[gmx-users] R: neighborsearching error

[Justin A. Lemkul]

Anna Marabotti wrote:
>  Dear Justin,
> many thanks for your answer. The removal of problematic water was just a "desperate" trial I made to find a
> solution to this problem. During my attempts, I changed:
> - the box size (for this particular system, I use a cubic box of 0.9 nm per side: I enlarged it until 1 nm,
> but nothing changed)

Do 0.9 nm and 1.0 nm correspond to the box "side" or to the distance between the 
solute and box edge (i.e., editconf -d)?

> - the number of steps in minimization (I use generally a maximum of 1000 steps: I increased until 2000, but
> nothing changed)

What is your target for minimization (emtol)?

> - the step size of minimization (I used steps of 0.5, 0.1, 0.05, 0.01 nm: sometimes the minimization arrived
> to convergence, sometimes not, but in any case I had the neighborsearching error on the following PR-MD). In
> any case, with or without final convergence, during the first steps of minimization GROMACS generates some
> stepXXX.pdb files. The .log file reports that some water molecules cannot be settled: I also checked these
> files, and found nothing strange on these water molecules.
> I also inspected the .pdb file containing protein+water molecules before and after minimization: I did not use
> ngmx but I used InsightII (I hope it would give the same result): apparently there are no bumps or contacts

I only suggested ngxm because it is a quick, convenient way to visualize how 
Gromacs "sees" your system, i.e., it will display the periodic box around your 
system to reassure you that solvent has filled the entire box, out to the edges.

> among water molecules, or between water and protein. The "problematic" water are external, at the boundary of
> the box, but apparently with no visible problems. The protein is completely immersed in the center of the box,
> with no contacts with the box side; water molecules appear to be homogeneously distributed, with no voids,
> crevices etc. The protocol I used to add the box and the water molecules is the same that I adopted for
> similar proteins (= of the same family and of the same size) I used previously with no problems: I create the
> topology with pdb2gmx, set the box size and shape with editconf, add water (spc216.gro model) with genbox, add
> counterions (na+) with genion, minimize. No error appear before minimization. I also restarted the entire
> procedure more and more times: the only thing I noted is that the number of water molecules changes even if I
> don't change the box size. By the way, I generally do not specify the number of water molecules to be added,

Now this is weird.  The number of water molecules you see is different each 
time, even for the same box?  Sounds like something with your genbox step is broken.

> nor their position, and I don't know if I can "direct" the solvation using some settings in editconf or genbox
> (but since I don't understand what happens in the system, I don't know how to avoid it!)
> Finally I also tried to minimize (with InsightII) the protein alone, without solvent, before submitting to
> GROMACS procedure, but nothing changed, again.
> I really do not have idea of how to proceed. Any suggestion will be really welcome!

Can you post your .mdp file for your PR MD step?

-Justin

> Anna
> 
> 
> Anna Marabotti wrote:
>> I'm resending the message without attachments, since the previous one was filtered because it was too big.
>> Please let me know how to send infos without being filtered, if more information are needed to help me. I
>> tried again and again, changing settings and options for minimization, but I had no results until now.
>> Best regards
>> Anna
>>
>> Dear all,
>> I'd need some help to understand a problem. I'm preparing my system (a protein of 157 residues in a box of
>> 7507 water molecules and 11 Na+ ions) for a MD. During energy minimization, the program saves some .pdb
> files
>> claiming that it was not able to settle some water molecules, but the minimization finishes regularly
>> producing a .pdb file, and the system converges to the required force in a number of steps lower than the
>> nsteps set in the em.mdp file. When I'm launching the PR-MD, the simulation stops with the
> "neighborsearching
>> error", and in the .log file there are alerts indicating that other water molecules cannot be settled. I
>> checked the files and found no problems on the minimized .pdb file: I cannot see any bump or collision.
>> However, to be sure, I removed the water molecules "not settled" as indicated in the .log file after
>> minimization, and repeated the PR-MD: error again, and a lot of "stepXXX.pdb" molecules. 
>> The potential energy of my system after minimization seems to be quite reasonable and the molecule seems to
> be
>> minimized to convergence. I tried adding other minimization steps, but nothing changes. The same if I reduce
>> the step. I inspected the protein, but I did not find anything anomalous on it; on the other hand, problems
>> seem to arise from water molecules only, not from protein, so I assume that its coordinates are correct.
>> However, when I'm inspecting the system, also water molecules appear to be correct. The only thing I notice
> is
>> that the "anomalous" water molecules are at the boundary of the box.
>>
> 
> Removing the "problematic" water molecules is not a solution; you should have 
> found that in the list archive.  Often these water molecules are the victim of 
> some larger perturbation in your system (i.e., other molecules going shooting 
> off through the box).
> 
> If the problems are always at the boundary, is your box size appropriate?  Do 
> you have voids where solvent should be, but is not?  Check out your system with 
> ngmx to make sure that you have fully solvated your unit cell.
> 
> -Justin
> 
>> Could you please suggest me how to manage this problem? I looked into the gmx-user list archive but found
>> nothing that seems applicable to my case.
>>
>> I'm attaching here the .log files from minimization and PR-MD (after minimization and after minimization and
>> water removal). Please let me know if you need some other infos.
>>
>> Many thanks and best regards.
>> Anna
>> ______________________________________________
>> Anna Marabotti, Ph.D.
>> Laboratory of Bioinformatics and Computational Biology
>> Institute of Food Science, CNR
>> Via Roma 64
>> 83100 Avellino (Italy)
>> Phone: +39 0825 299651
>> Fax: +39 0825 781585
>> Skype: annam1972
>> E-mail: amarabotti at isa.cnr.it
>> Web page: http://bioinformatica.isa.cnr.it/anna/anna.htm
>> ____________________________________________________
>> "If you think you are too small to make a difference, try sleeping with a mosquito"
>>
>> _______________________________________________
>> gmx-users mailing list    gmx-users at gromacs.org
>> http://www.gromacs.org/mailman/listinfo/gmx-users
>> Please search the archive at http://www.gromacs.org/search before posting!
>> Please don't post (un)subscribe requests to the list. Use the 
>> www interface or send it to gmx-users-request at gromacs.org.
>> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>>
> 

-- 
========================================

Justin A. Lemkul
Graduate Research Assistant
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================