[gmx-users] Error with equilibration of DPPC membrane with protein

[chris.neale at utoronto.ca]

Try gromacs 3.1.4 make_hole version, downloadable from the user  
contributions page. This works well for me. Alternatively, just select  
a bunch of lipids to remove based on g_mindist and equilibrate -- 50ns  
is cheap nowadays. The only reason that inflategro might be a huge  
advantage is if you plan to simulate a large number of different  
systems and you want to automate the procedure.

Chris.

-- original message --

Edvin Erdtman wrote:
> Hi
>
> We have tried with no cutoff, as I have written in former emails,  
> but that was when we got trouble with LINCS warnings. We then  
> thought that we could try to continue remove lipids in the  
> compression-steps to get rid of that LINCS warnings, and to have a  
> stable system!
> Is it maybe the protein that is the problem - need to be more minimized?

Maybe yes, maybe no.  If I remember your original post, you are doing the
annealing with Nose-Hoover as the thermostat.  IIRC, N-H does not respond well
to changes in temperature (it fluctuates a lot).  Maybe try your  
protocol using
a weak coupling scheme, Berendsen or V-rescale?

-Justin

> /Edvin
>
>
> Justin A. Lemkul wrote:
>>
>> Justin A. Lemkul wrote:
>>>
>>>
>>> Edvin Erdtman wrote:
>>>> Hi again
>>>> I don't know if you were aware of it, but I have commented some  
>>>> Justin's questions further down in the e-mail (my last e-mail  
>>>> wasn't only a thank-email). Since it took so long, and other  
>>>> similar discussions are still running I thought you have missed  
>>>> my comments (see below).
>>>>
>>>> Now we have tried with a Calpha-P cutoff of 5 Å (i.e.  perl  
>>>> inflategro.pl em1/confout.gro 0.97 DPPC 5 em2/input.gro 5  
>>>> em2/area.dat), and position restraints on the protein, I have  
>>>> also merged Cl and SOL in the same temp group, but it does not  
>>>> seem to work anyway. We still get the LINCS warnings.
>>>>
>>>
>>> Why are you using a cutoff during the compression phase?  You will  
>>> continue to delete lipids!  I have never had a problem if I scale  
>>> up by a factor of 4, with a 1.4-nm cutoff, then compress by a  
>>> factor of 0.95 (with no cutoff).
>>>
>>> Maybe that will make a difference?
>>>
>>> -Justin
>>>