[Spam:*****] [gmx-users] REMD analysis
Mark Abraham
mark.j.abraham at gmail.com
Thu May 16 16:24:33 CEST 2013
On Thu, May 16, 2013 at 2:04 PM, bharat gupta <bharat.85.monu at gmail.com>wrote:
> Dear Sir,
>
> Here's the result of three different runs :
>
> Temperature distribution for three trials
>
> Repeat-1 280 298 317 337 359 382 406 432 460 489 520 554 589 627
> Repeat-2 280 299 319 340 363 388 414 441 471 503 536 572 611
> Repeat-3 280 300 322 345 370 397 426 457 490 526 564 605 649
>
> md.log files output from three different trials:
>
> Repeat-1 .37 .28 .26 .30 .25 .29 .32 .35 .32 .35 .36 .32 .31
> Repeat-2 .30 .33 .30 .25 .19 .27 .30 .31 .27 .40 .34 .31
> Repeat-3 .18 .22 .26 .34 .26 .28 .25 .27 .27 .25 .27 .22
>
> I think as the required acceptance value all the three trials are fine, but
> trail 3 would be much better to continue the further runs and anlysis ??
>
Probably. But exchange acceptance is a poor proxy for sampling efficiency -
see recent discussions of REMD on this list.
Mark
So, is it fine to continue with the third simulation ?? But still the
> problem is that I am not getting the exact graphs with xmgrace??
>
>
> On Thu, May 16, 2013 at 5:36 PM, XAvier Periole <x.periole at rug.nl> wrote:
>
> >
> > You have to convince yourself, not me :)) But I can give you my opinion …
> >
> > On May 16, 2013, at 10:33 AM, bharat gupta <bharat.85.monu at gmail.com>
> > wrote:
> >
> > > Okay Sir, I will try two-three combinations this time and will report
> > back
> > > to you ...
> > >
> > >
> > > On Thu, May 16, 2013 at 5:25 PM, XAvier Periole <x.periole at rug.nl>
> > wrote:
> > >
> > >>
> > >> An acceptance ratio of 0.2/0.3 is normally best. The problem with high
> > >> acceptance ratio is that it means that a large portion of the
> exchanges
> > are
> > >> just back and forth exchanges between consecutive exchange and are
> thus
> > >> disturbing the system more that actually helping sampling.
> > >>
> > >> I do not know particularly the paper you mention but if you like what
> > they
> > >> do, it is your choice at the end.
> > >>
> > >> Why don;t you just increase the spacing between the replicas? You will
> > >> need less replicas and potentially you could run two simulations
> > instead of
> > >> one and evaluate the convergence ...
> > >>
> > >> On May 16, 2013, at 1:50 AM, bharat gupta <bharat.85.monu at gmail.com>
> > >> wrote:
> > >>
> > >>> The plots that I showed in my last mail were for all replicas. I
> tried
> > >>> plotting the first 500 ps of replica_index and replica_time files. I
> > >> think
> > >>> the plots look fine, and there could be problem with the plotting
> tool
> > .
> > >>> Here the link for both files ,
> > >>> https://www.dropbox.com/s/2g16mlxfsme4rx2/replica_temp.bmp
> > >>> https://www.dropbox.com/s/8jfs0b9whu6j7lo/replica_index.bmp
> > >>>
> > >>> Now regarding the high acceptance ratio which is 0.5 , I came across
> a
> > >>> paper (http://www.pnas.org/content/100/13/7587.full.pdf), here they
> > have
> > >>> mentioned that their average acceptance ratio ranged between 30 to
> > 80%. I
> > >>> have a question here, how did they calculate the range for the
> average
> > >>> acceptance ratio or is it average ratio for each replica . Actually,
> > this
> > >>> is the reference I am following. I am also interested in peptide
> > folding
> > >>> simulation, similar to this article.
> > >>>
> > >>> I want to know, whether the average acceptance ratio that I have got
> > for
> > >> my
> > >>> trial simulation is correct , together with the replica_temp and
> > >>> replica_remd plots. Can I proceed for large production runs to
> complete
> > >> my
> > >>> experiment ??
