Re: [AMBER] Optimal values for harm and output_freq in SMD

From: Feng Pan <fpan3.ncsu.edu>
Date: Thu, 24 Nov 2016 13:55:04 -0500

Hi, Ruth

The output_freq only affects the output so it should not affect the total
work. Are you sure you ran both cases with exact same conditions? The
random seed may also affect the
results of SMD.
Are you running the example? I think the harm strength 10kcal/mol is too
low because the CV path does not follow the Steering path very well. Try a
larger harm strength, could be better.
If the strength is too low, the steering does not take many effects, so the
total work may differ a lot from different runs.

Feng

On Thu, Nov 24, 2016 at 6:34 AM, Ruth Helena Tichauer <rhtichau.laas.fr>
wrote:

> Dear Amber users,
>
> I’ve been following tutorial A10 in order to perform SMD for my protein
> and its ligand using the LCOD method. In the tutorial, the spring constant
> for the proton transfer to occur is set to 1000 kcal/mol whose value,
> according to a few related questions on the mailing list, is too high.
>
> In order to find a reasonable value, I’ve run the simulation several times
> and the proton transfer takes place for “low” values such as 10 kcal/mol if
> the output_freq is set to 1000, like in the tutorial. But, if the
> output_freq is lowered (I’ve tried 500 and 50) the proton transfer doesn’t
> occur anymore (for harm=10kcal/mol).
>
> Moreover, in the output_file .txt the CV, handle_position and work are not
> the same for a given time during the MD, resulting in different values for
> the total work done.
>
> For harm=10 kcal/mol and output_freq=1000 I obtain:
>
> #
> # MD time (ps), CV, handle_position, spring_constant, work
> #
> 925.0000 0.50139387 0.70000000 10.00000000
> 0.00000000
> 925.5000 0.89660018 0.69958280 10.00000000
> 0.00030631
> 926.0000 0.83105576 0.69834240 10.00000000
> 0.00081418
> 926.5000 0.66819394 0.69629560 10.00000000
> 0.00204283
> 927.0000 0.60709575 0.69345920 10.00000000
> 0.00467988
> 927.5000 0.93546672 0.68985000 10.00000000
> 0.00719652
> 928.0000 0.56761201 0.68548480 10.00000000
> 0.01073920
> 928.5000 0.62737834 0.68038040 10.00000000
> 0.01357336
> 929.0000 0.92623955 0.67455360 10.00000000
> 0.01711340
> 929.5000 0.66861857 0.66802120 10.00000000
> 0.02039487
> 930.0000 0.91611501 0.66080000 10.00000000
> 0.02901486
> ...
> #
> # <> total work done: 0.7690942380
>
> For harm=10 kcal/mol and output_freq=500 I obtain:
>
> #
> # MD time (ps), CV, handle_position, spring_constant, work
> #
> 925.0000 0.50139387 0.70000000 10.00000000
> 0.00000000
> 925.2500 0.80711710 0.69989535 10.00000000
> 0.00022605
> 925.5000 1.00173702 0.69958280 10.00000000
> 0.00094337
> 925.7500 0.81672099 0.69906445 10.00000000
> 0.00198729
> 926.0000 0.81074280 0.69834240 10.00000000
> 0.00416796
> 926.2500 0.93641120 0.69741875 10.00000000
> 0.00658384
> 926.5000 0.79972556 0.69629560 10.00000000
> 0.00872512
> 926.7500 0.88288670 0.69497505 10.00000000
> 0.01112337
> 927.0000 0.99659450 0.69345920 10.00000000
> 0.01380751
> 927.2500 0.91748709 0.69175015 10.00000000
> 0.01664516
> 927.5000 0.98411987 0.68985000 10.00000000
> 0.01972750
> 927.7500 0.76115131 0.68776085 10.00000000
> 0.02316773
> 928.0000 0.82942938 0.68548480 10.00000000
> 0.02737620
> 928.2500 0.74668613 0.68302395 10.00000000
> 0.03189408
> 928.5000 0.95082231 0.68038040 10.00000000
> 0.03673629
> 928.7500 0.79738517 0.67755625 10.00000000
> 0.04177707
> 929.0000 1.03951245 0.67455360 10.00000000
> 0.04712890
> 929.2500 0.71139449 0.67137455 10.00000000
> 0.05324293
> 929.5000 0.85144838 0.66802120 10.00000000
> 0.05961067
> 929.7500 0.92964077 0.66449565 10.00000000
> 0.06619512
> 930.0000 0.84766718 0.66080000 10.00000000
> 0.07329916
> ...
> #
> # <> total work done: 11.2413859667
>
> So I wonder what exactly the output_freq achieves during the SMD? As I
> want to run the same kind of simulation for my protein and its ligand, I’d
> like to know how to choose this value.
>
> Thank you for any insight on this subject,
>
> Sincerely,
>
> Ruth
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>



-- 
Feng Pan
Ph.D. Candidate
North Carolina State University
Department of Physics
Email:  fpan3.ncsu.edu
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Nov 24 2016 - 11:00:03 PST
Custom Search