[gmx-users] REST simulation

Patrick Fuchs patrick.fuchs at univ-paris-diderot.fr
Tue Dec 13 14:25:38 CET 2011


Hi Otto,
do you know how to use the free energy code in GROMACS for running 
alchemical transformations (such as thermodynamic integration)? I 
strongly suggest first to be confortable with that code before trying to 
implement REST. In such alchemical calculations, you have to specify two 
topologies, one for state A and one for state B. Then simulations at 
each lambda value is a linear interpolation between state A and state B, 
such as H = (1-lambda)*H[A] + lambda*H[B]. In the case of REST, one 
simulation at an intermediate lambda value is also a linear 
interpolation between topology A (ref T, unscaled Hamiltonian) and 
topology B (highest T, scaled Hamiltonian). You thus have to build a 
topology for both A *and* B (how to do that is well described in the 
manual). Once you have this "dual" topology, you just have to set lambda 
for each replica.
Ciao,

Patrick

Le 13/12/2011 14:05, Otto Master a écrit :
> Hi Patrick,
> again thanks a lot for your valuable help. But I do not get my head
> around how to tell grompp that for lambda=0 I would like to use the the
> unscaled Hamiltonian and for lambda=1 the highest "temperature". In the
> tutorial the change in temperature is a number which is passed to the
> pre-processor, but for REST a number (lambda=0 and lambda=1) corresponds
> to different topology, which are written in a file. I do not know how to
> specify that. What is the identifier or should I pass something
> additionally to grompp, referring to the two topology files?
>
> Ciao,
> Otto
>
> On Tue, Dec 13, 2011 at 1:23 PM, Patrick Fuchs
> <patrick.fuchs at univ-paris-diderot.fr
> <mailto:patrick.fuchs at univ-paris-diderot.fr>> wrote:
>
>     Hi Otto,
>     you have to equilibrate at each lambda value! The unscaled
>     Hamiltionian is your lowest "temperature", say 300K, which
>     corresponds to lambda=0. You generate the highest "temperature" by
>     appropriately scaling the Hamiltonian, say 600K, which corresponds
>     to lambda=1. Then you create n directories, one for each replica. In
>     each directory, you set a different lambda value from 0 to 1 in the
>     mdp file and you equilibrate each replica. At the end, you obtain n
>     gro files that can serve as input to generate n tpr files for the
>     REMD run.
>     I think this is the standard flow for REMD as explained in:
>     http://www.gromacs.org/__Documentation/How-tos/REMD
>     <http://www.gromacs.org/Documentation/How-tos/REMD>. The only
>     difference is that you use a different lambda value, so a different
>     Hamiltonian, instead of a different temperature for each replica.
>     Ciao,
>
>     Patrick
>
>     Le 13/12/2011 11:46, Otto Master a écrit :
>
>         Hi Patrick,
>
>         thanks for your help. What I still do not understand is, how I can
>         set-up the replica simulation starting from the two equilibrated
>         systems. What do I have to put into the .mdp file and in the grompp
>         command to consider the two equilibrated configurations and further
>         obtain the tpr files for the different replica for different lambda
>         values to interpolate between the two configuration. I would be very
>         glad if you could help me on that.
>
>         All the best
>         Otto
>
>         On Mon, Dec 12, 2011 at 1:50 PM, Patrick Fuchs
>         <patrick.fuchs at univ-paris-__diderot.fr
>         <mailto:patrick.fuchs at univ-paris-diderot.fr>
>         <mailto:patrick.fuchs at univ-__paris-diderot.fr
>         <mailto:patrick.fuchs at univ-paris-diderot.fr>>> wrote:
>
>             Hi Otto,
>             yes I copied those two files you mentionned (also .rtp for
>         charges)
>             in some specific directory to apply the appropriate scaling. But
>             according to the authors this REST implementation, you just need
>             that for the highest "temperature" (for the lowest, the
>         Hamiltonian
>             is unchanged) and then intermediate "temperatures" are
>         interpolated
>             using the lambda factor. So for equilibrating each replica,
>         you just
>             need to set the appropriate lambda value.
>             Now I'd consider Mark's advice to use the -pp flag of grompp
>         which
>             might be convenient for scripting the scaling of the potential.
>             Ciao,
>
>             Patrick
>
>             Le 12/12/2011 12:56, Otto Master a écrit :
>
>                 Hi Patrick,
>                 Thanks a lot for your reply. Just to be sure, you create for
>                 every replicate a copy of the original force field, and
>         after
>                 you manipulate the parameter in ffnonbonded.itp and
>                 ffbonded.itp. Then you go for each replicate through the
>         usual
>                 simulation preparation steps (minimisation, nvt,
>         equilibration
>                 ...). The result of this you use for the replicate exchange
>                 simulation.