> > >>>
> > >>>
> > >>>
> > >>>
> > >>> On Tue, May 14, 2013 at 6:34 PM, XAvier Periole <x.periole at rug.nl>
> > >> wrote:
> > >>>
> > >>>>
> > >>>> The interval between the exchange trial affect the efficiency of
> REMD
> > >> but
> > >>>> not the the exchange ratio (at least in principle).
> > >>>>
> > >>>> In you case I am not sure what the plot are showing! Are these
> showing
> > >> all
> > >>>> the replicas? what are the units?
> > >>>>
> > >>>> On May 14, 2013, at 5:07 AM, bharat gupta <bharat.85.monu at gmail.com
> >
> > >>>> wrote:
> > >>>>
> > >>>>> Dear Sir,
> > >>>>>
> > >>>>> Here's the result for the REMD trial with large temperature gaps.
> > >>>>>
> > >>>>> Temp. distribution : 280.0 294.9 310.7 327.3 344.7 363.1 382.5
> 402.9
> > >>>> 424.4
> > >>>>> 447.1 471.0 496.1 522.6 550.5 579.9 610.8
> > >>>>>
> > >>>>> Out of md16.log :
> > >>>>>
> > >>>>> Replica exchange statistics
> > >>>>> Repl 249 attempts, 125 odd, 124 even
> > >>>>> Repl average probabilities:
> > >>>>> Repl 0 1 2 3 4 5 6 7 8 9 10 11
> > 12
> > >>>>> 13 14 15
> > >>>>> Repl .40 .34 .38 .43 .43 .36 .45 .40 .37 .48 .47
> .45
> > >> .47
> > >>>>> .44 .46
> > >>>>>
> > >>>>> Repl number of exchanges:
> > >>>>> Repl 0 1 2 3 4 5 6 7 8 9 10 11
> > 12
> > >>>>> 13 14 15
> > >>>>> Repl 50 42 46 52 57 40 58 49 42 53 61
> 63
> > >> 56
> > >>>>> 57 58
> > >>>>>
> > >>>>> Repl average number of exchanges:
> > >>>>> Repl 0 1 2 3 4 5 6 7 8 9 10 11
> > 12
> > >>>>> 13 14 15
> > >>>>> Repl .40 .34 .37 .42 .46 .32 .46 .40 .34 .43 .49
> .51
> > >> .45
> > >>>>> .46 .46
> > >>>>> Average acceptance ratio : 0.46
> > >>>>>
> > >>>>> But, the repli_index.xvg and replica_temp.xvg files still shows
> that
> > >> the
> > >>>>> replicas does not exchange equally well .
> > >>>>>
> > >>>>> https://www.dropbox.com/s/zkbwpuj7l2o282b/replica_index.png
> > >>>>> https://www.dropbox.com/s/0c8gp584v1hvlbx/replica_temp.png
> > >>>>>
> > >>>>> what could be wrong in this case?? Is it the mdp file settings or
> > >>>> implicit
> > >>>>> solvent setting. Does the time to replica to exhange also affects
> > their
> > >>>>> swapping ??
> > >>>>>
> > >>>>>
> > >>>>>
> > >>>>> On Tue, May 14, 2013 at 12:24 AM, XAvier Periole <x.periole at rug.nl
> >
> > >>>> wrote:
> > >>>>>
> > >>>>>>
> > >>>>>> You need to increase the temperature gaps indeed if you want
> > >> acceptance
> > >>>>>> ratio ~0.2/0.3. But again this won't work with the water …
> > >>>>>>
> > >>>>>> It is not clear what happens in your index file but probably a
> > problem
> > >>>>>> from grace to plot so many points … you can try to increase the
> "Max
> > >>>>>> drawing path length" in the preference menu of grace.
