[gmx-users] PMF profile using g_wham with pull_geometry = distance for ion channel

Thomas Piggot t.piggot at soton.ac.uk
Wed May 5 13:04:28 CEST 2010

Hi again,

Just before I reside myself to rerunning a large number of simulations I 
was wondering if it is possible to just rerun the simulations where the 
ion is above and below the reference using pull_geometry=position. Then 
could I combine these with the other simulations where 
pull_geometry=distance to get the complete PMF?


Jochen Hub wrote:
> Thomas Piggot wrote:
>> Jochen thanks for your help.
>> That was what I was afraid of you saying. Oh well, at least this will 
>> hopefully provide a useful reference so somebody doesn't make this 
>> same mistake.
>> On a side note I noticed a couple of small issues with g_wham. Firstly 
>> the -wcorr option says Input, opt. not Output, opt as I assume it 
>> should be.
> Ah thanks. I'll correct that soon.
>> Also when using -min and -max with the -noauto option, the program 
>> quits with the error:
>> Fatal error:
>> With -auto, do not give -min or -max
> Ok, I guess I never checked that case. But as you point out, you don't 
> have to use -noauto when providing min and max, the auto is switched off 
> automatically.
> Thanks for the feedback,
> Jochen
>> This is not a problem as with -min and -max (without specifying 
>> -noauto) g_wham switches automatically switches auto off, but I 
>> assumed that using -min and -max the -noauto option would have been 
>> the 'correct' way to do this.
>> Cheers
>> Tom
>> Jochen Hub wrote:
>>> Thomas Piggot wrote:
>>>> Hi,
>>>> I am trying to construct a PMF profile for a phosphate ion passing 
>>>> through a membrane protein using umbrella sampling with GROMACS 4.0.5.
>>>> I have performed the umbrella sampling simulations using the 
>>>> following mdp options and I am now attempting to construct the PMF 
>>>> using g_wham.
>>>> ; Pull code
>>>> pull                =  umbrella ; do umbrella sampling
>>>> pull_geometry       =  distance ; can't get PMF with direction
>>> Hi Tom,
>>> if you want to distinguish between above and below, 
>>> pull_geometry=position would have been your choice. What I would do 
>>> now is to run g_wham twice with only the histograms from below and a 
>>> second time only with the histograms from above your reference. 
>>> YOu'll have to make appropriate tpr-files.dat and pullx-files.dat 
>>> input files. Note however, that the PMF very close to your reference 
>>> will be ill-defined since the pull code checked the distance to the 
>>> reference and not the z-coordinate. Therefore I would remove 
>>> simulations in which the ion was sometimes above and sometimes below 
>>> the reference.
>>> To ge the full PMF along the z-axis (which is what you probably want) 
>>> you'll have to do the simulation again (with pull_geometry=position) 
>>> I'm afraid.
>>> I hope this helps,
>>> Jochen
>>>> pull_dim            =  N N Y    ; just in the z
>>>> pull_start          =  yes      ; add com to pull_init1
>>>> pull_ngroups        =  1        ; no. of groups to pull
>>>> pull_group0         =  Protein  ; reference group
>>>> pull_group1         =  PO4      ; pull group
>>>> pull_vec1           =  0 0 0    ;
>>>> pull_init1          =  0        ;
>>>> pull_rate1          =  0.0      ; no change in ref position
>>>> pull_k1             =  1000     ; force constant for restraint
>>>> pull_nstxout        =  1000     ; every 2 ps
>>>> pull_nstfout        =  1000     ; every 2 ps
>>>> The problem is that when running g_wham (using the following 
>>>> command) g_wham correctly reads all of the tpr and pullx files, 
>>>> however the data for the windows above and below the reference atom 
>>>> (the middle atom of the protein) are both included in the same part 
>>>> of the PMF profile.
>>>> g_wham_4.0.5 -ix -it -cycl weighted -b 2000
>>>> I assume that this is occuring due to the pull_geometry = distance 
>>>> option, as the distance to the reference does not matter if the ion 
>>>> is above or below the reference?
>>>> So my question is: Is there a way to construct the complete PMF 
>>>> profile  with the simulations I have already run?
>>>> I should also mention that I have tried using the -min and -max 
>>>> options with g_wham (rather than the auto determination of the 
>>>> boundaries) but this just produces outputs that have all the values 
>>>> as nan. I have also tried using g_wham 4.0.7 but with exactly the 
>>>> same issues ,as well as using the pullf.xvg files as input to g_wham.
>>>> I apologise if I am missing something obvious (or doing something 
>>>> silly!) but any help would be greatly appreciated.
>>>> Cheers
>>>> Tom

Dr Thomas Piggot
University of Southampton, UK.

More information about the gromacs.org_gmx-users mailing list