[gmx-users] Melting temperature for the lipid bilayer
Piggot T.
T.Piggot at soton.ac.uk
Thu Nov 24 12:19:46 CET 2016
Hi,
I'd suggest it is a combination of 4 and 5.
It is still not completely clear how you are determining if the membrane is classed as liquid disordered or not and also why you think it is too ordered in your simulations at 350K (are they in a gel phase or just in a more ordered state than you'd expect?). I understand lower APL and higher order parameters mean a more ordered membrane, but what are you expecting to see with DSPS above the phase transition temperature and why? I'm not aware of experimental information regarding these values for DSPS (although that doesn't mean the information isn't out there).
That said, it is known that PS doesn't perform particularly well in the CHARMM36 force field, even with the re-parameterisation of the ion interactions (see http://www.sciencedirect.com/science/article/pii/S0005273616303145) and will result in too ordered membranes. Combine this with the fact that you are using it within GROMACS (which results in slightly more ordered membranes with the CHARMM36 lipids compared to other simulation packages, as referred to in the paper you linked to), I'm not surprised that your membranes are more ordered than you think they should be.
As for cut-off's, switching off at 8A will help slightly increase disorder in your membrane. If you have a membrane only system, I'd definitely go for this. I believe the recommendation for using 10A (again in the paper your linked to) was to be consistent with other parts of the CHARMM36 force field (e.g. the protein parameters). The lipids were originally parameterised with the 8A switching.
If you'd like something that is more disordered for DSPS at 350K, I'd choose another force field. Slipids would be an easy choice (you should be able to use the starting structures you already have) and from some tests of POPS and DOPS I've done, it should give you a more disordered membrane. Otherwise, you will need to accept the limitations of the force field you have chosen.
Cheers
Tom
________________________________________
From: gromacs.org_gmx-users-bounces at maillist.sys.kth.se [gromacs.org_gmx-users-bounces at maillist.sys.kth.se] on behalf of Mohsen Ramezanpour [ramezanpour.mohsen at gmail.com]
Sent: 24 November 2016 03:48
To: Discussion list for GROMACS users
Subject: Re: [gmx-users] Melting temperature for the lipid bilayer
Dear Thomas,
Thanks for your comment,
Sorry all as the emails got a bit deviated from the main question.
Problem:
I want to do simulation on a DSPS bilayers with charmm36 ff in
"liquid_disordered" phase.
To do so, I need to find the right temperature that gives me DSPS in
liquid_disordered.
Since the reported experimental T is 341 K, I chose 350 and 355 K but it
did not give a liquid_disordered phase (till 250 ns) in these temperatures.
So, I tried higher T values of 360 and 365.
365 results in the right phase. Since 365 has a gap of 24 with experimental
value I got suspicious about it (probably there is something wrong in my
simulations
but I do not know which part)
So, there are different possibilities:
1) .mdp parameters are not right
Here is a copy of parameters I use for my simulations.
integrator = md
dt = 0.002
nsteps = 150000000
nstlog = 10000
nstxout = 500000
nstvout = 500000
nstfout = 500000
nstcalcenergy = 1000
nstenergy = 10000
;
cutoff-scheme = Verlet
nstlist = 20
rlist = 1.2
coulombtype = pme
rcoulomb = 1.2
vdwtype = Cut-off
vdw-modifier = Force-switch
rvdw_switch = 1.0
rvdw = 1.2
;
tcoupl = Nose-Hoover
tc_grps = lipids WI
tau_t = 1.0 1.0
ref_t = 360 360
;
pcoupl = Parrinello-Rahman
pcoupltype = semiisotropic
tau_p = 5.0
compressibility = 4.5e-5 4.5e-5
ref_p = 1.0 1.0
;
constraints = h-bonds
constraint_algorithm = LINCS
;
nstcomm = 100
comm_mode = linear
comm_grps = lipids WI
;
refcoord_scaling = com
2) force field parameters (.itp) for DSPS provided by Charmm-GUI are not
right (there is no specific study for DSPS. DSPS is made by PS headgroups
and DS tails)
As I checked, the parameters (at least the charges and bonds) for DS tails
and PS groups are consistent with what is in other lipids including either
PS or DS.
Maybe there are some parts hidden to my eyes and I miss them unconsciously.
I checked bonds as I thought there might be some bonds between two tails by
mistake. That was not the case, though.
3) sampling (250 ns) for each T is not enough to get the right phase.
I am testing this by going to 500 ns. 500 should be fairly enough to check
the idea.
