[Spam:*****] [gmx-users] REMD analysis
XAvier Periole
x.periole at rug.nl
Thu May 16 10:25:43 CEST 2013
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
More information about the gromacs.org_gmx-users
mailing list