc -o temp.xtc -center - boxcenter zero -pbc mol -n popc.ndx then convert temp.xtc to the final trajecory ### echo 3 0 | trjconv -s k*tpr -f temp.xtc -o final.xtc -center - boxcenter zero -pbc mol -n popc.ndx where groups 3 and 0
RE: [gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center
Hi, I also just recalled that we have a bug report open since two years already about drift of the COM: http://bugzilla.gromacs.org/show_bug.cgi?id=165 But in that case double precision did not change anything, so that does not seem to be a precision issue. Here tau_p was 1 ps, but up till now we did not manage to find the source of this problem. Thus it would be useful to see if a shorter tau_p fixes it in your case. Berk Date: Wed, 2 Sep 2009 17:21:46 +0200 Subject: Re: [gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center From: mariagorano...@gmail.com To: gmx-users@gromacs.org I will change the tau_p values, and report back. This might take more than a week though. maria On Wed, Sep 2, 2009 at 5:16 PM, Berk Hess g...@hotmail.com wrote: It might actually affect the center of mass motion removal, because you would be scaling your system with 1 +- 1 bit at every step. This could produce consistent rounding in one direction in single precision, causing the system to move in one direction. This is something we should check in general. Often people are using too small tau_p values, like 0.5 or 1 ps, so I advise them to use 5 or 10 ps. But if larger values cause problems in single precision we should be aware of this. Could you report back if changing tau_p solves the drifting problem? Berk Date: Wed, 2 Sep 2009 17:06:50 +0200 Subject: Re: [gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center From: mariagorano...@gmail.com To: gmx-users@gromacs.org Oh dear. That is not good. the missing decimal point in tau_p it is a typo all right. but it seems i have used it in the simulations too. thank you for noticing, Xavier. that forces redoing a lot of simulations. that said, it should still not impact the center of mass removal anyway? -maria On Wed, Sep 2, 2009 at 4:50 PM, XAvier Periole x.peri...@rug.nl wrote: your second value for tau_p is missing the . is this a typo? On Sep 2, 2009, at 4:45 PM, maria goranovic wrote: Here are the mdp parameters: title= POPC cpp = /usr/bin/cpp integrator = md tinit= 0.0 dt = 0.030 nsteps = 300 nstcomm = 1 comm-grps = Lipid W ; OUTPUT CONTROL OPTIONS = ; Output frequency for coords (x), velocities (v) and forces (f) = nstxout = 3 nstvout = 3 nstfout = 0 nstlog = 3 nstenergy= 3 ns_type = grid nstlist = 10 pbc = xyz rlist= 1.2 ; Method for doing electrostatics = coulombtype = Shift rcoulomb_switch = 0.0 rcoulomb = 1.2 epsilon_r= 15 vdw_type = Shift ; cut-off lengths= rvdw_switch = 0.9 rvdw = 1.2 DispCorr = No ; Temperature coupling = tcoupl = Berendsen tc-grps = Lipid W tau_t= 0.3 0.3 ref_t= 323 323 ; Pressure coupling = Pcoupl = berendsen Pcoupltype = semiisotropic tau_p = 3.030 compressibility = 3e-53e-5 ref_p = 1.01.0 constraints = none constraint_algorithm = Lincs unconstrained_start = no lincs_order = 4 lincs_warnangle = 30 On Wed, Sep 2, 2009 at 4:33 PM, Berk Hess g...@hotmail.com wrote: Hi, I am 99.99% sure that there is no problem with COM motion removal in Gromacs. Could you post your mdp parameters? Berk From: x.peri...@rug.nl To: gmx-users@gromacs.org Subject: Re: [gmx-users] Martini simulation problem in recentering trajectory so that the bilayer is at the center Date: Wed, 2 Sep 2009 16:04:39 +0200 I am not sure how to fix the trajectory that has drifted ... But if your bilayer drifts even if you use a removal of the COM for the water and bilayer separately that means there is problem in the code! And this should be fixed. XAvier. On Sep 2, 2009, at 3:36 PM, maria goranovic wrote: Dear Experts I had posted this earlier, but the problem was not solved by earlier suggestions. So am posting again. I am simulating a POPC bilayer using MARTINI. The simulation ran fine, but the bilayer drifted towards the edge of the box along the bilayer normal, and eventually some of the atoms crossed the box boundaries. In some cases, entire lipid molecules crossed the box boundaries. I tried to recenter the trajectory, so that the lipid bilayer would be at the center of the box at all times. But for some reason, this does not seem to work? I have tried simulations using a