4) using charmm36 ff for PS lipids (at least) in Gromacs package might have
problem (i.e. the optimized parameters in following article might not be
optimal for PS lipids)
The result of the following article has been implemented in Charmm-GUI
output .mdp parameters.
http://pubs.acs.org/doi/abs/10.1021/acs.jctc.5b00935
In this article, PS lipids seem to be more tricky than other studied lipids
when one aims to use charmm36 in Gromacs.
For example, simulation of PS lipids in Gromacs result in higher order
parameters compared to other packages (e.g. Amber, OpenMM, NAMD).
I am also testing cutt-offs of 8-12 to see what will happen. This cutt-off
has been suggested to give better results for lipids in this article.
Although, for the sake of consistency with other lipids, 10-12 has been
chosen as the optimal cut-off range.
5) Everything is right and it behaves as it should!
I wished :-) but it is less likely. 24 difference does not seem logical.
Hope the problem is clear now.
Thanks in advance for your suggestions and comments.
Cheers
Mohsen
On Wed, Nov 23, 2016 at 5:16 PM, Thomas Piggot <t.piggot at soton.ac.uk> wrote:
> Hi Moshen,
>
> PS lipids with the standard GROMOS54A7 force fields (which is what I'm
> assuming you mean but 53a7) are likely to not produce the correct behaviour
> of PS membranes. The GROMOS re-parameterisation of lipids for 54A7 was
> primarily based upon the modification of the CH3 choline head group
> parameters for PC lipids to make them behave well (plus the inclusion of
> the commonly used 'Chiu' united-atom lipids charges). Therefore using the
> same parameters for PS is almost certainly not going to work well, i.e.
> very likely to incorrectly produce far too ordered or even gel like
> membranes when they should be in the liquid crystal phase. I can't be sure
> as I haven't explicitly tested DSPS, but I have been testing some different
> PS force fields and can provide some GROMOS based PS parameters (tested for
> POPS and DOPS) should you wish to use a GROMOS based force field which is
> reasonable for such membranes.
>
> However, based upon the information you have provided, I can't be quite be
> sure of exactly what you are trying to do. If you are judging your CHARMM
> simulations based upon the fact that from your GROMOS simulations you can
> quickly make a gel or very ordered phase, I refer you to my above point
> about the accuracy of the GROMOS force field you are using. It doesn't
> surprise me that your GROMOS force field based simulations rapidly form gel
> or very ordered phases, as I think the parameters you may well be using for
> these simulations are likely to induce this due to force field
> deficiencies. In other words, the CHARMM simulations are likely to be
> behaving sensibly but your GROMOS ones not.
>
> I also completely agree with and want to further re-iterate some of the
> points already made by Justin and Chris. Trying to quantify a phase
> transition temperature of a lipid membrane from a simulation is a very
> difficult task. This because both heating or cooling simulations won't
> likely be anywhere near slow enough to measure this accurately and also
> because simulations at different fixed temperatures will also still likely
> suffer from sampling problems.
>
> As an aside, you need to make sure that for your different simulations
> (GROMOS or CHARMM), you are using appropriate simulation parameters (i.e.
> appropriate mdp's). Lipid membrane force fields in particular are very
> sensitive to van der Waals cut-off's, so you need to get this right before
> doing anything else.
>
> Cheers
>
> Tom
>
>
> On 23/11/16 21:08, Mohsen Ramezanpour wrote:
>
>> Lowering the temperature can push the system to gel phase or ordered phase
>> quickly (this is based on my simulations using Gromos53a7 on DSPS).
>>
>> Besides, I used the same system for different temperatures. This initial
>> structure was made by myself using genconf. So, it was not in
>> perfect "liquid state".
>> The approach for making the initial state would not be a problem as the
>> same approach gave liquid_disordered phase using gromos53a7 ff in 353.
>>
>> In these simulations, I used the .mdp files suggested by Charmm-GUI, as
>> well as .itp and .gro files for DSPS were also taken from Charmm-GUI
>> output.
>> I checked .mdp parameters with what suggested in recent articles as
>> follows:
>>
>> http://pubs.acs.org/doi/abs/10.1021/acs.jctc.5b00935
>> Everything seems okay regarding to parameters.
>>
>> Cheers
>> Mohsen
>>
>> On Wed, Nov 23, 2016 at 1:37 PM, Christopher Neale <
>> chris.neale at alum.utoronto.ca> wrote:
>>
>> You started a run with a liquid state conformation and it formed a gel on
>>> its own at 350K ? Fair enough. You're using the proper charmm force
>>> switching and cutoffs?
