[gmx-users] Weird structure after minimization (membrane protein simulation)

[chris.neale at utoronto.ca]

>> Dear Mark and Chris,
>>
>> Thanks a lot for your suggestions which gave me idea to solve the problem.
>>
>> I have run the minimization again, without the FLEXIBLE water molecules.
>> After the minimization, structure looks perfectly ok to me. please have a
>> look at the new snapshoot.
>> http://i269.photobucket.com/albums/jj58/gromacs/after_em.jpg
>
> Looks fine.
>
>> It is just my speculation that because I am using the TIP4P water model,
>> which has dummy atom (-1.04 charge), it might be possible that due to this
>> water hydrogen's (+0.52 charge) are attracted to the dummy atom of nearby
>> water molecules. That's why if I closely observed the structures (after
>> minimization using flexible water model) then, some water molecules
>> aggregates, and leave some vacuum at other places, that's why structure
>> was
>> looking weird. It is just my speculation please correct me if I am missing
>> something.
>
> It's a reasonable theory, but it seems unlikely that you're the first
> person to have encountered this problem, if indeed it is the only problem.
> Still, there's not much need for flexible waters in your EM, so you can
> keep them rigid and go on with life. I'd do that with a clean install, in
> case something weird you've forgotten about is causing the problem.
>

I use tip4p nearly exclusively and always use flexible waters for my  
minimization. I do the minimization with gromacs3.3 or 3.3.1 (even if  
I will use 3.1.2 for the make_hole routine). I admit that I would  
probably just move on at this point, but I would still be nervous  
about it and the fresh install sounds like a good idea.

>> I have also checked the bond and angle energy values and when I was using
>> the flexible water model it deviate a lot. For bond energy it increases
>> 8.88556e+04 to 4.10925e+07 and for angle energy 3.60214e+04 to 1.09253e+06
>> these energy compensated by short range Coulomb which changes -1.65898e+07
>> to -1.71941e+08 .
>
> Well that lends credence to your theory, so perhaps you should look at the
> paper that describes the development of TIP4P and see if they recommend
> anything. I've not used TIP4P, so know nothing much about it.
>
>> Without flexible water molecules there is no such huge deviation from the
>> original value.Also during the equilibration run I have not observed any
>> LINCS warnings. I am hoping my problem is solved now :-)
>
> Good :-)