Hi,
I just checked the MD output from your original post, and I have a few
comments that hopefully will help. I'll go in the order they appear in
the MDOUT file:
> ig = 1111,
This is dangerous. Unless you have a really good reason for doing so,
do *not* explicitly set a random seed. If you forget to change it in
successive runs you can end up with a sort of artificial resonance
that can mess up your system (see e.g.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2632580/ and related
work).
> tempi = 0.0, ! Initial temperature
> temp0 = 50.0, ! Target temperature
> gamma_ln = 1.0, ! Friction coefficient (ps^-1)
This is more of a recommendation, but I think a heating stage is
really unnecessary (unless you need it for some reason). You're
setting all your velocities to zero and asking the thermostat (with a
very small gamma!) to do a lot of heavy lifting in heating up your
system. I recommend just setting tempi to whatever temp0 will be and
use whatever random velocities get assigned. If you're looking for a
system relaxation protocol for MD I can recommend one :-)
https://aip.scitation.org/doi/abs/10.1063/5.0013849
> ntb = 1, ! pressure regulation (1: costant surface on yz plane)
This is more of a correction; ntb just refers to the kinds of boundary
conditions, not pressure regulation (0=none, 1 = NVT, 2 = NPT).
> ntr = 1,
> restraint_wt = 75.0,
> restraintmask ='((:1-1809)|(.Na+,Cl-))&!.H=', ! Mask to restrain during equili
This is where I think you may be getting into trouble. That force
constant is huge! Also, why are you restraining your ions? I could see
this potentially causing issues particularly with the SPFP code.
I recommend giving the relaxation protocol a try (there's even a
helpful companion script: https://github.com/drroe/AmberMdPrep). At
the very least, try using a smaller force constant and don't restrain
your ions (unless you really need to for some reason). Also, try doing
your initial relaxation with the CPU code (it is less susceptible to
overflows and will catch issues due to large velocities, i.e. "vlimit"
errors).
Hope this helps,
-Dan
On Wed, Apr 21, 2021 at 4:39 PM Neville Bethel <nevillebethel.gmail.com> wrote:
>
> Hi Kenneth,
>
> Yes, I have tried that. I've also turned iwrap=0 and the exact same issue
> is still there. The area sticking out of the box also distorts locally
> before shooting out over 100 nanometers, despite being restrained. The
> final distortion is also much larger than the unit cell length (see
> attached), so it's not just popping into an adjacent cell.
> The energies in the mdout file also spikes right before the protein heavily
> distorts.
> I have dealt with imaging issues in the past, and I've never seen anything
> quite like this.
>
>
>
> NSTEP = 2 TIME(PS) = 0.004 TEMP(K) = 6.93 PRESS =
> 0.0
>
> Etot = -1233660.5161 EKtot = 4504.6196 EPtot =
> -1238165.1357
>
> BOND = 876.6350 ANGLE = 3953.7721 DIHED =
> 6045.6337
>
> UB = 0.0000 IMP = 0.0000 CMAP =
> 2363.2836
>
> 1-4 NB = 7579.1417 1-4 EEL = 78180.4717 VDWAALS =
> 109000.5146
>
> EELEC = -1446165.1044 EHBOND = 0.0000 RESTRAINT =
> 0.5163
>
> EAMBER (non-restraint) = -1238165.6520
>
>
> ------------------------------------------------------------------------------
>
>
>
> NSTEP = 3 TIME(PS) = 0.006 TEMP(K) = 25.26 PRESS =
> 0.0
>
> Etot = -1233480.5499 EKtot = 16418.8125 EPtot =
> -1249899.3624
>
> BOND = 853.5182 ANGLE = 3654.1325 DIHED =
> 6041.8623
>
> UB = 0.0000 IMP = 0.0000 CMAP =
> 2351.2961
>
> 1-4 NB = 7507.8397 1-4 EEL = 78040.9065 VDWAALS =
> 108808.1071
>
> EELEC = -1457163.4803 EHBOND = 0.0000 RESTRAINT =
> 6.4556
>
> EAMBER (non-restraint) = -1249905.8180
>
>
> ------------------------------------------------------------------------------
>
>
>
> NSTEP = 4 TIME(PS) = 0.008 TEMP(K) =********* PRESS =
> 0.0
>
> Etot = ************** EKtot = ************** EPtot =
> -1091938.3276
>
> BOND = 813.7245 ANGLE = 3606.8652 DIHED =
> 6049.2609
>
> UB = 0.0000 IMP = 0.0000 CMAP =
> 2334.4278
>
> 1-4 NB = 11546.2703 1-4 EEL = 77840.2321 VDWAALS =
> 278138.1904
>
> EELEC = -1472294.4335 EHBOND = 0.0000 RESTRAINT =
> 27.1349
>
> EAMBER (non-restraint) = -1091965.4625
>
>
> ------------------------------------------------------------------------------
>
>
> Date: Tue, 20 Apr 2021 15:32:01 -0400
> From: Kenneth Huang <kennethneltharion.gmail.com>
> Subject: Re: [AMBER] protein sticking out of solvation box causing
> instability
> To: AMBER Mailing List <amber.ambermd.org>
> Message-ID:
> <CALeh7kDVpL-ozdoYbXeFvb1CY4-Lto6+_FR=P1uRzhxLBShL3Q.mail.gmail.com>
> Content-Type: text/plain; charset="UTF-8"
>
> Hi,
>
> It looks like an imaging issue- have you tried looking at it after imaging
> it with something like
>
> trajin equil.rst
> autoimage
> trajout test.pdb
>
> Alternatively, I think VMD can draw the other unit cells with an option
> under Graphics/Representations, since your protein tail is likely just
> sitting in the box next to the orginal.
>
> Best,
>
> Kenneth
>
> On Tue, Apr 20, 2021 at 1:36 PM Neville Bethel <nevillebethel.gmail.com>
> wrote:
>
> > Hi everyone,
> >
> > I came across an issue I haven't seen before.
> > I am trying to solvate and equilibrate a large triangular protein.
> > tleap seems to put one of the ends of the protein outside the original
> > solvation box (trimer_example_0.png attached).
> > This seems fine since there is still plenty of space between this end and
> > the next periodic image, but when I try to equilibrate, this edge of the
> > protein flies off into space (trimer_example_8.png).
> >
> > I also attached the output text from the equilibration run. Does anyone
> > know what's going on here?
> >
> > Thanks,
> > Neville
> >
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Apr 22 2021 - 07:00:02 PDT
