Re: [AMBER] convergence of Ref energies in cpHMD

From: Cruzeiro,Vinicius Wilian D <vwcruzeiro.ufl.edu>
Date: Thu, 23 Jan 2020 22:38:23 +0000

Hello Santo,

I noticed that you are using sander and I understand the complication of running longer simulations without GPU acceleration. However, I am taking my expectation regarding the simulation length from previous publications, like this one for example: https://pubs.acs.org/doi/abs/10.1021/acs.jctc.8b00935 . The simulations here were made with implicit solvent and for a system with only 270 atoms. As you can see, without REMD only 5 ns of simulation is clearly not nearly enough to achieve convergence (see Figure 5, for example). The --cumulative option in cphstats should confirm if this is the case for your system too (I would bet it is).

Running simulations with REMD will actually decrease a bit the computational efficiency (amount of ns/day that you can do), however, you will reach convergence much faster, which means you will be able to run much shorter simulations than in a serial CpHMD simulation. In addition to that, when you have information about more than one pH value the pKa predictions become more accurate, and this also helps finddgref.py in finding the correct DELTAGREF value. Therefore, at the end of the day, if you have access to more CPUs, it is a good idea to take advantage of pH-REMD.

I hope this helps,
Best,


Vinícius Wilian D. Cruzeiro

PhD Candidate
Department of Chemistry, Physical Chemistry Division
University of Florida, United States

Voice: +1(352)846-1633

________________________________
From: Kolattukudy P. Santo <santotheophys.gmail.com>
Sent: Wednesday, January 22, 2020 9:37 AM
To: Cruzeiro,Vinicius Wilian D <vwcruzeiro.ufl.edu>
Cc: AMBER Mailing List <amber.ambermd.org>
Subject: Re: [AMBER] convergence of Ref energies in cpHMD

[External Email]
Hi Vinicius
Thank you for your suggestion. 40 ns would be long time for serial mode. My system is of just 120 atoms, I am not sure parallel run would be more effective in terms of speed.
Santo

On Wed, 22 Jan 2020 at 02:44, Cruzeiro,Vinicius Wilian D <vwcruzeiro.ufl.edu<mailto:vwcruzeiro.ufl.edu>> wrote:
Hello Santo,

I suspect your issue is in the fact that you are running the production phase with only 4 ns. I suggest you to run a 4 ns simulation (outside finddgref.py) with one of the DELTAGREF values in the log file you sent and then use the --cumulative option in cphstats to check the convergence of your fraction of protonated species. I strongly suspect that this analysis will show that 4 ns is not enough to converge.

If that is indeed the case, you will need to increase the total number of steps in your mdin file (I would try 40 ns, as a first trial). If you have access to more CPUs, you would gain significantly in sampling and in accuracy by running finddgref.py in parallel mode, which makes use of pH-REMD.

I hope this helps,
Best,


Vinícius Wilian D. Cruzeiro

PhD Candidate
Department of Chemistry, Physical Chemistry Division
University of Florida, United States

Voice: +1(352)846-1633

________________________________
From: Kolattukudy P. Santo <santotheophys.gmail.com<mailto:santotheophys.gmail.com>>
Sent: Tuesday, January 21, 2020 9:13 AM
To: AMBER Mailing List <amber.ambermd.org<mailto:amber.ambermd.org>>
Subject: Re: [AMBER] convergence of Ref energies in cpHMD

[External Email]

Hi Vinicius
The mdin file is:
-------------------
Implicit solvent constant pH molecular dynamics
 &cntrl
   imin=0, irest=1, ntx=5,
   ntpr=1000, ntwx=1000, nstlim=2000000,
   dt=0.002, ntt=3, tempi=300,
   temp0=300, tautp=2.0, ig=-1,
   ntp=0, ntc=2, ntf=2, cut=30,
   ntb=0, igb=2, saltcon=0.1,
   nrespa=1, tol=0.000001, icnstph=1,
   solvph=4.5, ntcnstph=5,
   gamma_ln=5.0, ntwr=10000, ioutfm=1,
/
----------------------------
The initial log file was deleted. The currently running log file ( still
running ) is below. This has been running for about 15 hours.

