<html><head><style type="text/css"><!-- DIV {margin:0px} --></style></head><body><div style="font-family:times new roman, new york, times, serif;font-size:12pt"><DIV></DIV>
<DIV>>From: "David Mobley" <dmobley@gmail.com><BR>>Date: Fri, 12 May 2006 10:05:55 -0700<BR>>Subject: Re: [gmx-users] Re: Re: angle restraints</DIV>
<DIV style="FONT-SIZE: 12pt; FONT-FAMILY: times new roman, new york, times, serif">
<DIV>
<DIV>>Maarten,<BR><BR>>So you are saying that lambda is just controlling the restraints and<BR>>nothing else? That is, you are using lambda-dependent angle<BR>>restraints? If this is the case, the problem is the "fix" for the bug<BR>>that I referred you, as implemented in 3.3.1. From the "bugzilla"<BR>>page, here's the fix:<BR></DIV>
<DIV>yes, and I fixed it already.</DIV>
<DIV><BR>>>This was a nasty bug.<BR>>>The problem was that the angle-restraint parameters were stored<BR>>>incorrectly in the tpr file.<BR>>>I have fixed it completely in the CVS.<BR>>>I also made a fix for 3.3.1, but there the B-state parameters<BR>>>are always copied from the A-state.<BR><BR>>In other words, in 3.3.1, the claim is that angle restraint parameters<BR>>can't be made to be lambda-dependent; B state and A state parameters<BR>>are the same. I am confused why you would be getting a nonzero<BR>>dv/dlambda at all in this case.<BR>>According to Berk, the angle restraints have the correct lambda<BR>>dependence in the CVS version. So if you are really intent on doing<BR>>things this way (that is, using lambda-dependent angle restraints) you<BR>>may want the CVS version.<BR></DIV>
<DIV>Correct, but I made some changes to ensure correct storing of values in the tpr file. Correct me if I am wrong, but the storage of these values only occurs during the generation of a tpr file with grompp.</DIV>
<DIV> </DIV>
<DIV>I am not sure if there are other bugs in the angle restraints value. If so this might explain why the dG/dl crosses zero. I do not understand it otherwise.</DIV>
<DIV><BR>>Alternatively, you can do what I do, which is run simulations using<BR>>normal (non-lambda dependent) angle restraints with a bunch of<BR>>different spring constants, rather than using the free energy code.<BR>>Think of the spring constant like lambda. Then dV/dlambda=d(restraint<BR>>energy)/dK, where K is the spring constant. For harmonic restraints,<BR>>this is (restraint energy)/K. So you can compute dV/dlambda yourself<BR>>from the energies that are output by just taking the angle restraint<BR>>energy and dividing by the spring constant.</DIV>
<DIV> </DIV>
<DIV>Thanks, I will try this as well. It should be sufficient for my goal.<BR><BR>>Again, I'm still not clear that this is exactly what you're trying to<BR>>do. As far as I can tell if your only lambda dependence is in the<BR>>angle restraints, and you are using 3.3.1, you should be getting a<BR>>dv/dlambda that is zero everywhere, unless you've done something<BR>>strange like turning on soft core potentials or something.<BR><BR>>David<BR><BR><BR>On 5/11/06, Maarten Wolf <maarten_wolf@yahoo.co.uk> wrote:<BR>><BR>> David,<BR>><BR>> Thanks for your reply.<BR>><BR>> I am currently using 3.3.1 code.<BR>><BR>> I am increase the force, to be able to calculate the free energy<BR>> required of introducing the angle restraints in the system. I use small<BR>> force increments each run, using a delta_lambda of 0.<BR>><BR>> Decreasing or increasing in my view only changes the sign of the<BR>> overall graph, but it should
