Re: [AMBER] TI on pmemd CUDA, AMBER18

From: Hosein Geraili Daronkola <geraili.hsn.gmail.com>
Date: Fri, 21 Feb 2020 10:57:41 +0100

Hi all,

I changed the setup a bit, and it works better. I am not using softcore for
the charge and decharge state anymore. My question is what is the main
criteria for the TI results? I did it for three setups (exactly similar)
and the results are close (+-1 kcal), and the curves of charge and
decharge doesn't have any kinks, but still the O_matrix has some empty
squares, and also the vdW part has some kinks. Are the overlapping matrix
from alchemical_analysis for the TI of amber with the MBAR reliable, or a
good validation? Or just the curves are enough? I simulated these for 60 ns
with backbone restrained.

I also added the result for the first 30 ns of decharge state for the
lambda 1.

Best regards

 dhdl_TI_charge.pdf
<https://drive.google.com/file/d/1zlZ5-GsFXDJ7upul92GnxD24NnltVfyI/view?usp=drive_web>
 dhdl_TI_decharge.pdf
<https://drive.google.com/file/d/1g8rt1ko0KwO7vSa_dHD1m2bsWymLJ2b4/view?usp=drive_web>
 dhdl_TI_vdw.pdf
<https://drive.google.com/file/d/1q9JE73k9S4vMzUVXJWT-cfN-V0kyZHC5/view?usp=drive_web>
 O_MBAR_charge.pdf
<https://drive.google.com/file/d/11zCAuIF2zW1PUsY-W5AodwwOFsyqsUN7/view?usp=drive_web>
 O_MBAR_decharge.pdf
<https://drive.google.com/file/d/1-9moODrIAFoyNQZqYWK2SdTw7Vmn3_qW/view?usp=drive_web>
 O_MBAR_vdw.pdf
<https://drive.google.com/file/d/1c7jbs_Edi6_mfmwOx6VU21XA2_zn-qHO/view?usp=drive_web>
 ti001.out
<https://drive.google.com/file/d/1Z1WBp1fJ7blQNrAJPgtcQC-rPhVCqTVK/view?usp=drive_web>




On Tue, Feb 11, 2020 at 3:39 PM Taisung Lee <accuratefreeenergy.gmail.com>
wrote:

> Hi Hosein,
>
> Without all input files (crd, prmtop, mdin), it is hard to verify what
> is really wrong. Nevertheless, it seems to me the problem could be the
> definition of de/re-charge regions. Improper choices of the de/re-charge
> atoms could lead to the well-known end-point problems--it seems it is the
> case in your du/dl curves. One easy thing to try is change sc_beta to a
> larger value (say 17) and see if there is any difference. A yes would be
> strong evidence of the end-point problem.
>
> If possible, could you post all your input files and at least one output
> file, say the lambda=1 result of the de-charge step?
>
> Thanks!
>
> Taisung
>
>
> On Mon, Feb 10, 2020 at 5:14 AM Hosein Geraili Daronkola <
> geraili.hsn.gmail.com> wrote:
>
> > Hi all,
> >
> > What can be possibly wrong in this, that it never works? A mutation from
> > Asp to Asn with 21 lambdas, and separated to three steps of charging Asn,
> > decharging Asp, and vdW. It's already equilibrated for sure. I am using
> the
> > "alchemical_analysis" to get the result, and the O_MBAR and the TI graph
> > are almost smooth and good for vdW part, but the result for the
> > decharge and charge is attached and not acceptable. What did I do wrong?
> >
> >
> >
> > This is the simulation details for the decharge of Asp:
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > *TI simulation &cntrl imin = 0, nstlim = 15000000, irest = 1, ntx = 5,
> dt
> > = 0.001, ntt = 3, temp0 = 298.0, gamma_ln = 5.0, ig = -1, !ntc = 2,
> ntf
> > = 1, cut=10, ntp = 1, pres0 = 1.0, taup = 2, ioutfm = 1, iwrap =
> 1,
> > ntwe = 500, ntwx = 50000, ntpr = 500, ntwr = 500, icfe = 1, clambda =
> > 0.20, scalpha = 0.5, scbeta = 12.0, logdvdl = 0, ifmbar = 1,
> > mbar_states = 21, mbar_lambda = 0.00, 0.05, 0.10, 0.15, 0.20, 0.25,
> 0.30,
> > 0.35, 0.40, 0.45, 0.50, 0.55, 0.60, 0.65, 0.70, 0.75, 0.80, 0.85, 0.90,
> > 0.95, 1.00, nmropt = 1, ntr = 1, restraint_wt = 50.0,
> > restraintmask='!(:WAT,K+,Cl-) & (@N,CA,C,O)', timask1 = ':34',
> timask2 =
> > ':130', ifsc=1, scmask1=':34.CB,HB2,HB3,CG,OD1,OD2',
> > scmask2=':130.CB,HB2,HB3,CG,OD1,OD2', crgmask =
> > ':34.CB,HB2,HB3,CG,OD1,OD2', !method 2: !scmask1 = '', scmask2 =
> > ':163.HG=' / &ewald skinnb = 5, dsum_tol = 1E-06, / &wt type = 'END' /*
> >
> >
> >
> >
> > best regards
> > _______________________________________________
> > 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
>


-- 
Hosein Geraili Daronkola
Ph.D. student,
Wissenschaftspark Potsdam-Golm
Max Planck Institute of Colloids and Interfaces
Theory and Bio-Systems Department
Office K-1.119
Am M├╝hlenberg 1 OT Golm
14476 Potsdam
Germany
Phone: +49-(0)331 567-9611
Fax: +49-(0)331 567-9602
E-mail: hosein.geraili.mpikg.mpg.de
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Feb 21 2020 - 02:00:02 PST
Custom Search