Re: simulation of membrane-protein systems

From: Marcos Sotomayor (sotomayo_at_ks.uiuc.edu)
Date: Wed Dec 13 2006 - 11:59:53 CST

No problem, the script should be something like this:

package require solvate
solvate yourprotein.psf yourprotein.pdb -o yourprotein_water_TEMP -b 1.5 -minmax {{-48 -48 -48} {48 48 48}}
set all [atomselect top all]
$all set beta 0
set seltext "segid WT1 to WT99 and same residue as abs(z) <10"
set sel [atomselect top $seltext]
$sel set beta 1
set badwater [atomselect top "name OH2 and beta > 0"]
set seglist [$badwater get segid]
set reslist [$badwater get resid]
mol delete all
package require psfgen
readpsf kcsa_dppc_water_TEMP.psf
coordpdb kcsa_dppc_water_TEMP.pdb
foreach segid $seglist resid $reslist {
delatom $segid $resid
}
writepdb yourproteinw.pdb
writepsf yourproteinw.psf
file delete yourprotein_water_TEMP.psf
file delete yourprotein_water_TEMP.pdb
exit

- Note that solvate loads the resulting system so the command in the third
line is working on yourportein_water_TEMP.

- The key selection is in the fifth line, where we used segid's to avoid
deleting other water molecules (crystallographic or others) and the abs(z)
command permits you to delete water molecules on the hydrophobic region of
your whole membrane. You will have to modify this selection to avoid
selecting water molecules present in a channel, for off-center membranes,
etc. Take into account that you want to keep water molecules in the
hydrophilic regions of the membrane, don't include them in this selection!

Hope it helps,
Marcos

On Wed, 13 Dec 2006, L. Michel Espinoza-Fonseca wrote:

