AMBER Archive (2006)

Subject: Re: AMBER: problem with calcpka.pl

From: Amber admin (amber-admin_at_scripps.edu)
Date: Tue Oct 31 2006 - 18:33:09 CST


Date: Mon, 30 Oct 2006 20:06:35 -0800
From: "John Mongan"
To: amber_at_scripps.edu
Subject: Re: AMBER: problem with calcpka.pl

The "stabilization" isn't an equilibration, it's a statistical
convergence when you have sufficient sampling. So it doesn't matter
how long you simulate prior to your sampling period, you still need a
fairly extensive sampling period. You should also keep in mind that if
different portions of your conformational ensemble have different
micro-pKas for your residues (this is frequently the case) you won't
get good results unless your sampling period is long enough to have
converged sampling over the relevant portions of your conformational
ensemble. In summary: 1 ns of sampling is probably barely enough if
you have a fairly rigid system, likely not nearly enough if your
system is flexible. And simulation at a single pH value is adequate
only if you're sure you have Henderson-Hasslebalch titration curves
for all your residues of interest *and* their pKa's are all pretty
close to the simulation pH value.

Hope this helps,

John

On 10/30/06, Benjamin Juhl <itbbju_at_itb.uni-stuttgart.de> wrote:
> Dear John,
>
> in your paper, you wrote that for large systems like HEWL, you performed
> titration of 1 ns to allow for stabilization of predicted pKa values. So
> prior to the analysed 100 ps, i simulated 1.25 ns at the pH of interest
> (pH 7 for testing) to reach that stabilization. so the 100 ps are the
> timeframe from 1.25 to 1.35 ns. I will run another 900 ps if the 100 ps
> of analyzed simulation time are the problem, but the protonation states
> after 1.35 ns of simulation were still way off from expected values, so
> i wondered if there was a methodological error in what i was doing.
>
> Thanks for your help,
> Benjamin
>
>
> David A. Case schrieb:
> >> Date: Wed, 25 Oct 2006 17:14:20 -0700
> >> From: "John Mongan"
> >> Subject: Re: AMBER: problem with calcpka.pl
> >>
> >> You seem to have run 100 ps of simulation at a single pH value. This
> >> is not nearly sufficient sampling to get optimum predictions, even
> >> within the (significant) limitations of the force-field and GB model.
> >> The sampling regimen (1 ns at every 0.5 pH across the pKa range of
> >> interest) used in the JCC paper referenced in the manual should
> >> probably be considered a bare minimum for sampling, and more would
> >> certainly be needed if your system is particularly flexible.
> >>
> >> Constant pH methods are still somewhat in their infancy and there are
> >> many potential pitfalls. If you haven't read the paper describing the
> >> method, I would strongly recommend that you do so prior to investing
> >> much time with the method (though as the author I'm somewhat
> >> biased...)
> >>
> >> Hope this helps,
> >>
> >> John
> >>
> >> On 10/24/06, Benjamin Juhl <itbbju_at_itb.uni-stuttgart.de> wrote:
> >>
> >>> Dear all,
> >>>
> >>>
> >>> i was interested on the effect of different pH values on a lipase i am
> >>> studying and wanted to use the simulations at constant pH as described
> >>> in the manual (chapter 6.11 - Amber9).
> >>> When i want to analyse the protonation states of some residues from the
> >>> cpout files with calcpka.pl i get following output at a ph=7:
> >>>
> >>> AS4 6: Offset 1.549 Pred 8.549 Frac Prot 0.973 Transitions 2
> >>> AS4 17: Offset 1.546 Pred 8.546 Frac Prot 0.972 Transitions 2
> >>> AS4 49: Offset 1.966 Pred 8.966 Frac Prot 0.989 Transitions 16
> >>> AS4 75: Offset 1.514 Pred 8.514 Frac Prot 0.970 Transitions 6
> >>> GL4 81: Offset 2.157 Pred 9.157 Frac Prot 0.993 Transitions 12
> >>> AS4 126: Offset -2.558 Pred 4.442 Frac Prot 0.003 Transitions 2
> >>> AS4 134: Offset 2.299 Pred 9.299 Frac Prot 0.995 Transitions 2
> >>> AS4 145: Offset 2.467 Pred 9.467 Frac Prot 0.997 Transitions 2
> >>> AS4 187: Offset -Inf Pred -Inf Frac Prot 0.000 Transitions 0
> >>> GL4 188: Offset 1.095 Pred 8.095 Frac Prot 0.926 Transitions 55
> >>> AS4 200: Offset 2.162 Pred 9.162 Frac Prot 0.993 Transitions 2
> >>> AS4 223: Offset 0.656 Pred 7.656 Frac Prot 0.819 Transitions 94
> >>> HIP 224: Offset 0.980 Pred 7.980 Frac Prot 0.905 Transitions 84
> >>> AS4 252: Offset 1.139 Pred 8.139 Frac Prot 0.932 Transitions 40
> >>> AS4 257: Offset 1.851 Pred 8.851 Frac Prot 0.986 Transitions 20
> >>> AS4 265: Offset 0.578 Pred 7.578 Frac Prot 0.791 Transitions 116
> >>> GL4 269: Offset 1.191 Pred 8.191 Frac Prot 0.939 Transitions 48
> >>> GL4 294: Offset -1.982 Pred 5.018 Frac Prot 0.010 Transitions 6
> >>> AS4 296: Offset 1.253 Pred 8.253 Frac Prot 0.947 Transitions 34
> >>> Can't call method "ResName" on an undefined value at
> >>> /share/apps/amber9_lampath/exe/calcpka.pl line 65.
> >>>
> >>> The shown pKa values are very unrealistic and i wonder if this could be
> >>> due to the error in the calcpka.pl script and would appreciate any help.
> >>>
> >>>
> >>> I used the following input file for the analysed part of the simulation:
> >>> &cntrl
> >>> igb=1, cut=30.0, ig = 12703,
> >>> imin=0, ntx=5, irest=1, ntpr=100, ntwx=200, ntwr=100,
> >>> ntc=2, tol=0.00000001, ntb=0, ntp=0,
> >>> tautp=2.0,
> >>> nstlim=100000, dt=0.001, ntr=0,
> >>> temp0=300.0, ntt=1,
> >>> ntf=2,
> >>> icnstph = 1, solvph = 7, ntcnstph = 2,
> >>> scee = 1.2, saltcon = 0.1, nrespa = 1,
> >>> /
> >>>
> >>>
> >>> Thanks,
> >>> Benjamin Juhl
> >>> -----------------------------------------------------------------------
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu