Re: [gmx-users] Re: Re: Re: umbrella potential
Stefan Hoorman wrote: Well, I tried again, this time simulating 1ns each window. I thought about simulating more, but thought of making a test before making longer simulations for nothing. And again, my histogram looks like a chromatographic peak, but the range this time is different. The rest of You should have overlapping roughly Gaussian distributions. I can't tell if that's what you have, or if you have independent peaks. If the latter is the case, then your windows do not overlap sufficiently to run WHAM. the histogram is a straight line parallel to the x axis set in zero. I Parallel lines indicate discontinuities in the WHAM analysis. I really think you need to post an image of this online so we can see. Without seeing what you're dealing with, it's very hard to be helpful. didn't understand what you meant by "getting what you think you are setting up". I have checked the distance between my groups, both with g_dist and by visual aid. The separate all right, there is no doubt about that. Each window is what it should be, a window for the 1.5nm distance starts at 1.5, decreases a little bit in the first picoseconds and then stabilizes at 1.47 or 1.45. When I start g_wham analysis, the verbose states that my initial distance is 0.99nm and then it says that "Determined boundaries to 1.046511 and 2.140948". I tried this time Right, because your windows appear to range between these two values, so the closest window is restraining your structures at roughly 1 nm, and the furthest window is restraining the strucutres at 2.14 nm COM separation. Is this what you expect? I thought before you were dealing with distances out to 2.5 nm. simulating longer periods, and got the same result as before. The profile graphic stars as a descending curve from 0 to -5 kCal, then it starts rising in a "up the hill" way until it reaches +45kCal. I really don't know what I am doing wrong this time. Can you re-post the .mdp options you're using for pulling? Also, images of both your histogram and profile would be extremely helpful. I'm having a hard time picturing what you're describing with respect to the histograms. -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 -- 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
[gmx-users] Re: Re: Re: umbrella potential
> Stefan Hoorman wrote: > > > I simulate 400 ps for each window. I have a total of 20 windows. My > > 400 ps is relatively short, especially given the speed of current hardware > and > of Gromacs 4.0. I generally see longer time periods in the literature. > > > histogram looks like a chromatographic peak ranging from 0.74 and 0.91 > > Then you are not getting the separation you described before (>2.5 nm). It > looks like you are only pulling a distance of 0.17 nm total. > > > in the x axis and the count (y axis) goes up to 3. Is there a way to > > index my histogram.xvg or my profile.xvg file and send it to the gromacs > > user's list? Or is it not necessary? > > The best way to send this information is to generate image files and post > them > to a free site where others can see them (photobucket, etc). Up to you to > determine whether its necessary. At this point, I think you are just not > getting what you think you are setting up. > > -Justin > > Well, I tried again, this time simulating 1ns each window. I thought about simulating more, but thought of making a test before making longer simulations for nothing. And again, my histogram looks like a chromatographic peak, but the range this time is different. The rest of the histogram is a straight line parallel to the x axis set in zero. I didn't understand what you meant by "getting what you think you are setting up". I have checked the distance between my groups, both with g_dist and by visual aid. The separate all right, there is no doubt about that. Each window is what it should be, a window for the 1.5nm distance starts at 1.5, decreases a little bit in the first picoseconds and then stabilizes at 1.47 or 1.45. When I start g_wham analysis, the verbose states that my initial distance is 0.99nm and then it says that "Determined boundaries to 1.046511 and 2.140948". I tried this time simulating longer periods, and got the same result as before. The profile graphic stars as a descending curve from 0 to -5 kCal, then it starts rising in a "up the hill" way until it reaches +45kCal. I really don't know what I am doing wrong this time. ___ 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] Re: Re: Re: umbrella potential
Stefan Hoorman wrote: I have a test set which is comprised of the same constituents of my real system, except the solvent (i mean, it is in vacuum. This test set I use to (as the name says) test different different parameters without having to wait too much. I know it would not serve as a very quantitative set, but even though, when I changed the force_constant of the umbrella potentia to a lower value (from 35 to 10) I got something out of the WHAM that looked more like a "rollercoaster", I little bit edgy perhaps, but at least not a downstream line like the one before. When I changed it again to 5 kJ, the graphic still looked like a rollercoaster but it differed in 1 or 2 kcal more negative than the one with 10 kJ as umbrella force constant. Perhaps, in addition to space my intervals as a function of distance, my spring constant is wrong. Does that make any sense? Well, a "vacuum" run might be a bit strange, anyway, so I wouldn't base much on it, since the force fields weren't designed for that kind of usage. Rough edges and transitions might be present, but could be smoothed with proper sampling and window distribution. As for the force constant, I generally see much greater numbers in the literature (500-1000 kJ mol^-1 nm^-2). I don't know if there's a rule of thumb for assigning this value. Ahh...by the way, the link for oldwiki is not working. I will try it later. Indeed, it has been running very slowly, at least. You can access the same content here: http://www.gromacs.org/Documentation/How-tos/Making_Commands_Non-Interactive -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 -- 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
[gmx-users] Re: Re: Re: umbrella potential
> > > Stefan Hoorman wrote: > > > > > > > The distances between the two structures in each of the windows > > are (in nm): > > > 0ps = 1.56 ; 500ps = 1.54; 1000ps = 1.56; 1500 = 1.56; 2000 = > > 1.52; 2500 > > > = 1.65; 3000 = 1.66; 3500 = 1.77; 4000 = 1.63; 4500 = 1.62; 5000 > > = 1.7; > > > 5250 = 1.62; 5500 = 1.6; 5750 = 1.73; 6000 = 1.85; 6250 = 1.86; > > 6500 = > > > 1.91; 6750 = 1.84; 7000 = 1.88; 7250 = 1.82; 7500 = 1.93 ; 7750 = > > 2.04; > > > 8000 = 2.16 ; 8250 = 2.3 ; 8500 = 2.32 ; 8750 = 2.45 ; 9000 = 2.52 > ; > > > 9250 = 2.5 ; 9500 = 2.62 ; 9750 = 2.65 ; 1 = 2.66 > > > > There is substantial overlap and jumping back and forth (i.e., 1.56, > > decrease to > > 1.54, increase to 1.56, then back to 1.52, etc) in these distances, > > so I think > > they are not optimal for umbrella sampling. Instead of more regular > > spacing in > > time, I would suggest more regular spacing in terms of distance. > Most > > literature on the topic cites how far apart windows were in terms of > > distance, > > which is important for WHAM analysis. > > > > -Justin > > > > > > Is there a way to do this automatically in gromacs? To extract > > structures from the simulation based on the distance between them? > > > > > > No, but you can use a script that iteratively calls g_dist on separate > structures from your trajectory (trjconv -sep). See, for example: > > http://oldwiki.gromacs.org/index.php/Making_Commands_Non-Interactive > > -Justin > > > I have a test set which is comprised of the same constituents of my real system, except the solvent (i mean, it is in vacuum. This test set I use to (as the name says) test different different parameters without having to wait too much. I know it would not serve as a very quantitative set, but even though, when I changed the force_constant of the umbrella potentia to a lower value (from 35 to 10) I got something out of the WHAM that looked more like a "rollercoaster", I little bit edgy perhaps, but at least not a downstream line like the one before. When I changed it again to 5 kJ, the graphic still looked like a rollercoaster but it differed in 1 or 2 kcal more negative than the one with 10 kJ as umbrella force constant. Perhaps, in addition to space my intervals as a function of distance, my spring constant is wrong. Does that make any sense? Ahh...by the way, the link for oldwiki is not working. I will try it later. ___ 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