Re: [gmx-users] About g_msd (and noise)
I've been playing with g_msd myself recently, and been seeing weird results toward the end of the simulation. From the below post, it looks like I was doing it correctly (apart from analysing the leaflets separately). Previous posts in the mailing list have implied that increased noise towards the end of a simulation is inherent in the algorithm, I just wanted to check that I was interpreting those posts correctly? And if this is so, do people just not show the results towards the end? - Original Message From: David van der Spoel [EMAIL PROTECTED] To: Discussion list for GROMACS users gmx-users@gromacs.org Sent: Wednesday, January 16, 2008 6:48:32 AM Subject: Re: [gmx-users] About g_msd Justin A. Lemkul wrote: Hi Alan, Thanks for the reply. My initial trajectory showed several of the lipids jumped across the box and continued through the bilayer from there, which resulted in a large displacement, so I processed the trajectory with trjconv -pbc nojump. There is still a rather large initial displacement (within the first several nanoseconds out of 100, likely due to my equilibration procedure of packing the lipids tightly around the peptide), so I attempted to analyze the last 75 ns and 90 ns of the trajectory, using the structures at those times as the reference (in g_msd -s). Still the same result, a large value of D. Any ideas? please go back to your original trajectory and do normal g_msd for the P atoms only. (no mol flags etc.) Thanks again. -Justin Quoting Alan Dodd [EMAIL PROTECTED]: What happens if you visualise the trajectory? Two orders of magnitude in scale of lipid movement should stick out like a sore thumb. - Original Message From: Justin A. Lemkul [EMAIL PROTECTED] To: gmx-users@gromacs.org Sent: Wednesday, January 16, 2008 12:27:45 AM Subject: [gmx-users] About g_msd Hello again, I'm back with a few more questions about g_msd (version 3.3, in case I hadn't mentioned that before). Thanks to Xavier's message earlier, I have abandoned use of ordered trajectories to analyze my lipids. I will deal with lipid shells in the future. For now I am approaching the problem of lateral diffusion coefficients from a slightly different angle. My system contains a helical peptide that is oriented asymmetrically with respect to the DPPC bilayer. It is tilted and only partially embedded into the intracellular leaflet of the bilayer (at the beginning of the simulation). Due to the asymmetry, I would like to study the properties of the leaflets separately, including, among other parameters, the lateral diffusion coefficients of the component lipids. I have found a few papers that have simulated pure DPPC bilayers, and am using them as somewhat of a reference point for the magnitude of the lateral diffusion coefficients that I am determining: E. Lindahl and O. Edholm (2001) J. Chem. Phys. 115 (10), and U. Essmann and M. L. Berkowitz (1999) Biophys. J. 76. For the top leaflet of my bilayer, I am getting a value of D = (4.0+/-2.2)x10^-7 cm^2/sec (reasonable, in terms of order of magnitude, I think), but for the bottom leaflet, I am getting roughly (355.7+/-551.5)x10^-7 cm^2/sec. I figured this enormous number was due to artefacts of PBC, so I tried every iteration of trjconv -pbc, but to no avail. Every result is quite similar. I tried starting g_msd at a later time (10 ns, 25 ns) to determine if any large initial movements of lipids were responsible for the result, but I'm still coming up with the enormous value of D (albeit slightly lower, ~200+/-400) I am using g_msd -mol, with an index file that contains molecule numbers, and then using g_analyze on the output .xvg file to get the values of D. Has anyone ever experienced anything similar? Am I missing something obvious? Thanks in advance, as always, especially if you read the entirety of my lengthy message. -Justin Justin A. Lemkul Graduate Research Assistant Department of Biochemistry Virginia Tech Blacksburg, VA [EMAIL PROTECTED] | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin/ ___ gmx-users mailing listgmx-users@gromacs.org http://www.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 [EMAIL PROTECTED] Can't post? Read http://www.gromacs.org/mailing_lists/users.php Never miss a thing. Make Yahoo your home page. http://www.yahoo.com/r/hs ___ gmx-users mailing listgmx-users@gromacs.org http://www.gromacs.org/mailman/listinfo/gmx-users Please search the
Re: [gmx-users] About g_msd (and noise)
