Re: [AMBER] Progress with testing GB-neck2 MD with nucleic acids

From: Hai Nguyen <nhai.qn.gmail.com>
Date: Thu, 30 Jun 2016 16:48:16 -0400

On Thu, Jun 30, 2016 at 4:16 PM, Kasprzak, Wojciech (NIH/NCI) [C] <
kasprzaw.mail.nih.gov> wrote:

> Carlos, Hai,
>
> I believe things are getting sorted out on our side. A clean login to the
> server and fresh initializations of all paths
> got the right binary running. I do see the mbondi3 section in the prmtop
> file now, and I do see the output
> containing b_alpha_hnu, gb_beta_hnu, etc. (The updates were installed in
> a temporary area, in parallel with
> the "production" binaries, hence some confusion). The product of
> minimization also looks clean.
> The equilibration steps are under way now. Thank you very much for
> immediate help and hand-holding!
>

cool. It's great that we spotted to the right point.


>
> Regarding equilibration protocol, ours looks like the one described in
> your paper, with gradually diminishing
> restraints going down to 0.1 kcal/mol/A^2 before un-restrained MD begins,
> but I am curious if
> you are using and/or recommending varying gamma_ln (for the ntt=3) at
> various stages of it. What
> setting do you use?


varying `gamma_ln` might be working for other GB models but I think for
igb8 that does not matter much (from my extensive testing)
We normally use gamma_ln=1.

But you may need to vary gamma_ln for your system., which is much more
complicated that the one we tested (canonical RNA, DNA duplexes, RNA/RNA
hairpins (folding))
Per production run, lower gamma_ln can accelerate the conformation sampling
more (http://www.ncbi.nlm.nih.gov/pubmed/25762327) but I have never tried
myself.


> Also, is gbsa option any more important to GB-neck2 than before? Setting
> gbsa=1 slows
> down my runs dramatically while the listed values are very small. My
> current test run is with gbsa=0.
>

Yes, you should always use gbsa=0 with igb8 unless there would be much
accurate and faster (GPU) gbsa model.
(If you're interested in protein folding, please check our paper with
igb8/gbsa=0 with millisecond aggregated data
http://pubs.acs.org/doi/abs/10.1021/ja5032776)


Let me know if I miss any question.

cheers

Hai


> To clarify my past e-mails, I am dealing with DNA and RNA nanocubes. In
> GB-HCT DNA cubes looked more
> stable than the RNA cubes, and I am trying to see if GB-neck2 can improve
> the results. Both are quite stable
> in explicit solvent simulations (PME, TIP3P)
>


>
> Thank you very much again!
> Best regards, Voytek Kasprzak
>
> Wojciech (Voytek) Kasprzak (Contractor)
> Analyst Programmer,
> Basic Science Program,
> Leidos Biomedical Research, Inc.
> Frederick National Laboratory for Cancer Research (FNLCR)
> Frederick, MD 21702
> (301) 846 5537
> http://binkley2.ncifcrf.gov/users/kasprzak
>
> ________________________________________
> From: Hai Nguyen [nhai.qn.gmail.com]
> Sent: Thursday, June 30, 2016 3:23 PM
> To: AMBER Mailing List
> Subject: Re: [AMBER] GB-neck2 MD with nucleic acids
>
> On Thu, Jun 30, 2016 at 3:09 PM, Kasprzak, Wojciech (NIH/NCI) [C] <
> kasprzaw.mail.nih.gov> wrote:
>
> > Hai,
> >
> > I am trying to compare the GB-neck2 protocol to the GB-HCT (for which I
> > see decent, but
> > not perfect results, as would be expected). Based on your 2015 paper, the
> > results should
> > improve with the new mbondi3/igb=8. The fact that my RNA nanocube begins
> > to fall apart
> > in minimization and does not even look like RNA after a few equilibration
> > steps makes me
> > suspicious about the parameters read-into leap and/or binary I am using.
> >
> >
>
> > I do not see any of the gb_alpha_hnu = 0.53705, gb_beta_hnu =
> > 0.36286,
> > gb_gamma_hnu = 0.11670 in the output file, but it is not clear to me if
> > things failed in leap or
> > are ignored in pmemd.cuda.
> >
>
> This indicates that you did not use updated igb8 parameters for nucleic.
> Please tell your admin to recompile AMBER after applying updates.
>
>
>
> > Is there something I could check in the prmtop or inpcrd files created in
> > tleap to isolate the issue
> > better?
> >
>
> No, there is no special thing about tleap with igb8, except that you need
> to use mbondi3.
> You can use vim to open your prmtop and search for mbondi3 keyword. If you
> found it, it's good to go.
> But please recompile AMBER code and
> make sure you DO see gb_alpha_hnu, gb_beta_hnu, ... nucleic acid parameters
> in your mdout.
>
> Hai
>
>
> >
> > Thank you, Voytek
> >
> > Wojciech (Voytek) Kasprzak (Contractor)
> > Analyst Programmer,
> > Basic Science Program,
> > Leidos Biomedical Research, Inc.
> > Frederick National Laboratory for Cancer Research (FNLCR)
> > Frederick, MD 21702
> > (301) 846 5537
> > http://binkley2.ncifcrf.gov/users/kasprzak
> >
> > ________________________________________
> > From: Hai Nguyen [nhai.qn.gmail.com]
> > Sent: Thursday, June 30, 2016 2:48 PM
> > To: AMBER Mailing List
> > Subject: Re: [AMBER] GB-neck2 MD with nucleic acids
> >
> > Hi
> >
> > On Thu, Jun 30, 2016 at 2:29 PM, Kasprzak, Wojciech (NIH/NCI) [C] <
> > kasprzaw.mail.nih.gov> wrote:
> >
> > > Dear Amber Users,
> > >
> > > Following the advice from Carlos and Hai on the latest GB-neck2 option
> > > for running nuclei acids simulations with
> > > implicit solvent, our applications admin updated Amber14 to update.12
> > > (and more), but when I used tleap and ran
> > > a quick equilibration, the RNA fell apart immediately, indicating wrong
> > > parameters.
> > >
> > > In response to the tleap command "source leaprc.ff14SB" I see the
> ususal
> > > libraries for RNA, a list of *12.lib files and
> > > atomic_ions.lib and solvents.lib.
> > >
> > > Command " set default PBradii mbondi3" response is "Using ArgH and
> > > AspGlu0 modified Bondi2 radii." There is not hint
> > > of any nucleic acid-specific modifications.
> >
> >
> > Sorry, I forgot to mention that mbondi2 and mbondi3 are identical for
> > nucleic acid (only different for protein as you saw). So you did
> correctly.
> >
> >
> >
> > > GB was run with igb=8, ntt=3, gamma_ln=1.0 and saltcon=1.0. I
> > understand
> > > that offset does not need to be specified (unlike in the GB-HCT
> > protocol).
> > >
> > >
> > Yes, users should always use default igb=8 parameters.
> >
> > Do you see below line in your mdout (to make sure you actually use
> updated
> > code)
> >
> > gb_alpha_hnu = 0.53705, gb_beta_hnu = 0.36286, gb_gamma_hnu =
> > 0.11670
> >
> > If yes, how does your RNA look like?
> > Simulation of nucleic acid in implicit solvent is very challenging, so it
> > is likely that igb=8 does not do a good job here.
> >
> > Hai
> >
> >
> >
> > _______________________________________________
> > 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
>
> _______________________________________________
> 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 Jun 30 2016 - 14:00:02 PDT
Custom Search