[gmx-users] Follow up (How to define pull-init for two pull groups) How to concatenate pullx.xvg files

[Justin Lemkul]

On 2/19/14, 6:08 PM, Rini Gupta wrote:
> Dear gmx-users,
>
> Thanks for the nice suggestion.
>
> -noappend option in mdrun works fine and
>   I have successfully extended the simulation of each of the umbrella
> sampling windows (Total 44 windows with spacing of 0.1 nm) up to 25ns in
> steps of 5ns.
> So, now I have 5 sets of .tpr, .xvg files corresponding to each umbrella
> sampling window for 25 ns simulation.
> Now, in order to get PMF following tutorial, I ran g_wham command as:
>
> g_wham_mpi_d -it tpr-files.dat -if pullf-files.dat -sym -o profilejsym.xvg
> -hist histojsym.xvg -unit kJ
>
> I have prepared the list of .tpr files, pullf.xvg  and pullx.xvg file
> corresponding to  last run ( i.e 20 ns to 25 ns ) for each of the umbrella
> sampling.
> How can I concatenate  .tpr ,   pullf.xvg  and pullx.xvg  for full 25ns
> length of simulation run. Is there some gromacs tool like trjcat for
> joining .xvg files?

No.  You don't need to concatenate .tpr files.  Any of them will serve the 
purpose for g_wham, which simply needs to read pull geometry parameters.  The 
.xvg files are just text; standard unix utilities like cat, awk, grep, head, 
tail, etc. are designed to manipulate files like these.

> I have searched across mailing list but did not get any satisfactory answer
> of this question except the use of -b and -e option in g-wham command.
>
> However, considering only last frames of each of the umbrella window , I
> got PMF which have some broken points especially near centre of the lipid
> bilayer.
> Can anyone please tell is this an artefact of poor sampling or real nature
> of the system.

Yes, that probably reflects poor sampling, but if you're not considering all 
relevant data then it may not be a problem; you just need to consider more of 
the time series.

>   The other output histo.xvg, shows sufficient overlap between each of the
> umbrella sampling windows,  but it looks noisy.
> Here, I have used the default value of number of bins in histogram i.e.
> 200. Is this correct?

200 bins usually works well.

> Please find attached here with profile.xvg and histo.xvg
>

Attachments are not accepted by the list.  Posting images on public file-sharing 
servers is generally a more efficient means of conveying this information.

> Please suggest me if this is right or I should further extend these
> simulation in order to get better convergence.
>

It isn't clear that you've even properly assessed the convergence of the data 
that you have.  Proper use of -b and -e to assess different windows of time over 
the whole time series would be the proper approach.

-Justin

-- 
==================================================

Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 601
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441
http://mackerell.umaryland.edu/~jalemkul

==================================================