Alan Dodd wrote: I've been playing with g_msd myself recently, and been seeing weird results toward the end of the simulation. From the below post, it looks like I was doing it correctly (apart from analysing the leaflets separately). Previous posts in the mailing list have implied that increased noise towards the end of a simulation is inherent in the algorithm, I just wanted to check that I was interpreting those posts correctly? And if this is so, do people just not show the results towards the end? DESCRIPTION --- g_msd computes the mean square displacement (MSD) of atoms from their initial positions. This provides an easy way to compute the diffusion constant using the Einstein relation. The time between additional starting points for the MSD calculation is set with -trestart. The diffusion constant is calculated by least squares fitting a straight line through the MSD from -beginfit to -endfit. An error estimate given, which is the difference of the diffusion coefficients obtained from fits over the two halfs of the fit interval. Have you tried these option? - Original Message From: David van der Spoel [EMAIL PROTECTED] To: Discussion list for GROMACS users gmx-users@gromacs.org Sent: Wednesday, January 16, 2008 6:48:32 AM Subject: Re: [gmx-users] About g_msd Justin A. Lemkul wrote: Hi Alan, Thanks for the reply. My initial trajectory showed several of the lipids jumped across the box and continued through the bilayer from there, which resulted in a large displacement, so I processed the trajectory with trjconv -pbc nojump. There is still a rather large initial displacement (within the first several nanoseconds out of 100, likely due to my equilibration procedure of packing the lipids tightly around the peptide), so I attempted to analyze the last 75 ns and 90 ns of the trajectory, using the structures at those times as the reference (in g_msd -s). Still the same result, a large value of D. Any ideas? please go back to your original trajectory and do normal g_msd for the P atoms only. (no mol flags etc.) Thanks again. -Justin Quoting Alan Dodd [EMAIL PROTECTED]: What happens if you visualise the trajectory? Two orders of magnitude in scale of lipid movement should stick out like a sore thumb. - Original Message From: Justin A. Lemkul [EMAIL PROTECTED] To: gmx-users@gromacs.org Sent: Wednesday, January 16, 2008 12:27:45 AM Subject: [gmx-users] About g_msd Hello again, I'm back with a few more questions about g_msd (version 3.3, in case I hadn't mentioned that before). Thanks to Xavier's message earlier, I have abandoned use of ordered trajectories to analyze my lipids. I will deal with lipid shells in the future. For now I am approaching the problem of lateral diffusion coefficients from a slightly different angle. My system contains a helical peptide that is oriented asymmetrically with respect to the DPPC bilayer. It is tilted and only partially embedded into the intracellular leaflet of the bilayer (at the beginning of the simulation). Due to the asymmetry, I would like to study the properties of the leaflets separately, including, among other parameters, the lateral diffusion coefficients of the component lipids. I have found a few papers that have simulated pure DPPC bilayers, and am using them as somewhat of a reference point for the magnitude of the lateral diffusion coefficients that I am determining: E. Lindahl and O. Edholm (2001) J. Chem. Phys. 115 (10), and U. Essmann and M. L. Berkowitz (1999) Biophys. J. 76. For the top leaflet of my bilayer, I am getting a value of D = (4.0+/-2.2)x10^-7 cm^2/sec (reasonable, in terms of order of magnitude, I think), but for the bottom leaflet, I am getting roughly (355.7+/-551.5)x10^-7 cm^2/sec. I figured this enormous number was due to artefacts of PBC, so I tried every iteration of trjconv -pbc, but to no avail. Every result is quite similar. I tried starting g_msd at a later time (10 ns, 25 ns) to determine if any large initial movements of lipids were responsible for the result, but I'm still coming up with the enormous value of D (albeit slightly lower, ~200+/-400) I am using g_msd -mol, with an index file that contains molecule numbers, and then using g_analyze on the output .xvg file to get the values of D. Has anyone ever experienced anything similar? Am I missing something obvious? Thanks in advance, as always, especially if you read the entirety of my lengthy message. -Justin Justin A. Lemkul Graduate Research Assistant Department of Biochemistry Virginia Tech Blacksburg, VA [EMAIL PROTECTED] | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin/ ___ gmx-users mailing listgmx-users@gromacs.org http://www.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/search before
Re: [gmx-users] About g_msd (and noise)