>
>                 Thanks a lot
>                 Otto
>
>                 On 12 Dec 2011, at 10:42, Patrick
>                 Fuchs<patrick.fuchs at univ-____paris-diderot.fr
>         <mailto:patrick.fuchs at univ-__paris-diderot.fr>
>         <mailto:patrick.fuchs at univ-__paris-diderot.fr
>         <mailto:patrick.fuchs at univ-paris-diderot.fr>>>  wrote:
>
>
>                     Hi Otto,
>                     in my lab we tried to implement this REST variant in
>         GROMACS
>                     as proposed by those authors. We figured out that it was
>                     easier to manipulate directly the parameters files
>         in the
>                     top directory. There you know exactly what you are
>         doing;
>                     recall that some interactions (i.e. solvent/solvent)
>         mustn't
>                     be scaled whereas some others have to be scaled
>                     (solute/solute and solute/solvent).
>                     It's probably possible to do it in the tpr file, but it
>                     looked less trivial to me: i) you have to know how
>         atoms are
>                     coded in the file (e.g. in the functype[???]=LJ_SR[...]
>                     matrix, you have to understand how atom numbers are
>         coded
>                     there), ii) you have to regenerate a tpr from plain text
>                     file; it's probably doable, but I don't know how.
>         Actually,
>                     maybe some developers can tell if it's possible.
>                     Good luck,
>
>                     Patrick
>
>                     Le 08/12/2011 19:01, Otto Master a écrit :
>
>                         Dear gromacs users,
>
>                         Recently I stumbled over following paper:
>                         T. Terakawa, T. Kameda, and S. Takada, On Easy
>                         Implementation of a
>                         Variant of the Replica Exchange with Solute
>         Tempering in
>                         GROMACS.
>                         Journal of Computational Chemistry 32 (2011)
>         1228-1234.
>
>                         The authors suggested an easy way to run this
>         kind of
>                         simulation with
>                         Gromacs, without even changing the code. The
>         only thing
>                         that is need, is
>                         the the rescaling of the parameters in the parameter
>                         file. Since the
>                         reduction of the replica number is quite
>         appealing to me
>                         I wonder which
>                         file I have to change? Actually, I thought of
>                         manipulating the .tpr file
>                         or to rescale and creating the force fields for
>         every
>                         replicate. Is this
>                         feasible, or is there a better way?
>
>                         Manipulating the .tpr file could be easier, since it
>                         unifies (right?)
>                         the parameters from the different force fields,
>         before
>                         sending it to the
>                         mdrun application. But for this I would like to
>                         understand the tpr file
>                         first.There are quite a lot of entries and first
>         I try
>                         to understand LJ
>                         interactions and how they are defined in this
>         file. I
>                         found two entries
>
>                         LJ14
>                                   functype[154]=LJ14, c6A=
>         0.00000000e+00, c12A=
>                         0.00000000e+00,
>                         c6B= 0.00000000e+00, c12B= 0.00000000e+00
>                                   functype[155]=LJ14, c6A=
>         4.46680887e-03, c12A=
>                         4.74702711e-06,
>                         c6B= 4.46680887e-03, c12B= 4.74702711e-06
>
>                         which corresponds to following interactions
>
>                                LJ-14:
>                                   nr: 876
>                                   iatoms:
>                                      0 type=154 (LJ14) 0 4
>                                      1 type=155 (LJ14) 0 5
>
>                         When I tried to calculate the parameters from the
>                         combination rules (in
>                         this case Gromos 53A6 force field), I found (the
>                         highlighted columns
>                         contain the original parameters for the specific
>         atom
>                         groups from the
>                         Gromos  documentation and the calculated value for
>                         combining the two
>                         parameters:
>
>
>
>
>                             sqrt(C6i) (from ff)    sqrt(C6j) (from ff)
>                           sqrt(C6i)*sqrt(C6j)    value
>                         from tpr file
>                         functype[154]=LJ14,    c6A=    CH3    H
>           0.09805    0
>                             0    0.00E+00
>                         functype[155]=LJ14,    c6A=    CH3    CH1    0.09805
>                           0.0779    0.007638095    4.47E-03
>                         functype[156]=LJ14,    c6A=    C    CH2    0.04838
>                           0.08642    0.004181    3.33E-03
>                         functype[157]=LJ14,    c6A=    C    C    0.04838
>                           0.04838    0.002340624    2.34E-03
>
>
>                         The values for N, C, O, H seems to be OK, but I have
>                         problems to get the
>                         same value, when CH1, CH2, CH3 are involved.