> > >>>>>>
> > >>>>>> On May 13, 2013, at 4:22 PM, bharat gupta <
> bharat.85.monu at gmail.com
> > >
> > >>>>>> wrote:
> > >>>>>>
> > >>>>>>> Dear Sir,
> > >>>>>>>
> > >>>>>>> I repeated the simulation again for 25 replicas with the
> following
> > >>>> temp.
> > >>>>>>> distribution .
> > >>>>>>>
> > >>>>>>> 280
> > >>>>>>> 289.1
> > >>>>>>> 298.5
> > >>>>>>> 308.2
> > >>>>>>> 318.2
> > >>>>>>> 328.6
> > >>>>>>> 339.3
> > >>>>>>> 350.3
> > >>>>>>> 361.7
> > >>>>>>> 373.5
> > >>>>>>> 385.6
> > >>>>>>> 398.1
> > >>>>>>> 411.1
> > >>>>>>> 424.4
> > >>>>>>> 438.3
> > >>>>>>> 452.5
> > >>>>>>> 467.2
> > >>>>>>> 482.4
> > >>>>>>> 498.1
> > >>>>>>> 514.3
> > >>>>>>> 531.0
> > >>>>>>> 548.3
> > >>>>>>> 566.1
> > >>>>>>> 584.5
> > >>>>>>> 603.5
> > >>>>>>> 623.2
> > >>>>>>>
> > >>>>>>> The output of md.log file is :-
> > >>>>>>>
> > >>>>>>> Replica exchange statistics
> > >>>>>>> Repl 24999 attempts, 12500 odd, 12499 even
> > >>>>>>> Repl average probabilities:
> > >>>>>>> Repl 0 1 2 3 4 5 6 7 8 9 10 11
> > >> 12
> > >>>>>>> 13 14 15 16 17 18 19 20 21 22 23 24 25
> > >>>>>>> Repl .63 .63 .62 .62 .61 .61 .60 .60 .59 .59 .58
> > .59
> > >>>> .59
> > >>>>>>> .60 .60 .61 .62 .62 .63 .64 .64 .65 .65 .66 .66
> > >>>>>>>
> > >>>>>>> Repl number of exchanges:
> > >>>>>>> Repl 0 1 2 3 4 5 6 7 8 9 10 11
> > >> 12
> > >>>>>>> 13 14 15 16 17 18 19 20 21 22 23 24 25
> > >>>>>>> Repl 7822 7752 7816 7760 7639 7628 7511 7442 7375 7332 7312
> > 7424
> > >>>> 7408
> > >>>>>>> 7410 7522 7559 7684 7697 7878 7927 7917 8073 8151 8208 8266
> > >>>>>>>
> > >>>>>>> Repl average number of exchanges:
> > >>>>>>> Repl 0 1 2 3 4 5 6 7 8 9 10 11
> > >> 12
> > >>>>>>> 13 14 15 16 17 18 19 20 21 22 23 24 25
> > >>>>>>> Repl .63 .62 .63 .62 .61 .61 .60 .60 .59 .59 .58
> > .59
> > >>>> .59
> > >>>>>>> .59 .60 .60 .61 .62 .63 .63 .63 .65 .65 .66 .66
> > >>>>>>>
> > >>>>>>> The average acceptance ration is around 0.6 which is still high.
> > >>>>>>>
> > >>>>>>> The link for replica_temp,replica_index :
> > >>>>>>> https://www.dropbox.com/s/c7soajnwc3uww8j/replica_temp.png
> > >>>>>>> https://www.dropbox.com/s/wvx82m4c6cnsfit/replica_index.png
> > >>>>>>>
> > >>>>>>> The temp files look better but the index file looks weird ...
> > >>>>>>>
> > >>>>>>> Do i need to experiment with the gap difference in order to get
> the
> > >>>>>>> required ration of 0.2-0.3 ?? There is some problem with the .mdp
> > >> file
> > >>>>>>> settings??
> > >>>>>>>
> > >>>>>>> --
> > >>>>>>> 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
> > >>
> > >> --
> > >> 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
>
More information about the gromacs.org_gmx-users
mailing list