Re: [gmx-users] OPLS + ffgmx

[Alok]

Hi Mark and Chris,
Thanks a lot to both of you for going through my query, for suggesting the 
positive and negative points of this procedure.

I will think and then decide how to go about it.

I am really thankful to Chris for the detailed suggestion regarding the 
procedure. I will come back to you later as I go about it.


Thanking you all again,
Regards,
Alok.

- Original Message - 
From: Mark Abraham [EMAIL PROTECTED]

To: Discussion list for GROMACS users gmx-users@gromacs.org
Sent: Wednesday, September 06, 2006 7:47 AM
Subject: Re: [gmx-users] OPLS + ffgmx



Hi all,

I am trying to do a membrane protein simulation. I want to use OPLS - AA
force field for protein and ffgmx (modified ffgmx force filed with lipid
parameters from user contribution section ) force field for POPE lipids.

Is it is possible and advisable to use two different force fields for
protein and lipids?


In general, this way madness lies. Force fields are mathematical
constructs that are optimized to approximately reproduce some experimental
properties *in cooperation with itself*. There is no reason to expect that
force field parameters for bond strength have any great correlation with
an experimentally determined bond strength, and thus to have no
correlation with a bond strength from another force field. Accordingly
there's no reason to expect that a mixture of parts of different force
fields will work well together. It's somewhat like taking half of a soccer
team and half of a gridiron team and expecting them to be able to play
rugby together.


Or else, Is there any all atom force field availble for lipid molecules
which I can use?


CHARMM parameter sets optimized for combined protein  lipid calculations
exist. Simplest would then be to use CHARMM or NAMD. If you want to use
GROMACS, you can get such a force field, use my scripts available here
http://www.gromacs.org/contributed_by_users/task,doc_details/gid,59/ to
convert them to gromacs format, obtain some .rtp files elsewhere, test
carefully, and simulate to your heart's content.

Mark

___
gmx-users mailing listgmx-users@gromacs.org
http://www.gromacs.org/mailman/listinfo/gmx-users
Please don't post (un)subscribe requests to the list. Use the
www interface or send it to [EMAIL PROTECTED]
Can't post? Read http://www.gromacs.org/mailing_lists/users.php

___
gmx-users mailing listgmx-users@gromacs.org
http://www.gromacs.org/mailman/listinfo/gmx-users
Please don't post (un)subscribe requests to the list. Use the 
www interface or send it to [EMAIL PROTECTED]

Can't post? Read http://www.gromacs.org/mailing_lists/users.php

Re: [gmx-users] OPLS + ffgmx

[chris . neale]

I posted the procedure and not the modified files mostly because this is my
modification of somebody else's files. Without permission, it is not reasonable
for me to redistribute them. In order to both help you and still treat the
originals with due respect, I have attached snippets that will help you to make
your own modifications. It should not take very long for you to copy the 
procedure.

Regarding the advisability of combining the Berger lipids and the OPLS-AA force
field... Read this: Tieleman et. al. J. Phys.:Condens. Matter 18 (2006) S1221-34
As for my own interpretation (of course that depends on your system and what
question you are trying to answer): If you are focusing mostly on the protein
and just want a pretty good lipid, then yes, I personally think it's a good
combo. If you are focusing on the lipid behaviour around the protein then you
run into more difficulty, but then again there really aren't any good fields for
that as far as I know. I am not sure how good the Berger lipid / gromos field
combination represents that either.

Follow the procedure here:
http://www.gromacs.org/pipermail/gmx-users/2006-May/021416.html

When it says:

1. Added [atomtypes] from lipid.itp to ffoplsaanb.itp -- after changing c6/c12
to sigma/epsilon. Also added atomtype H from olsa_369 to match H expected by
pope.itp
  - sigma   = (c12/c6)^1/6
  - epsilon = c6/(4*sigma^6)

Do this:

a) Using some script of your own or MS Excel convert the c6 and c12 values to
sigma and epsilon values as per the noted equations.

b) Take the [atomtypes] section from lipid.itp to the [atomtypes] section of
ffoplsaanb.itp. Since ffoplsaanb.itp has only one section, and it is
[atomtypes], just paste it to the bottom. 

When it says:

2. Added [pairtypes] from lipid.itp to ffoplsaanb.itp -- after changing c6/c12
to sigma/epsilon. (gives effective fudgeLJ of 0.125). Also changed all reference
to OW to opls_116 (opls spc water oxygen) and simply removed any with reference
to HW as it will be zero regardless.

Do this:

c) Make the same c6/c12 - sigma/epsilon changes and paste the section
([pairtypes] identifier and all) to the bottom of the new ffoplsaanb.itp file.


