[gmx-users] Umbrella Sampling - good histogram but no result in profile

Jochen Hub jhub at gwdg.de
Wed Mar 14 20:53:50 CET 2018 


Am 14.03.18 um 16:26 schrieb Ben Tam:
> Hi Neha,
> 
> I have tried that, I have reduced the number of reaction coordinate (from 42 to 10 windows) and still it gives me "nan". I have check each specific pullf.xvg and pullx.xvg, there are non zero value inside. Hence forth I don't understand what else going on.
> 
> Hi Mark,
> 
> Okay I will do that. However what file should I need to attach with the issue?
> 
> Thank you both for your answer. However I have one more question. When using gmx wham, it is either pullf.dat or pullx.dat are need to provide. As far as I know, you need the force act upon the position of the molecules. What I don't understand is, why do I only need to provide pullx or pullf but not both? how does "gmx wham" understand at what position that the force is applied or vice versa?  is it because it read it off .tpr?

WHAM needs the position, which can (obviously) be taken from pullx, but 
the position can also be computed from pullf when you know the force 
constant (since F = -k*dx). In doubt, I would always use pullf, since 
they files are usually smaller so reading is faster. Parsing pullf files 
is also much easier (so less bug-prone).

Please check your histograms carefully, do they really well overlap? You 
could also play with gmx wham -min -max. This can help to take out 
single data points in the first or last histograms (very right or very 
left) that are "disconnected" from the histograms.

Hope this helps,
Jochen

> 
> Thank you very much,
> 
> Ben
> ________________________________
> From: gromacs.org_gmx-users-bounces at maillist.sys.kth.se <gromacs.org_gmx-users-bounces at maillist.sys.kth.se> on behalf of Awasthi, Neha <neha.awasthi at biologie.uni-goettingen.de>
> Sent: Wednesday, March 14, 2018 15:16
> To: gmx-users at gromacs.org
> Subject: Re: [gmx-users] Umbrella Sampling - good histogram but no result in profile
> 
> Before labeling this as a bug and reporting on redmine, I suggest you try :
> 
> 
> 1) calculate the profile over part(s) of your reaction coordinate, i.e. reduce the number of windows you work with.
> 
> In other words,  start with small number of windows in pullx-files.dat & tpr-files.dat.
> 
> If you get partial profiles, it would point towards some windows (along your reaction coordinate) that cause problems.  One needs to patiently test increasing number of windows.
> 
> 
> 2) A small region of poor overlap can lead  to profiles with nan.
> 
> high k value should not be a problem provided you have the required number of windows and the histograms overlap (check about how much overlap is good overlap) . Sometimes, along a steep profile, histograms are difficult to assign, and tend to shift leading to small regions of poor overlap.
> 
> 
> Hope this helps.
> 
> 
> ________________________________
> From: gromacs.org_gmx-users-bounces at maillist.sys.kth.se <gromacs.org_gmx-users-bounces at maillist.sys.kth.se> on behalf of Mark Abraham <mark.j.abraham at gmail.com>
> Sent: Wednesday, March 14, 2018 3:51 PM
> To: gmx-users at gromacs.org
> Subject: Re: [gmx-users] Umbrella Sampling - good histogram but no result in profile
> 
> Hi,
> 
> That would generally suggest something divided by zero, or similar, eg
> because no data was in a histogram bin, etc.
> 
> That behavior is a bug that we could perhaps fix if you would be so kind as
> to report it at https://redmine.gromacs.org.
> 
> But the core problem remains that something about how you sampled didn't
> meet the expectations of gmx wham.
> 
> Mark
> 
> On Wed, Mar 14, 2018, 15:45 Ben Tam <btam125 at hotmail.co.uk> wrote:
> 
>> Dear all,
>>
>> I have a rather strange problem. Currently I am trying to get a PMF of a
>> water molecule through a hydrophobic materials. At the end of using "gmx
>> wham", I get a rather good histogram and it cover the whole range of the
>> system. However at the making profile.xvg. I get the answer "nan" inside
>> the file. Therefore I really cannot think of why is it giving me that?
>>
>> Currently I set it as :
>> ;Pull code
>> pull = yes
>> ;pull-cylinder-r = 0.5
>> pull_ngroups = 2
>> pull_ncoords = 1
>> pull_group1_name = SOL
>> pull_group2_name = MOL
>> pull_coord1_type = umbrella ; harmonic biasing force
>> pull_coord1_geometry = direction-periodic ; simple distance increase,
>> direction-periodic
>> pull_coord1_vec = 1 1 1
>> pull_coord1_groups = 1 2
>> pull_coord1_dim = Y Y Y
>> pull_coord1_rate = 0.00 ; 0.01 nm per ps = 10 nm per ns
>> pull_coord1_k = 10000 ; kJ mol^-1 nm^-2
>> pull_coord1_start = yes ; define initial COM distance > 0
>>
>> Also this appears when I used the command "gmx wham":
>>
>> Switched to exact iteration in iteration 1
>>             1) Maximum change -1.000000e+20
>> Converged in 2 iterations. Final maximum change -1e+20
>>
>> Therefore I really don't understand what I have to do in order to get a
>> profile.xvg. I understand the k value is high, but without that high value,
>> the water won't settle within the assign window.
>>
>> Am I doing something wrong? because I got a good histogram but no result
>> in the profile.xvg? Thank you very much for your help.
>>
>> Best regards,
>>
>> Ben
>>
>>
>> --
>> Gromacs Users mailing list
>>
>> * Please search the archive at
>> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
>> posting!
>>
>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>
>> * For (un)subscribe requests visit
>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
>> send a mail to gmx-users-request at gromacs.org.
>>
> --
> Gromacs Users mailing list
> 
> * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting!
> 
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> 
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-request at gromacs.org.
> --
> Gromacs Users mailing list
> 
> * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting!
> 
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> 
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-request at gromacs.org.
> 

-- 
---------------------------------------------------
Dr. Jochen Hub
Computational Molecular Biophysics Group
Institute for Microbiology and Genetics
Georg-August-University of Göttingen
Justus-von-Liebig-Weg 11, 37077 Göttingen, Germany.
Phone: +49-551-39-14189
http://cmb.bio.uni-goettingen.de/
---------------------------------------------------

More information about the gromacs.org_gmx-users mailing list