Hi Daniel,
I only have a little experience with WHAM analysis, so I would wager some
of the more veteran users can weigh in here if I miss anything.
Am I understanding you correctly that you varied your r1 and r4 values
across the different windows? If you go back through the list serve there
is a good amount of discussion on what to set the r1 and r4 values, and I
*believe* they should be constant across all of your windows ( I faced
similar issues as you when I varied my r1 and r4 across windows ). If this
is wrong I would love for someone to correct me.
Additionally it is typically suggested that your r1 and r4 values be 'far
away' from your r2/r3 values, to ensure your potential is harmonic at the
sampled values of your defined coordinate. It is common to set r1 = 1.0 or
0.0, and r4 to be some very large number (99.0 or 999). Check out the Amber
list serve archive for more details on how others have done this, like this
thread: http://archive.ambermd.org/200705/0113.html.
It is possible that having the r1 and r4 values vary across your windows,
as well as them being so close to your sampled region, is what is causing
you trouble. Without knowing much else about your system or setup, that is
where I would look first. Unfortunately that likely means rerunning your
simulations.
Finally, it is striking that your system requires a *minimum *of 50
kcal/(mol*A^2) to maintain your fixed distances. That is a big energy
penalty, and obviously it may make sense in the context of your system, but
it sticks out to me as unusual -- whatever barrier you are sampling must be
huge.
Hope this helps, best of luck!
--Zachary Fallon
On Sat, Apr 13, 2019 at 9:20 AM Daniel Fernández Remacha <
dnlfr1994.gmail.com> wrote:
> Hi!!
>
> I am a student doing a Master Program in Pharmaceutical Chemistry and I am
> writing to get some advice on understanding the results given by WHAM after
> doing an Umbrella Sampling of a dissociation process in a protein (600
> residues approx.)-peptide interaction.
>
> To separate these molecules, a force constant of 50kcal/mol/ang2 (the
> smallest that can keep the distance reasonably steady) is set, and r1 and
> r4 values fixed at -1 and +1 unit of each r2=r3 value for every (5 ns
> production) window. The increment of distance separation from window to
> window is 0.2 angstroms.
>
> The combination of the MD gives a comprehensive separation movement when
> visualized in VMD and umbrella sampling looks nice since the overlap of the
> simulated windows looks adequate. It is much more full of dots than the
> example of the tutorial.
>
> However, I am having difficulties to understand WHAM results.
> When looking at the result.dat (as it is called in the AMBER advanced
> tutorial 17) we get the following data.
>
> #Coor Free +/- Prob +/-
> 7.492857 0.000000 0.000089 0.078953 0.000015
> 7.678571 0.021163 0.000114 0.076200 0.000018
> 7.864286 0.048638 0.000050 0.072768 0.000005
> 8.050000 0.035474 0.000201 0.074392 0.000012
> 8.235714 0.025529 0.000203 0.075644 0.000032
> 8.421429 0.074532 0.000235 0.069675 0.000034
> 8.607143 0.010492 0.000132 0.077576 0.000021
> 8.792857 0.031564 0.000090 0.074882 0.000009
> 8.978571 0.138947 0.000127 0.062539 0.000005
> 9.164286 0.160852 0.000055 0.060283 0.000007
> 9.350000 0.122887 0.000140 0.064246 0.000019
> 9.535714 0.063552 0.000138 0.070970 0.000021
> 9.721429 0.063538 0.000176 0.070971 0.000025
> 9.907143 0.064124 0.000422 0.070902 0.000016
> #Window Free +/-
> #0 0.000000 0.000000
> #1 0.000000 0.000000
> #2 0.000000 0.000000
> #3 0.015216 0.000000
> #4 0.000000 0.000000
> #5 0.000000 0.000000
> #6 0.000000 0.000000
> #7 0.000000 0.000000
> #8 0.010285 0.000000
> #9 0.000000 0.000000
> #10 0.000000 0.000000
> #11 0.000000 0.000000
> #12 0.000000 0.000000
> #13 0.034936 0.000000
>
> The main problem comes when looking at the Free energy, which is used to
> see whether the system has converged or not. In the WHAM manual, it is
> mentioned that when this F value is smaller than the specified tolerance
> the system has converged. The tolerance value set for these results is
> 10^-6.
>
> Moreover, when doing MonteCarlo Bootstrap analysis to look for similarities
> with the information given in the manual, we get 3 iterations; very
> different from the information given in the WHAM manual.
>
> I expected the system to be convergent due to a small increment value,
> enough simulation time and good overlapping between windows. However, I
> find strange to see these F values. The files present in the amber tutorial
> do not show these results ( #Window Free and +/-) and the WHAM manual does
> not give information about the example used so we do not have a solid
> criterion to understand these results.
>
> I would be really grateful if you could help me understand the meaning of
> these results for this system and provide any suggestion to perform a
> successful umbrella sampling.
>
> Thanks in advance,
>
> Daniel Fernández
>
> PD: We modified the Perl script for minimization, equilibration and
> production steps and our own script reads coordinates from the end of the
> last production .rst file window performed; and not from the original
> coordinates as shown in the tutorial. I am not sure if this is correct but
> we thought that this way the system would be less distorted when reaching
> high distance values for separation.
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
Zachary Fallon, PhD Candidate
Dr. Carlos Simmerling Laboratory
The Laufer Center for Physical and Quantitative Biology
The Department of Chemistry, Stony Brook University
Stony Brook, New York 11794
Phone: (914) 703-1010 <(914)+703+1010> Email: zachary.fallon.stonybrook.
<zachary.fallon.stonybrook.edu>edu
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Apr 16 2019 - 12:30:02 PDT