>>>
>>> ________________________________________
>>> From: gromacs.org_gmx-users-bounces at maillist.sys.kth.se <
>>> gromacs.org_gmx-users-bounces at maillist.sys.kth.se> on behalf of Mohsen
>>> Ramezanpour <ramezanpour.mohsen at gmail.com>
>>> Sent: 23 November 2016 15:31:53
>>> To: Discussion list for GROMACS users
>>> Subject: Re: [gmx-users] Melting temperature for the lipid bilayer
>>>
>>> Hi Christopher,
>>>
>>> I took the second approach by running simulations for 5 different
>>> temperatures of 5 degrees intervals. (350, 355, 360, 365)
>>> I will not pick any specific temperature as melting T. However, I can see
>>> the shift between 355 and 365 K, representative of going to a new phase
>>> (Liquid_disorder) in 365 K.
>>> Visualization is also enough to see that 350 is in gel phase.
>>>
>>> Cheers
>>>
>>> On Wed, Nov 23, 2016 at 12:28 PM, Christopher Neale <
>>> chris.neale at alum.utoronto.ca> wrote:
>>>
>>> I presume you're heating your bilayer up slowly to assess the melting
>>>> temperature? This method will give you an estimated melting temperature
>>>> that is equal to the true melting temperature of that Hamiltonian in
>>>> simulation only if you go infinitely slowly. At finite heating speeds,
>>>>
>>> your
>>>
>>>> observed melting temperature will be greater than or equal to the true
>>>> simulation melting temperature.
>>>>
>>>> You have to go slower than you might expect. For instance, even 0.15
>>>> K/ns
>>>> overestimates the melting temperature (on average, of course).
>>>>
>>>> On the other hand, if you're evaluating the Tm by running simulations at
>>>> various constant temperatures and you call the Tm the temperature at
>>>>
>>> which
>>>
>>>> you see 50% gel and 50% liquid dynamically interconverting many times
>>>>
>>> then
>>>
>>>> your estimate should be correct for the Hamiltonian you selected (though
>>>>
>>> I
>>>
>>>> expect this approach to be way beyond current computational abilities).
>>>>
>>>> ________________________________________
>>>> From: gromacs.org_gmx-users-bounces at maillist.sys.kth.se <
>>>> gromacs.org_gmx-users-bounces at maillist.sys.kth.se> on behalf of Mohsen
>>>> Ramezanpour <ramezanpour.mohsen at gmail.com>
>>>> Sent: 22 November 2016 16:40:08
>>>> To: Discussion list for GROMACS users
>>>> Subject: Re: [gmx-users] Melting temperature for the lipid bilayer
>>>>
>>>> It is worth mentioning that with Gromos53a7, the model was made by
>>>>
>>> merging
>>>
>>>> head group and tails, I got T= 353 K for the right phase.
>>>>
>>>> On Tue, Nov 22, 2016 at 2:37 PM, Mohsen Ramezanpour <
>>>> ramezanpour.mohsen at gmail.com> wrote:
>>>>
>>>> Hi Justin,
>>>>>
>>>>> Comments interspersed.
>>>>>
>>>>> On Tue, Nov 22, 2016 at 1:18 PM, Justin Lemkul <jalemkul at vt.edu>
>>>>>
>>>> wrote:
>>>
>>>>
>>>>>> On 11/21/16 3:28 PM, Mohsen Ramezanpour wrote:
>>>>>>
>>>>>> Dear gromacs users,
>>>>>>>
>>>>>>> Running simulation on a lipid bilayer made by Charmm-GUI, there is
>>>>>>>
>>>>>> about
>>>>
>>>>> 24
>>>>>>> degrees difference between the reported temperature in Avanti and
>>>>>>>
>>>>>> what
>>>
>>>> I
>>>>
>>>>> see in my simulations for getting a liquid disordered (L_disorder)
>>>>>>> bilayer.
>>>>>>>
>>>>>>>
>>>>>>> How are you quantifying the transition in the simulation?
>>>>>>
>>>>> Mainly by diffusion constant and diffusion graph. However, diffusion
>>>>> constant, bilayer thickness and area per lipid all have a shift.
>>>>> Order parameters are also higher for lower temperatures.
>>>>>
>>>>>
>>>>>> I am using all-atom Charmm36 FF for simulations.
>>>>>> What is the lipid?
>>>>>>
>>>>>> DSPS
>>>>>
>>>>> I know that it is not possible to get an exact match between
>>>>>>
>>>>> experimental
>>>>
>>>>> and simulation T values for L_alpha phase of bilayers, but I am not
>>>>>>>
>>>>>> sure
>>>>
>>>>> if
>>>>>>> these gap seems reasonable or there is something wrong in my
>>>>>>> simulations. I
>>>>>>> have checked the bonds in tails and apparently, everything looks fine
>>>>>>> with
>>>>>>> topology.