Hi Alan, see Fig 3 in the paper I was mentionning in my previous post (http://dx.doi.org/10.1063/1.2393240); from the legend the authors say 'deviation from linearity at long times is due to poor statistics'. So to answer your question, that's probably true, most papers just show the beginning of the plot where the MSD is linear. As quoted before by David, one has just to take care about fitting in the linear region (with the -beginfit and -endfit flags). Cheers, Patrick Alan Dodd a écrit : I've been playing with g_msd myself recently, and been seeing weird results toward the end of the simulation. From the below post, it looks like I was doing it correctly (apart from analysing the leaflets separately). Previous posts in the mailing list have implied that increased noise towards the end of a simulation is inherent in the algorithm, I just wanted to check that I was interpreting those posts correctly? And if this is so, do people just not show the results towards the end? - Original Message From: David van der Spoel [EMAIL PROTECTED] To: Discussion list for GROMACS users gmx-users@gromacs.org Sent: Wednesday, January 16, 2008 6:48:32 AM Subject: Re: [gmx-users] About g_msd Justin A. Lemkul wrote: Hi Alan, Thanks for the reply. My initial trajectory showed several of the lipids jumped across the box and continued through the bilayer from there, which resulted in a large displacement, so I processed the trajectory with trjconv -pbc nojump. There is still a rather large initial displacement (within the first several nanoseconds out of 100, likely due to my equilibration procedure of packing the lipids tightly around the peptide), so I attempted to analyze the last 75 ns and 90 ns of the trajectory, using the structures at those times as the reference (in g_msd -s). Still the same result, a large value of D. Any ideas? please go back to your original trajectory and do normal g_msd for the P atoms only. (no mol flags etc.) Thanks again. -Justin Quoting Alan Dodd [EMAIL PROTECTED]: What happens if you visualise the trajectory? Two orders of magnitude in scale of lipid movement should stick out like a sore thumb. - Original Message From: Justin A. Lemkul [EMAIL PROTECTED] To: gmx-users@gromacs.org Sent: Wednesday, January 16, 2008 12:27:45 AM Subject: [gmx-users] About g_msd Hello again, I'm back with a few more questions about g_msd (version 3.3, in case I hadn't mentioned that before). Thanks to Xavier's message earlier, I have abandoned use of ordered trajectories to analyze my lipids. I will deal with lipid shells in the future. For now I am approaching the problem of lateral diffusion coefficients from a slightly different angle. My system contains a helical peptide that is oriented asymmetrically with respect to the DPPC bilayer. It is tilted and only partially embedded into the intracellular leaflet of the bilayer (at the beginning of the simulation). Due to the asymmetry, I would like to study the properties of the leaflets separately, including, among other parameters, the lateral diffusion coefficients of the component lipids. I have found a few papers that have simulated pure DPPC bilayers, and am using them as somewhat of a reference point for the magnitude of the lateral diffusion coefficients that I am determining: E. Lindahl and O. Edholm (2001) J. Chem. Phys. 115 (10), and U. Essmann and M. L. Berkowitz (1999) Biophys. J. 76. For the top leaflet of my bilayer, I am getting a value of D = (4.0+/-2.2)x10^-7 cm^2/sec (reasonable, in terms of order of magnitude, I think), but for the bottom leaflet, I am getting roughly (355.7+/-551.5)x10^-7 cm^2/sec. I figured this enormous number was due to artefacts of PBC, so I tried every iteration of trjconv -pbc, but to no avail. Every result is quite similar. I tried starting g_msd at a later time (10 ns, 25 ns) to determine if any large initial movements of lipids were responsible for the result, but I'm still coming up with the enormous value of D (albeit slightly lower, ~200+/-400) I am using g_msd -mol, with an index file that contains molecule numbers, and then using g_analyze on the output .xvg file to get the values of D. Has anyone ever experienced anything similar? Am I missing something obvious? Thanks in advance, as always, especially if you read the entirety of my lengthy message. -Justin Justin A. Lemkul Graduate Research Assistant Department of Biochemistry Virginia Tech Blacksburg, VA [EMAIL PROTECTED] | (540) 231-9080 http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin/ ___ gmx-users mailing listgmx-users@gromacs.org http://www.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