-----------------
 Checking cpin file and/or cein file.
 We are going to find DELTAGREF for a pH titratable residue without using
Replica Exchange.
 The solvent pH value is 4.500 and was loaded from the mdin file (
finddg.in<https://urldefense.proofpoint.com/v2/url?u=http-3A__finddg.in&d=DwMFaQ&c=sJ6xIWYx-zLMB3EPkvcnVg&r=vg8iTdivJL1PEwCvcH8P2DFF-Rtc9lAQvIqnaSWm1Pc&m=EcnZCsFUR8BsdZelb6CMjXxpz6p4wFO0YFYVXSwET8k&s=q4N-TBQOO5KADRfjvSxYjjVBjh3f4pSjwRIhIermaB4&e=>).
 The temperature is 300.00 K and was loaded from the mdin file (finddg.in<https://urldefense.proofpoint.com/v2/url?u=http-3A__finddg.in&d=DwMFaQ&c=sJ6xIWYx-zLMB3EPkvcnVg&r=vg8iTdivJL1PEwCvcH8P2DFF-Rtc9lAQvIqnaSWm1Pc&m=EcnZCsFUR8BsdZelb6CMjXxpz6p4wFO0YFYVXSwET8k&s=q4N-TBQOO5KADRfjvSxYjjVBjh3f4pSjwRIhIermaB4&e=>
).
 According to the mdin file (finddg.in<https://urldefense.proofpoint.com/v2/url?u=http-3A__finddg.in&d=DwMFaQ&c=sJ6xIWYx-zLMB3EPkvcnVg&r=vg8iTdivJL1PEwCvcH8P2DFF-Rtc9lAQvIqnaSWm1Pc&m=EcnZCsFUR8BsdZelb6CMjXxpz6p4wFO0YFYVXSwET8k&s=q4N-TBQOO5KADRfjvSxYjjVBjh3f4pSjwRIhIermaB4&e=>), the value of ntcnstph is 5.
 According to the mdin file (finddg.in<https://urldefense.proofpoint.com/v2/url?u=http-3A__finddg.in&d=DwMFaQ&c=sJ6xIWYx-zLMB3EPkvcnVg&r=vg8iTdivJL1PEwCvcH8P2DFF-Rtc9lAQvIqnaSWm1Pc&m=EcnZCsFUR8BsdZelb6CMjXxpz6p4wFO0YFYVXSwET8k&s=q4N-TBQOO5KADRfjvSxYjjVBjh3f4pSjwRIhIermaB4&e=>), the value of nstlim is 2000000.

 The program will try to find a range of values for DELTAGREF
automatically, as the argument -dgrefrange was not given

 AMBER execution #1: running 50 MD steps for DELTAGREF = 0.000000 kcal/mol
   The fraction of protonated species is 0.00% for the Residue 'UAP 2'
 AMBER execution #2: running 50 MD steps for DELTAGREF = -100.000000
kcal/mol
   The fraction of protonated species is 0.00% for the Residue 'UAP 2'
 AMBER execution #3: running 50 MD steps for DELTAGREF = 100.000000 kcal/mol
   The fraction of protonated species is 100.00% for the Residue 'UAP 2'

 The value of DELTAGREF in which 20% > fraction of protonated species > 80%
for solvent pH = 4.5 is between 0.000000 and 100.000000 kcal/mol.

 AMBER execution #4: running 200 MD steps for DELTAGREF = 50.000000 kcal/mol
   The fraction of protonated species is 100.00% for the Residue 'UAP 2'
 AMBER execution #5: running 200 MD steps for DELTAGREF = 25.000000 kcal/mol
   The fraction of protonated species is 100.00% for the Residue 'UAP 2'
 AMBER execution #6: running 200 MD steps for DELTAGREF = 12.500000 kcal/mol
   The fraction of protonated species is 5.10% for the Residue 'UAP 2'
 AMBER execution #7: running 200 MD steps for DELTAGREF = 18.750000 kcal/mol
   The fraction of protonated species is 100.00% for the Residue 'UAP 2'
 AMBER execution #8: running 200 MD steps for DELTAGREF = 15.625000 kcal/mol
   The fraction of protonated species is 28.20% for the Residue 'UAP 2'

 From this step on, we will make more accurate estimatives of DELTAGREF
values and we will keep increasing the number of steps.
 The convergence criterium is: 48.50% >= fraction >= 51.50% for pH = 4.5.