not allow dG/dl to cross zero.<BR>><BR>> Thanks,<BR>> Maarten<BR>><BR>> > > From: "David Mobley" <dmobley@gmail.com><BR>> > To: "Discussion list for GROMACS users" <gmx-users@gromacs.org><BR>> > Date: Thu, 11 May 2006 09:34:28 -0700<BR>> > Subject: Re: [gmx-users] angle restraints<BR>> ><BR>> > Maarten,<BR>> ><BR>> > What version of the code are you using? There was a bug in the angle<BR>> > restraints code until 3.3.1 (or 3.3 cvs) which caused angle<BR>> > restraints<BR>> > to turn off gradually as a function of lambda in free energy<BR>> > calculations, if I remember correctly. (See bugzilla for details:<BR>> > <A href="http://bugzilla.gromacs.org/show_bug.cgi?id=43" target=_blank>http://bugzilla.gromacs.org/show_bug.cgi?id=43</A>).<BR>> ><BR>> > However, I'm not sure that relates to the problem you're describing,<BR>> > but it probably does,
depending on the version you are using.<BR>> ><BR>> > What exactly are you changing in the free energy calculation? Are you<BR>> > changing the angle restraint as a function of lambda? If so, which<BR>> > way<BR>> > are you changing it? Decreasing or increasing restraints?<BR>> ><BR>> > David<BR>> ><BR>> ><BR>> ><BR>> > On 5/11/06, Maarten Wolf <maarten_wolf@yahoo.co.uk> wrote:<BR>> > > Dear users,<BR>> > ><BR>> > > I have a small system with a number of small peptides. I want to<BR>> > > constraint the angle between the N-term to C-term vectors of<BR>> > different<BR>> > > peptides. This works well, but if I do a FEP calculation the dG/dl<BR>> > > becomes smaller than zero at some point.<BR>> > > I expected it to be at least allways greater than zero (zero in the<BR>> > > limit that the restraint does not have to do
anything), since a<BR>> > force<BR>> > > has to be aplied to restraint the angles.<BR>> > > Can anybody explain where I am going wrong?<BR>> > ><BR>> > > Thanks<BR>> > > Maarten<BR>> > ><BR>> > > Send instant messages to your online friends<BR>> > <A href="http://uk.messenger.yahoo.com/" target=_blank>http://uk.messenger.yahoo.com</A><BR>> > > _______________________________________________<BR>> > > gmx-users mailing list gmx-users@gromacs.org<BR>> > > <A href="http://www.gromacs.org/mailman/listinfo/gmx-users" target=_blank>http://www.gromacs.org/mailman/listinfo/gmx-users</A><BR>> > > Please don't post (un)subscribe requests to the list. Use the<BR>> > > www interface or send it to gmx-users-request@gromacs.org.<BR>> > > Can't post? Read <A href="http://www.gromacs.org/mailing_lists/users.php"
target=_blank>http://www.gromacs.org/mailing_lists/users.php</A><BR>> > ><BR>> ><BR>> > > _______________________________________________<BR>> > gmx-users mailing list<BR>> > gmx-users@gromacs.org<BR>> > <A href="http://www.gromacs.org/mailman/listinfo/gmx-users" target=_blank>http://www.gromacs.org/mailman/listinfo/gmx-users</A><BR>> ><BR>><BR>><BR>> MvG Maarten<BR>><BR>> ******************************************************<BR>> ** Work Adress Julianalaan 136 **<BR>> ** 2628 BL Delft **<BR>>
** Netherlands **<BR>> ** +31(0)15-2789382 **<BR>> ******************************************************<BR>><BR>><BR>><BR>> ___________________________________________________________<BR>> Switch an email account to Yahoo! Mail, you could win FIFA World Cup tickets. <A href="http://uk.mail.yahoo.com/" target=_blank>http://uk.mail.yahoo.com</A><BR>> _______________________________________________<BR>> gmx-users mailing list gmx-users@gromacs.org<BR>> <A
href="http://www.gromacs.org/mailman/listinfo/gmx-users" target=_blank>http://www.gromacs.org/mailman/listinfo/gmx-users</A><BR>> Please don't post (un)subscribe requests to the list. Use the<BR>> www interface or send it to gmx-users-request@gromacs.org.<BR>> Can't post? Read <A href="http://www.gromacs.org/mailing_lists/users.php" target=_blank>http://www.gromacs.org/mailing_lists/users.php</A><BR>></DIV></DIV></DIV></div></body></html>