>         Since I do
>                         not have too
>                         much experience, I would like to know how the
>         value from
>                         the .tpr file
>                         can be derived.
>
>                         The other entry for LJ potential is the short
>         range term
>                         LJ_SR (.tpr file
>
>                             ffparams:
>                                atnr=11
>                                ntypes=170
>                                   functype[0]=LJ_SR, c6= 9.61380266e-03,
>         c12=
>                         2.66462448e-05
>                                   functype[1]=LJ_SR, c6= 4.74365894e-03,
>         c12=
>                         1.14699596e-05
>                                   functype[2]=LJ_SR, c6= 4.66325786e-03,
>         c12=
>                         5.16199998e-06
>
>                         Unfortunately, I do not find the section where the
>                         function is assigned
>                         to a specific pair of interaction. Where are these
>                         functions assigned to
>                         a specific interaction? Furthermore, is it
>         possible to
>                         distinguish
>                         between intra-nonbonded (solute-solute) and
>         inter-bonded
>                         (water-solute)
>                         interaction?
>
>                         For you this might be an easy question to
>         answer, and
>                         you immediately
>                         realize there is a beginner at work, but
>         nevertheless I
>                         would appreciate
>                         any help.
>
>                         All the best
>                         Otto
>
>
>
>                     --
>
>           _______________________________________________________________________________
>
>                     Patrick FUCHS
>                     Dynamique des Structures et Interactions des
>         Macromolécules
>                     Biologiques
>                     INTS, INSERM UMR-S665, Université Paris Diderot,
>                     6 rue Alexandre Cabanel, 75015 Paris
>                     Tel : +33 (0)1-44-49-30-57
>         <tel:%2B33%20%280%291-44-49-30-57>
>         <tel:%2B33%20%280%291-44-49-__30-57> - Fax : +33
>                     (0)1-43-06-50-19 <tel:%2B33%20%280%291-43-06-__50-19>
>                     E-mail address:
>         patrick.fuchs at univ-paris-____diderot.fr
>         <mailto:patrick.fuchs at univ-paris-__diderot.fr>
>         <mailto:patrick.fuchs at univ-__paris-diderot.fr
>         <mailto:patrick.fuchs at univ-paris-diderot.fr>>
>                     Web Site: http://www.dsimb.inserm.fr/~____fuchs
>         <http://www.dsimb.inserm.fr/%7E__fuchs>
>         <http://www.dsimb.inserm.fr/%__7Efuchs
>         <http://www.dsimb.inserm.fr/%7Efuchs>>
>
>                     --
>                     gmx-users mailing list gmx-users at gromacs.org
>         <mailto:gmx-users at gromacs.org>
>         <mailto:gmx-users at gromacs.org <mailto:gmx-users at gromacs.org>>
>         http://lists.gromacs.org/____mailman/listinfo/gmx-users
>         <http://lists.gromacs.org/__mailman/listinfo/gmx-users>
>
>         <http://lists.gromacs.org/__mailman/listinfo/gmx-users
>         <http://lists.gromacs.org/mailman/listinfo/gmx-users>>
>                     Please search the archive at
>         http://www.gromacs.org/____Support/Mailing_Lists/Search
>         <http://www.gromacs.org/__Support/Mailing_Lists/Search>
>
>         <http://www.gromacs.org/__Support/Mailing_Lists/Search
>         <http://www.gromacs.org/Support/Mailing_Lists/Search>> before
>                     posting!
>                     Please don't post (un)subscribe requests to the
>         list. Use
>                     the www interface or send it to
>         gmx-users-request at gromacs.org <mailto:gmx-users-request at gromacs.org>
>         <mailto:gmx-users-request at __gromacs.org
>         <mailto:gmx-users-request at gromacs.org>>.