 AMBER execution #9: running 200000 MD steps of equilibration for
DELTAGREF = 16.182147 kcal/mol
 AMBER execution #9: running 2000000 MD steps of production for
DELTAGREF = 16.182147 kcal/mol
   The fraction of protonated species is 99.90% for the Residue 'UAP 2'
 AMBER execution #10: running 200000 MD steps of equilibration for
DELTAGREF = 12.064631 kcal/mol
 AMBER execution #10: running 2000000 MD steps of production for
DELTAGREF = 12.064631 kcal/mol
   The fraction of protonated species is 7.00% for the Residue 'UAP 2'
 AMBER execution #11: running 200000 MD steps of equilibration for
DELTAGREF = 13.606706 kcal/mol
 AMBER execution #11: running 2000000 MD steps of production for
DELTAGREF = 13.606706 kcal/mol
   The fraction of protonated species is 95.50% for the Residue 'UAP 2'
 AMBER execution #12: running 200000 MD steps of equilibration for
DELTAGREF = 11.785415 kcal/mol
 AMBER execution #12: running 2000000 MD steps of production for
DELTAGREF = 11.785415 kcal/mol
   The fraction of protonated species is 3.80% for the Residue 'UAP 2'
 AMBER execution #13: running 200000 MD steps of equilibration for
DELTAGREF = 13.711856 kcal/mol
 AMBER execution #13: running 2000000 MD steps of production for
DELTAGREF = 13.711856 kcal/mol
   The fraction of protonated species is 82.70% for the Residue 'UAP 2'
 AMBER execution #14: running 200000 MD steps of equilibration for
DELTAGREF = 12.779159 kcal/mol
 AMBER execution #14: running 2000000 MD steps of production for
DELTAGREF = 12.779159 kcal/mol
   The fraction of protonated species is 35.50% for the Residue 'UAP 2'
 AMBER execution #15: running 200000 MD steps of equilibration for
DELTAGREF = 13.135145 kcal/mol
 AMBER execution #15: running 2000000 MD steps of production for
DELTAGREF = 13.135145 kcal/mol
   The fraction of protonated species is 78.30% for the Residue 'UAP 2'
 AMBER execution #16: running 200000 MD steps of equilibration for
DELTAGREF = 12.370134 kcal/mol
 AMBER execution #16: running 2000000 MD steps of production for
DELTAGREF = 12.370134 kcal/mol
   The fraction of protonated species is 12.90% for the Residue 'UAP 2'
 AMBER execution #17: running 200000 MD steps of equilibration for
DELTAGREF = 13.508694 kcal/mol
 AMBER execution #17: running 2000000 MD steps of production for
DELTAGREF = 13.508694 kcal/mol
   The fraction of protonated species is 54.20% for the Residue 'UAP 2'
 AMBER execution #18: running 200000 MD steps of equilibration for
DELTAGREF = 13.408303 kcal/mol
 AMBER execution #18: running 2000000 MD steps of production for
DELTAGREF = 13.408303 kcal/mol
   The fraction of protonated species is 51.70% for the Residue 'UAP 2'
 AMBER execution #19: running 200000 MD steps of equilibration for
DELTAGREF = 13.367748 kcal/mol
 AMBER execution #19: running 2000000 MD steps of production for
DELTAGREF = 13.367748 kcal/mol
   The fraction of protonated species is 60.90% for the Residue 'UAP 2'
 AMBER execution #20: running 200000 MD steps of equilibration for
DELTAGREF = 13.103584 kcal/mol
 AMBER execution #20: running 2000000 MD steps of production for
DELTAGREF = 13.103584 kcal/mol
   The fraction of protonated species is 35.60% for the Residue 'UAP 2'
 AMBER execution #21: running 200000 MD steps of equilibration for
DELTAGREF = 13.456968 kcal/mol
 AMBER execution #21: running 2000000 MD steps of production for
DELTAGREF = 13.456968 kcal/mol
   The fraction of protonated species is 92.30% for the Residue 'UAP 2'
 AMBER execution #22: running 200000 MD steps of equilibration for
DELTAGREF = 11.976217 kcal/mol
 AMBER execution #22: running 2000000 MD steps of production for
DELTAGREF = 11.976217 kcal/mol



Thanks
Santo


On Mon, 20 Jan 2020 at 17:27, Cruzeiro,Vinicius Wilian D <vwcruzeiro.ufl.edu<mailto:vwcruzeiro.ufl.edu>>
wrote:

> Hello Santo,
>
> May you send the mdin file you are using in the simulation and your
> current finddg.log file?
>
> Thanks,
>
>
> Vinícius Wilian D Cruzeiro
>
> PhD Candidate
> Department of Chemistry, Physical Chemistry Division
> University of Florida, United States
>
> Voice: +1(352)846-1633<tel:+1(352)846-1633>
>
> On Jan 20, 2020, at 12:42 PM, Kolattukudy P. Santo <
> santotheophys.gmail.com<mailto:santotheophys.gmail.com><mailto:santotheophys.gmail.com<mailto:santotheophys.gmail.com>>> wrote:
>
> finddg.log
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org<mailto:AMBER.ambermd.org>
> https://urldefense.proofpoint.com/v2/url?u=http-3A__lists.ambermd.org_mailman_listinfo_amber&d=DwIGaQ&c=sJ6xIWYx-zLMB3EPkvcnVg&r=vg8iTdivJL1PEwCvcH8P2DFF-Rtc9lAQvIqnaSWm1Pc&m=cXy6VPDsXJaEMYpImZJu4R0K7Ri2m1idUh27vINEsVs&s=athAkYfSK3HAZiHL5IbmbwIO8OMH4u1ukbIHGrjkn2U&e=
>


--
*Dr. Kolattukudy P.  Santo*
*Post doctoral Associate *
*Department of Chemical and Biochemical Engineering*
*Rutgers, The State University of New Jersey*
*New Brunswick, New jersey*
*USA*
_______________________________________________
AMBER mailing list
AMBER.ambermd.org<mailto:AMBER.ambermd.org>
https://urldefense.proofpoint.com/v2/url?u=http-3A__lists.ambermd.org_mailman_listinfo_amber&d=DwIGaQ&c=sJ6xIWYx-zLMB3EPkvcnVg&r=vg8iTdivJL1PEwCvcH8P2DFF-Rtc9lAQvIqnaSWm1Pc&m=cXy6VPDsXJaEMYpImZJu4R0K7Ri2m1idUh27vINEsVs&s=athAkYfSK3HAZiHL5IbmbwIO8OMH4u1ukbIHGrjkn2U&e=
--
Dr. Kolattukudy P.  Santo
Post doctoral Associate
Department of Chemical and Biochemical Engineering
Rutgers, The State University of New Jersey
New Brunswick, New jersey
USA
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Jan 23 2020 - 15:00:02 PST
Custom Search