[gmx-users] Re: Re[7]: Flat energy profile in g_wham
chris.neale at utoronto.ca
chris.neale at utoronto.ca
Tue Aug 31 14:31:29 CEST 2010
Sorry Alex, but there is no good substitute what I have already
suggested. I helped you find a lead and I suggested a way to simplify
your system to discover what is going on. Now it's up to you to do
that work.
It might end up being a bug, but I doubt it. I have done things
similar to what you are doing and have no problems. And even if it is
a bug, then the developers will need you to find a simple version of
the problem in any event.
Make the water box, pull two waters, try to match up the force and
energy values from your pull output files.
Note that you have forgotten the 1/2 factor in your energy function so
there is no factor of 2 in your force. Also, the force is the negative
derivative. Check the manual for the exact equations.
Chris.
Quoting alexander yakovenko <yakovenko_a at ukr.net>:
> It is 2 options why I can’t understand relation between pullf and
> pullx: either I have no imagine how they should related or they
> related so that one can’t even imagine (say due to bug or file
> format). Till now my view was that we need a force to keep two
> molecules at asked separation. The force is F=2*k*x (or 2*k*|r| in
> case of Y Y Y) and work spent on this is k*x^2. Thus in pullx we
> have either a) coordinates of COMs of both groups or b) COM of
> reference group and relative displacement of pulling group and in
> pullf we have 2*k*x values. The problem is that I can’t substitute
> data from pullx so to get data from pullf and thus I don’t realize
> these relations. Say pullx line is 11984.0000
>
> 3.12086 4.37951 -1.63297 1.98567 2.64212
> -3.25943 so |r| is a) 4.641936 if dx, dy and dz are displacements
> or b) 2.63676933 if dx, dy and dz are just coordinates of
> pulling group COM. Suggesting |ro| to be 2.383+1.0=3.383 nm
> (equilibrated complex position + displacement of 1.0 nm) and k=1000
> kJ/mol/nm^2 we have force of a) 9283.871 or b)
> 5273.53865 kJ/mol/nm. Unfortunately neither of these values
> looks like 332.586 from pullf file... So can you please help
> me to choose my option: a) mdrun is corrupted, b) I don’t
> understand what I am doing or c) pullf and pullx writes some
> specific format of output that should be processed by g_wham to
> make some sense? Because if it is a) I have to post bugzilla, if b)
> I have to deal with water, if c) I have to post gromacs-users in
> order to launch g_wham.
>
> Regards,
>
> Alex
>
> >
>
> >
>
> So now we have the first real lead. Next is for you to simplify your
> system to the point where you *can* understand the relationship
> between the pullf and pullx files. For example, I always suggest a
> water bath of explicit water with the sd integrator and apply the pull
> code to two waters.
>
> It's time for you to bash your head against this one on your own until
> you can find a simple system that has no problem and then a slightly
> more complex system that has the problem. Bring those back to the list
> and we can help again if you can;t figure it out.
>
> PS: you should file a bugzilla or otherwise try to get the gromacs
> list fixed so that you can post directly.
>
> Chris.
>
> Quoting alexander yakovenko < yakovenko_a at ukr.net >:
>
> > Sorry for mixing things, I do pull into N Y N direction and then
> > performed umbrella simulation at Y Y Y for each distances window.
> > This is very good question, actually I don't realize it too. The only
> > thing what I can say is that I haven't edited/renamed these files,
> > just copy them. So this is pure output of mdrun, hopes it knows what
> > it writes... I can provide a tpr of a window for example so you can
> > check couple first strings (for reproduction, I carried out all
> > calculations on 8x Xeons at 2.66 workstations).
> > Alex
> > >
> >
> > First, you confuse me when you say "pulling direction was along
> > Y-axis" since your .mdp file indicates pull_dim = Y Y Y and
> so you are
> > pulling in *all* dimensions.
> >
> > Your profile doesn't look like mine at all, but then again I
> took mine
> > from a force profile. Now that I can see your forces and
> > displacements, I can't find any match between them. For example:
> > cneale at cneale-desktop:~/work/August2010/integral/u10$ tail
> pullf_u10.xvg
> > 11982.0000 255.769
> > 11984.0000 332.586
> > 11986.0000 291.367
> > 11988.0000 264.031
> > 11990.0000 330.69
> > 11992.0000 336.757
> > 11994.0000 327.46
> > 11996.0000 475.078
> > 11998.0000 440.766
> > 12000.0000 342.814
> > cneale at cneale-desktop:~/work/August2010/integral/u10$ tail
> pullx_u10.xvg
> > 11982.0000 3.05189 4.39188 -1.23643 2.03259
> 2.62201 3.35563
> > 11984.0000 3.12086 4.37951 -1.63297 1.98567
> 2.64212 -3.25943
> > 11986.0000 3.13274 4.37699 -1.57303 1.97354
> 2.64089 -3.32607
> > 11988.0000 3.13754 4.38542 -1.52275 1.94917
> 2.63444 -3.38366
> > 11990.0000 3.12712 4.38416 -1.61647 1.93608
> 2.63691 -3.29597
> > 11992.0000 3.11342 4.37694 -1.59735 1.95931
> 2.61705 -3.28952
> > 11994.0000 3.0957 4.384 -1.5691 1.99226 2.57634 -3.3149
> > 11996.0000 3.08384 4.38684 5.24141 2.00727 2.57168 -3.09875
> > 11998.0000 2.99618 4.41111 5.25558 2.08776 2.53869 -3.1227
> > 12000.0000 2.79826 4.41958 5.22444 2.29862 2.53084 -3.12472
> >
> >
> > Can you explain to me how these can match up (with an
> equation please)?
> >
> >
> > Quoting alexander yakovenko < yakovenko_a at ukr.net >:
> >
> > > Yep, it was mistake with a sign, is
> > >
> >
> http://picasaweb.google.ca/117558205101329348732/G_wham#5511351478636719042
> > > looks like your one? Btw my temperature was 310K.
> > > The dollars you would lost (as I used only 1 mdp for all runs via
> > > script and thus avoiding humans mistakes like with pictures) is
> > > probably because complicated landscape - the dissociation
> is coupled
> > > to (partial) unfolding and relative sub-domains shift (actually
> > > this experiment is aimed to support this hypothesis). I.e. it does
> > > not look too suspicious for me; something closer to manual was
> > > expected though. Anyway I attach logs from all my simulations and
> > > the mdp.
> > > As for box, it has 10x16x7 edges and pulling direction was along
> > > Y-axis to bring molecules only 5.5nm apart (DNA and
> protein of ~2 nm
> > > radii) so there is definitely sufficient distances.
> > > Anyway the only bug in g_wham is an explanation I can give to
> > > ABSOLUTELY flat line of PMF profile.
> > > As for colors, can we please stick on windows ids instead? I guess
> > > you told about w4, w6 and w8 so I attached their logs from
> mdrun and
> > > for w10 too.
> > > P.S. I still not understand if there is a way to make
> g_wham working
> > > for me (suggesting all other calculations were made fairly)?
> > > Alex
> > >
> >
>
>
>
>
More information about the gromacs.org_gmx-users
mailing list