>                     Can't post? Read
>         http://www.gromacs.org/____Support/Mailing_Lists
>         <http://www.gromacs.org/__Support/Mailing_Lists>
>         <http://www.gromacs.org/__Support/Mailing_Lists
>         <http://www.gromacs.org/Support/Mailing_Lists>>
>
>
>             --
>
>           _______________________________________________________________________________
>
>             Patrick FUCHS
>             Dynamique des Structures et Interactions des Macromolécules
>         Biologiques
>             INTS, INSERM UMR-S665, Université Paris Diderot,
>             6 rue Alexandre Cabanel, 75015 Paris
>             Tel : +33 (0)1-44-49-30-57
>         <tel:%2B33%20%280%291-44-49-30-57>
>         <tel:%2B33%20%280%291-44-49-__30-57> - Fax
>             : +33 (0)1-43-06-50-19 <tel:%2B33%20%280%291-43-06-50-19>
>         <tel:%2B33%20%280%291-43-06-__50-19>
>             E-mail address: patrick.fuchs at univ-paris-____diderot.fr
>         <mailto:patrick.fuchs at univ-paris-__diderot.fr>
>         <mailto:patrick.fuchs at univ-__paris-diderot.fr
>         <mailto:patrick.fuchs at univ-paris-diderot.fr>>
>             Web Site: http://www.dsimb.inserm.fr/~____fuchs
>         <http://www.dsimb.inserm.fr/%7E__fuchs>
>         <http://www.dsimb.inserm.fr/%__7Efuchs
>         <http://www.dsimb.inserm.fr/%7Efuchs>>
>
>             --
>             gmx-users mailing list gmx-users at gromacs.org
>         <mailto:gmx-users at gromacs.org>
>         <mailto:gmx-users at gromacs.org <mailto:gmx-users at gromacs.org>>
>         http://lists.gromacs.org/____mailman/listinfo/gmx-users
>         <http://lists.gromacs.org/__mailman/listinfo/gmx-users>
>
>         <http://lists.gromacs.org/__mailman/listinfo/gmx-users
>         <http://lists.gromacs.org/mailman/listinfo/gmx-users>>
>             Please search the archive at
>         http://www.gromacs.org/____Support/Mailing_Lists/Search
>         <http://www.gromacs.org/__Support/Mailing_Lists/Search>
>
>         <http://www.gromacs.org/__Support/Mailing_Lists/Search
>         <http://www.gromacs.org/Support/Mailing_Lists/Search>> before
>         posting!
>             Please don't post (un)subscribe requests to the list. Use
>         the www
>             interface or send it to gmx-users-request at gromacs.org
>         <mailto:gmx-users-request at gromacs.org>
>         <mailto:gmx-users-request at __gromacs.org
>         <mailto:gmx-users-request at gromacs.org>>.
>             Can't post? Read
>         http://www.gromacs.org/____Support/Mailing_Lists
>         <http://www.gromacs.org/__Support/Mailing_Lists>
>         <http://www.gromacs.org/__Support/Mailing_Lists
>         <http://www.gromacs.org/Support/Mailing_Lists>>
>
>
>
>
>
>     --
>     ___________________________________________________________________________
>     Patrick FUCHS
>     Dynamique des Structures et Interactions des Macromolécules Biologiques
>     INTS, INSERM UMR-S665, Université Paris Diderot,
>     6 rue Alexandre Cabanel, 75015 Paris
>     Tel : +33 (0)1-44-49-30-57 <tel:%2B33%20%280%291-44-49-30-57> - Fax
>     : +33 (0)1-43-06-50-19 <tel:%2B33%20%280%291-43-06-50-19>
>     E-mail address: patrick.fuchs at univ-paris-__diderot.fr
>     <mailto:patrick.fuchs at univ-paris-diderot.fr>
>     Web Site: http://www.dsimb.inserm.fr/~__fuchs
>     <http://www.dsimb.inserm.fr/%7Efuchs>
>     --
>     gmx-users mailing list gmx-users at gromacs.org
>     <mailto:gmx-users at gromacs.org>
>     http://lists.gromacs.org/__mailman/listinfo/gmx-users
>     <http://lists.gromacs.org/mailman/listinfo/gmx-users>
>     Please search the archive at
>     http://www.gromacs.org/__Support/Mailing_Lists/Search
>     <http://www.gromacs.org/Support/Mailing_Lists/Search> before posting!
>     Please don't post (un)subscribe requests to the list. Use the www
>     interface or send it to gmx-users-request at gromacs.org
>     <mailto:gmx-users-request at gromacs.org>.
>     Can't post? Read http://www.gromacs.org/__Support/Mailing_Lists
>     <http://www.gromacs.org/Support/Mailing_Lists>
>
>
>
>

-- 
_______________________________________________________________________
Patrick FUCHS
Dynamique des Structures et Interactions des Macromolécules Biologiques
INTS, INSERM UMR-S665, Université Paris Diderot,
6 rue Alexandre Cabanel, 75015 Paris
Tel : +33 (0)1-44-49-30-57 - Fax : +33 (0)1-43-06-50-19
E-mail address: patrick.fuchs at univ-paris-diderot.fr
Web Site: http://www.dsimb.inserm.fr/~fuchs



More information about the gromacs.org_gmx-users mailing list