When it says:

3. Added [dihedraltypes] from lipid.itp to ffoplsaabon.itp. 
  - Prior to running ensure that the non-RB dihedral does not exist for these
groups.

Do this:

d) paste the following segment to the bottom of ffoplsaabon.itp:

; Added by Chris Neale April 16 2006 based on
; lipid.inp from
http://moose.bio.ucalgary.ca/index.php?page=Structures_and_Topologies
; Based on Berger et al, Biophys. J. (1997) 72, pp. 2002-2013.
; Copy and paste from lipid.inp
;[ bondtypes ]
;[ constrainttypes ]
;[ angletypes ]
[ dihedraltypes ]
;  ij func coefficients
  LP2   LP23  9.278912.156-13.120   -3.0597   26.240-31.495
etc...


e) Then make your topology file according to what was already laid out in
section 4 of the previous post.

#

The relevant addition from my ffoplsaanb.itp file

; Added by Chris Neale April 16 2006 based on
; lipid.inp from
http://moose.bio.ucalgary.ca/index.php?page=Structures_and_Topologies
; Based on Berger et al, Biophys. J. (1997) 72, pp. 2002-2013.
; Copy and paste from lipid.inp then duplicate first column and add zero in 3rd
column
; Then comment out initial version (with C6 and C12) and replace with sigma and
epsilon
; NOTE considered the use of epsilon LOS 0.711 (Berger/OPLS?) not 0.879 
(Tieleman)
; HOWEVER, dihedrals etc are set up for 0.879 therefore use that one
; Also added name = H from opls_369 to match H expected by pope.itp
; Also changed 3rd column from all zeros to all ones
   LO   LO 115.9994  0.000   A2.96000e-01  8.87864e-01
;carbonyl O, OPLS
  LOM  LOM 115.9994  0.000   A2.96000e-01  8.87864e-01
;carbonyl O, OPLS
  LNL  LNL 114.0067  0.000   A3.25000e-01  7.11280e-01
;Nitrogen, OPLS
etc...


; Added by Chris Neale May 1 2006 based on
; lipid.inp from
http://moose.bio.ucalgary.ca/index.php?page=Structures_and_Topologies
; Based on Berger et al, Biophys. J. (1997) 72, pp. 2002-2013.
; Copy and paste from lipid.inp into excel then convert c6 and c12 to sigma and
epsilon
; sigma=power(c12/c6,1/6) and epsilon=c6/(4*power(sigma,6))
; NOTE direct conversion therefore using LOS epsilon 0.879 as Tieleman
; Removed all SPC hydrogen as they will be calculated as zero anyway
; This was required so as to give LJ14 values involving lipid-lipid a fudgeLJ
value of 0.125
[ pairtypes ]
;   i   j   funct   sigma   epsilon
LO  LO  1   2.96E-011.10E-01
LO  LOM 1   2.96E-011.10E-01
LO  opls_1161   3.06E-019.47E-02
LO  LNL 1   3.10E-019.88E-02
LO  LC  1   3.33E-017.76E-02
etc...
___
gmx-users mailing list

Re: [gmx-users] OPLS + ffgmx

[Mark Abraham]

 Hi all,

 I am trying to do a membrane protein simulation. I want to use OPLS - AA
 force field for protein and ffgmx (modified ffgmx force filed with lipid
 parameters from user contribution section ) force field for POPE lipids.

 Is it is possible and advisable to use two different force fields for
 protein and lipids?

In general, this way madness lies. Force fields are mathematical
constructs that are optimized to approximately reproduce some experimental
properties *in cooperation with itself*. There is no reason to expect that
force field parameters for bond strength have any great correlation with
an experimentally determined bond strength, and thus to have no
correlation with a bond strength from another force field. Accordingly
there's no reason to expect that a mixture of parts of different force
fields will work well together. It's somewhat like taking half of a soccer
team and half of a gridiron team and expecting them to be able to play
rugby together.

 Or else, Is there any all atom force field availble for lipid molecules
 which I can use?

CHARMM parameter sets optimized for combined protein  lipid calculations
exist. Simplest would then be to use CHARMM or NAMD. If you want to use
GROMACS, you can get such a force field, use my scripts available here
http://www.gromacs.org/contributed_by_users/task,doc_details/gid,59/ to
convert them to gromacs format, obtain some .rtp files elsewhere, test
carefully, and simulate to your heart's content.

Mark

___
gmx-users mailing listgmx-users@gromacs.org
http://www.gromacs.org/mailman/listinfo/gmx-users
Please don't post (un)subscribe requests to the list. Use the
www interface or send it to [EMAIL PROTECTED]
Can't post? Read http://www.gromacs.org/mailing_lists/users.php