[gmx-users] Re: REMD analysis

[Mark Abraham]

On Sat, May 18, 2013 at 5:17 PM, bharat gupta <bharat.85.monu at gmail.com>wrote:

> Dear Sir,
>
> My main objective of carrying out REMD is to study peptide folding and if
> possible to get some insight in protein design and folding. I read some
> articles related to my work and they always show temp (replica_index)
> graphs for 2-3 replicas , saying that the sufficient sampling had been
> achieved. In my case I checked the replica_temp for first three replicas (
> https://www.dropbox.com/s/gkbu0g0e1r37l57/replica_temp0.png) (
> https://www.dropbox.com/s/sffq8rwghjublu0/replica_temp1.png)
>  https://www.dropbox.com/s/ulccw8xabj66ktm/replica_temp2.png)
>

You've shown the same graph three times.


> Almost all the three graphs are similar and the last 15ns shows that there
> is no enough exchange (if I have analyzed correctly) ?? In this case the
> acceptance ratio was fine and PE overlap was also good, but problem lies
> with efficient sampling ??


Like I've been saying, acceptance does not assure enhanced sampling.


> What shall I do in such a case ??
>

Do some of the reading I've been suggesting? One of the papers shows that
it is hard to get good sampling with just Met-enkephalin unless you design
the temperature range for it.

Mark

On Sat, May 18, 2013 at 12:27 AM, Mark Abraham <mark.j.abraham at gmail.com
> >wrote:
>
> > On Fri, May 17, 2013 at 4:26 PM, bharat gupta <bharat.85.monu at gmail.com
> > >wrote:
> >
> > > Dear Sir,
> > >
> > > The the default bin width is 0.1 which I used for plotting the graphs.
> > >
> >
> > That's nice. You need to decide what you need to do about it if you want
> > graphs that look like those you see reported :-)
> >
> >
> > > Another question is about your last reply to my thread "exchange
> > acceptance
> > > is a poor proxy for sampling efficiency". Sorry to ask this, but how to
> > > check whether the sampling efficiency is optimal or not (what should be
> > > optimal sampling efficiency) ??
> > >
> >
> > Ah, now here's the real question :-) Spacing the replicas for optimal
> > *flow* is a difficult problem, even for toy peptides, see e.g. papers by
> > Nadler and Hansmann. Merely accepting exchanges does not imply flow. The
> > belief is that getting flow enhances sampling, but the latter is hard to
> > demonstrate without showing that "simulation time to converged sampling"
> > actually reduces. I'm not aware of anybody who's actually done that - but
> > it would certainly be an advantage if your application is interested in
> > data at a range of temperatures.
> >
> > Mark
> >
> > On Fri, May 17, 2013 at 11:10 PM, Mark Abraham <mark.j.abraham at gmail.com
> > > >wrote:
> > >
> > > > Histograms 101: The smaller your bin width, the more variations you
> > see.
> > > > The more samples you have, the fewer variations you see. A histogram
> > that
> > > > does not mention either of this is a work of fiction.
> > > >
> > > > The number of degrees of freedom in the potential energy distribution
> > is
> > > > also a factor in whether the distribution will look smooth for a
> given
> > > bin
> > > > width and number of samples.
> > > >
> > > > Mark
> > > >
> > > > On Fri, May 17, 2013 at 3:51 PM, bharat gupta <
> > bharat.85.monu at gmail.com
> > > > >wrote:
> > > >
> > > > > Dear Sir,
> > > > >
> > > > > I tried plotting the PE overlap using the following way :-
> > > > >
> > > > > 1. extract PE of each replica using g_energy
> > > > > 2. get the PE distribution using g_analyze -f potential_0.xvg -dist
> > > > > pot0.xvg
> > > > > 3. used xmgrace to plot all the PE distribution graphs together.
> > > > >
> > > > > The same thing I did for temperature distribution for each replica.
> > > > >
> > > > > Here's the file for both PE overlap (
> > > > > https://www.dropbox.com/s/895f1bi0hkuy884/pe_dist.png)
> > > > > temp distribution (
> > > > https://www.dropbox.com/s/ed66uop16blgqwa/temp_dist.png
> > > > > )
> > > > > Is this the correct way ??
> > > > >
> > > > >
> > > > > But the plot doesnot look like this (
> > > > >
> https://www.dropbox.com/s/fsuabkl7zrydnib/sample%20PE%20overlap.jpg
> > ).
> > > > Do i
> > > > > have to normalize the data and then plot in order to get a smooth
> > plot
> > > > like
> > > > > this one??
> > > > >
> > > > > --------
> > > > > Bharat
> > > > > --
> > > > > gmx-users mailing list    gmx-users at gromacs.org
> > > > > http://lists.gromacs.org/mailman/listinfo/gmx-users
> > > > > * Please search the archive at
> > > > > http://www.gromacs.org/Support/Mailing_Lists/Search before
> posting!
> > > > > * Please don't post (un)subscribe requests to the list. Use the
> > > > > www interface or send it to gmx-users-request at gromacs.org.
> > > > > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> > > > >
> > > > --
> > > > gmx-users mailing list    gmx-users at gromacs.org
> > > > http://lists.gromacs.org/mailman/listinfo/gmx-users
> > > > * Please search the archive at
> > > > http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
> > > > * Please don't post (un)subscribe requests to the list. Use the
> > > > www interface or send it to gmx-users-request at gromacs.org.
> > > > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> > > >
> > > --
> > > gmx-users mailing list    gmx-users at gromacs.org
> > > http://lists.gromacs.org/mailman/listinfo/gmx-users
> > > * Please search the archive at
> > > http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
> > > * Please don't post (un)subscribe requests to the list. Use the
> > > www interface or send it to gmx-users-request at gromacs.org.
> > > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> > >
> > --
> > gmx-users mailing list    gmx-users at gromacs.org
> > http://lists.gromacs.org/mailman/listinfo/gmx-users
> > * Please search the archive at
> > http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
> > * Please don't post (un)subscribe requests to the list. Use the
> > www interface or send it to gmx-users-request at gromacs.org.
> > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> >
> --
> gmx-users mailing list    gmx-users at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-users
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
> * Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>