> This thing about a tcl script to remove such waters sounds good too.
> Do you already have it available? It'll be nice if you could share it
> with some of us :)
>
> Michel
>
> 2006/12/13, Marcos Sotomayor <sotomayo_at_ks.uiuc.edu>:
>>
>>
>>
>> On Wed, 13 Dec 2006, L. Michel Espinoza-Fonseca wrote:
>>
>> > Thank you for your comments.
>> >
>> > I think I wrote that I was measuring the size of the lipid bilayer,
>> > but I was wrong. I'm indeed measuring the distance of the whole box
>> > (water+lipid), but I wrote the opposite :).
>>
>> Michel, if you are measuring the size of the box with the minmax command
>> of VMD, DO NOT include the lipids in your selection, as the outcome will
>> give you a box that's bigger in the xy plan that what it really is and
>> you will have a gap between periodic images that will be filled with water.
>>
>> >
>> > About the answer, yes, I guessed this behavior is not right, specially
>> > because I don't want to simulate "floating discs" (i.e., a
>> > lipid-protein system fully surrounded by water). When I build my
>> > original system it looks good -no waters are found at the edges. Right
>> > now I'm re-equilibrating everything and hope to get the right results
>> > this time.
>>
>> Check your periodic images with VMD first, you may save some time by
>> setting a smaller box in the xy plane.
>>
>> >
>> > Now that this topic was raised, I was wondering how to add more water
>> > to your lipid-protein-water system with "solvate". Usually, when I do
>> > the following:
>> >
>> > solvate mypsf.psf mypdb.pdb -o myoutput -b 3.5 +z 20 -z 20
>> >
>> > But I still get some water molecules in the edges. Maybe you have some
>> > hints about how to avoid that!
>>
>> Peter's hint is quite good. You can also use the regular solvate over the
>> whole system and then delete the water molecules at the edges using a tcl
>> script. A tutorial on how to do this and how to simulate membrane proteins
>> will be released soon.
>>
>> Marcos
>>
>> >
>> > Michel
>> >
>> > 2006/12/13, Marcos Sotomayor <sotomayo_at_ks.uiuc.edu>:
>> >>
>> >> Cesar is likely right, and you can easily check if the size of your
>> >> cell box is OK by showing the periodic images of your system in VMD (if
>> >> the periodic cell info is not included in the dcd, use the vmd command
>> >> "molinfo top set a xxx", where xxx is the value you set for size of the
>> >> periodic cell in the x direction, same thing with b and c for y and z,
>> >> respectively; then use the periodic tab in the graphical representations
>> >> window to visualize periodic images).
>> >>
>> >> Just to answer your original questions, the behavior you are observing
>> is
>> >> not normal (unless you want to simulate a disc), and you should not
>> >> observe water molecules going into the hydrophobic region of the
>> membrane
>> >> (not even at the edges). You should also check that your initial
>> >> condition is right, i.e., you don't have water molecules already at the
>> >> edges of your simulation cell at the level of the hydrophobic region of
>> >> your membrane.
>> >>
>> >> Marcos
>> >>
>> >>
>> >> On Wed, 13 Dec 2006, Cesar Luis Avila wrote:
>> >>
>> >> > Well, indeed taking minmax from lipid selection is a common mistake.
>> You
>> >> > should take minmax from water box. This is because of the periodic
>> nature
>> >> of
>> >> > the box. When writing the coordinates for the system some of the
>> lipids
>> >> that
>> >> > were split in the boundaries get wrapped together resulting in a wider
>> >> box
>> >> > for lipids than it really is. Since water molecules are smaller, they
>> are
>> >> not
>> >> > affected that much.
>> >> > Regards
>> >> > Cesar
>> >> >
>> >> > L. Michel Espinoza-Fonseca escribió:
>> >> >> Dear Peter and Cesar,
>> >> >>
>> >> >> Thank you for your answers.
>> >> >>
>> >> >> Peter: Yes, I'm using pressure controls, but instead of
>> >> >> useConstantRatio I'm using useConstantArea... What do you think?
>> Maybe
>> >> >> I should modify this and see what I get. I don't think I'll be a
>> >> >> problem, since my system is actually in the x-y plane.
>> >> >>
>> >> >> Cesar: I'm using the x-y dimensions of the lipid to assign the length
>> >> >> of my periodic box, so I think the problem is not actually being
>> >> >> caused by this. Thank you anyway for the reminder!
>> >> >>
>> >> >> Cheers,
>> >> >> Michel
>> >> >>
>> >> >> 2006/12/13, Peter Freddolino <petefred_at_ks.uiuc.edu>:
>> >> >>> Hi Michel,
>> >> >>> are you using pressure controls? If so, you may want to try adding
>> >> >>> useConstantRatio to keep your x and y cell dimensions identical to
>> each
>> >> >>> other (this assumes that your membrane is in the x-y plane, so you
>> may
>> >> >>> need to rotate your system).
>> >> >>> Peter
>> >> >>>
>> >> >>> L. Michel Espinoza-Fonseca wrote:
>> >> >>> > Hi people,
>> >> >>> >
>> >> >>> > I have been performing a few simulations of protein-membrane
>> systems
>> >> >>> > using a flexible cell and PBC. I used the "membrane" plugin to
>> build
>> >> >>> > the membranes. I subjected such membranes to minimization and
>> >> >>> > equilibration for a period of 0.5 ns. I get a pretty good
>> >> equilibrated
>> >> >>> > slab, so no problem there. The "problem" (I really don't know if
>> >> >>> > that's a problem) is that after continuing my simulation for about
>> 10
>> >> >>> > ns, the shape of the lipid bilayer looks more like a "disc" than a
>> >> >>> > "box". Moreover, water molecules start to surround the Z-axis
>> edges
>> >> of
>> >> >>> > the membrane. Now my question is, is that normal? According to
>> what I
>> >> >>> > believe, it is not. how can I avoid this?
>> >> >>> > All comments are very appreciated.
>> >> >>> >
>> >> >>> > Thanks a lot!
>> >> >>> > Michel
>> >> >>>
>> >> >>
>> >> >
>> >>
>> >
>>
>

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:42:56 CST