>>>>>>>
>>>>>>>
>>>>>>> A gap of 24 degrees is quite substantial. I would not consider that
>>>>>> outcome to be sufficiently accurate.
>>>>>>
>>>>>> The melting temperature for DSPS is 68 centigrade (341 K). The
>>>>> liquid_disordered one is about 365 K (at least one with reasonable
>>>>> characteristics of liquid_disorder phase).
>>>>>
>>>>> However, based on this, doing simulation in L_alpha phase requires
>>>>>>
>>>>> hight T
>>>>
>>>>> values (close to 85 or 90 degrees of centigrade) which makes me worry
>>>>>>> about
>>>>>>> the water molecules in the system. It is a high temperature for water
>>>>>>> molecules.
>>>>>>>
>>>>>>>
>>>>>>> Agreed, but TIP3P has a ton of other problems, so this is probably
>>>>>> the
>>>>>> least of our concerns at the moment.
>>>>>>
>>>>>> -Justin
>>>>>>
>>>>>> Its worth mentioning that bilayer structural properties like lipid
>>>>>>
>>>>> order
>>>
>>>> parameters match available experimental data for these high T values.
>>>>>>>
>>>>>>> Please let me know your opinion.
>>>>>>>
>>>>>>> Cheers
>>>>>>> Mohsen
>>>>>>>
>>>>>>>
>>>>>>> --
>>>>>> ==================================================
>>>>>>
>>>>>> Justin A. Lemkul, Ph.D.
>>>>>> Ruth L. Kirschstein NRSA Postdoctoral Fellow
>>>>>>
>>>>>> Department of Pharmaceutical Sciences
>>>>>> School of Pharmacy
>>>>>> Health Sciences Facility II, Room 629
>>>>>> University of Maryland, Baltimore
>>>>>> 20 Penn St.
>>>>>> Baltimore, MD 21201
>>>>>>
>>>>>> jalemkul at outerbanks.umaryland.edu | (410) 706-7441
>>>>>> http://mackerell.umaryland.edu/~jalemkul
>>>>>>
>>>>>> ==================================================
>>>>>> --
>>>>>> Gromacs Users mailing list
>>>>>>
>>>>>> * Please search the archive at http://www.gromacs.org/Support
>>>>>> /Mailing_Lists/GMX-Users_List before posting!
>>>>>>
>>>>>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>>>>>
>>>>>> * For (un)subscribe requests visit
>>>>>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
>>>>>> send a mail to gmx-users-request at gromacs.org.
>>>>>>
>>>>>>
>>>>>
>>>>> --
>>>>> *Rewards work better than punishment ...*
>>>>>
>>>>>
>>>>
>>>> --
>>>> *Rewards work better than punishment ...*
>>>> --
>>>> Gromacs Users mailing list
>>>>
>>>> * Please search the archive at http://www.gromacs.org/
>>>> Support/Mailing_Lists/GMX-Users_List before posting!
>>>>
>>>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>>>
>>>> * For (un)subscribe requests visit
>>>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
>>>> send a mail to gmx-users-request at gromacs.org.
>>>> --
>>>> Gromacs Users mailing list
>>>>
>>>> * Please search the archive at http://www.gromacs.org/
>>>> Support/Mailing_Lists/GMX-Users_List before posting!
>>>>
>>>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>>>
>>>> * For (un)subscribe requests visit
>>>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
>>>> send a mail to gmx-users-request at gromacs.org.
>>>>
>>>>
>>>
>>> --
>>> *Rewards work better than punishment ...*
>>> --
>>> Gromacs Users mailing list
>>>
>>> * Please search the archive at http://www.gromacs.org/
>>> Support/Mailing_Lists/GMX-Users_List before posting!
>>>
>>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>>
>>> * For (un)subscribe requests visit
>>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
>>> send a mail to gmx-users-request at gromacs.org.
>>> --
>>> Gromacs Users mailing list
>>>
>>> * Please search the archive at http://www.gromacs.org/
>>> Support/Mailing_Lists/GMX-Users_List before posting!
>>>
>>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>>
>>> * For (un)subscribe requests visit
>>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
>>> send a mail to gmx-users-request at gromacs.org.
>>>
>>>
>>
>>
> --
> Dr Thomas Piggot
> Visiting Fellow
> University of Southampton, UK.
>
>
> --
> Gromacs Users mailing list
>
> * Please search the archive at http://www.gromacs.org/Support
> /Mailing_Lists/GMX-Users_List before posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.
>
--
*Rewards work better than punishment ...*
--
Gromacs Users mailing list
* Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting!
* Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
* For (un)subscribe requests visit
https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-request at gromacs.org.
More information about the gromacs.org_gmx-users
mailing list