Re: [AMBER] GIST: nan (not a number) in unreferenced water-water energy of the grid

From: Daniel Roe <daniel.r.roe.gmail.com>
Date: Wed, 7 Dec 2016 16:07:50 -0500

On Wed, Dec 7, 2016 at 2:41 PM, Bruno Falcone
<brunofalcone.qo.fcen.uba.ar> wrote:
> It seems that the water coordinates are getting very large so they don't
> fit into the rst7 or other ascii based formats, but they do in the
> netcdf format. I guess GIST internally uses the ascii trajectory format
> coordinate size limit?

No, all coordinate calculations in GIST (and in cpptraj as a whole)
are done with full double precision. As long as there are no issues
with your input coordinates there should be no coordinate-based issues
with the GIST calculation.

> I tried centering the netcdf file with these cpptraj commands but the
> 'nan' remain in the coordinates (when I convert them to rst7 to test that).

Did you try testing the frames converted thusly with the 'check'
command? This can tell you if the problem is with the trajectory
itself.

> I also tried to align the centered trajectory with rms but the problem
> persists.

RMS-fitting is not recommended prior to GIST (or with anything that
may use imaged distances). Rotating coordinates will mess up image
calculations.

> Any idea how I can solve this issue and obtain a trajectory that can be
> converted to ascii based so that I can re-run gist?

I recommend checking some of your problem frames with the cpptraj
'check' command to make sure they are not the issue.

-Dan

>
> Thanks,
> Bruno
>
> On 06/12/16 15:44, Bruno Falcone wrote:
>> Hi Steve and Dan,
>>
>> Thank you very much for your suggestions. The problem arose indeed due
>> to problems in the trajectories, which originated from establishing the
>> restraintmask as !:WAT instead of !(:WAT). With the former the
>> restraints were not working.
>>
>> I'm re-running the calculations now, will let you know how it goes.
>>
>> Bruno
>>
>> On 06/12/16 15:38, Daniel Roe wrote:
>>> On Tue, Dec 6, 2016 at 12:27 PM, Bruno Falcone
>>> <brunofalcone.qo.fcen.uba.ar> wrote:
>>>> I checked the trajectories and I think I'm getting closer to finding out
>>>> what the problem may be. I used the NetCDF format for the trajectories
>>>> but I cannot visualize them with VMD. I converted them to pdb, and
>>>> cannot open them either. Inspecting the pdb file I can see that many of
>>>> the water coordinates have nan values in them. Perhaps the coordinate
>>> Having 'nan' in your coordinates is never a good thing. Do you get any
>>> warnings during trajectory process of the format "Warning: Set X
>>> coords 1 & 2 overlap at origin; may be corrupt."? You may want to run
>>> the 'check' command to look for bad coordinates throughout your system
>>> (although be aware this can be an expensive and slow calculation, even
>>> when using cpptraj.OMP). If you do have bad frames you can use 'check'
>>> in conjunction with 'skipbadframes' to skip these problem frames.
>>>
>>> Hope this helps,
>>>
>>> -Dan
>>>
>>>> numbers don't fit in the pdb columns so they appear as pdb? I converted
>>>> the frames to rst7 format and I also get nan in the coordinates.
>>>>
>>>>
>>>> Hope this helps,
>>>>
>>>> Thanks,
>>>> Bruno
>>>>
>>>> On 06/12/16 13:59, Steven Ramsey wrote:
>>>>> Hi Bruno,
>>>>>
>>>>> That's a strange output for GIST. Some information would help me get to the
>>>>> bottom of this issue and therefore make suggestions on how to resolve it
>>>>> (the first question will likely clarify the remaining questions if the
>>>>> answer is yes):
>>>>>
>>>>> Are these GIST calculations performed within the tutorial itself? (meaning
>>>>> the calculations are run on the given files/trajectories and are exactly
>>>>> following the instructions)
>>>>> What solvent model is in use?
>>>>> Which version of GIST is in use?
>>>>> How many frames are being analyzed by GIST?
>>>>> Was there any alignments being used before using GIST?
>>>>> Are we seeing nan's in every voxel or a few here and there? (both are bad,
>>>>> but it will help me figure out what went wrong)
>>>>> Are there any other files with nan values (particularly Esw)?
>>>>>
>>>>> Thanks,
>>>>>
>>>>> --Steve
>>>>>
>>>>> On Tue, Dec 6, 2016 at 9:40 AM, Bruno Falcone <brunofalcone.qo.fcen.uba.ar>
>>>>> wrote:
>>>>>
>>>>>> Hi, I'm performing Grid Inhomogeneous Solvation Theory (GIST)
>>>>>> calculations with cpptraj, of small organic compounds in water.
>>>>>>
>>>>>> I followed the instructions in the GIST tutorial and performed a
>>>>>> restrained MD to then analyse with GIST.
>>>>>> http://ambermd.org/tutorials/advanced/tutorial25/
>>>>>>
>>>>>> Although most of the calculations succeed, I'm getting "nan" (not a
>>>>>> number) in the "Total unreferenced water-water energy of the grid: Eww"
>>>>>> term. Checking the .dx files, this arises from the gist-Eww-dens.dx file
>>>>>> containing all nan in the output.
>>>>>>
>>>>>> Was wondering why this might be the case, and if you could point out to
>>>>>> any suggestions to fix this problem.
>>>>>>
>>>>>> Thanks for your help,
>>>>>>
>>>>>> Bruno
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> AMBER mailing list
>>>>>> AMBER.ambermd.org
>>>>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>>>>
>>>>> _______________________________________________
>>>>> AMBER mailing list
>>>>> AMBER.ambermd.org
>>>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>> _______________________________________________
>>>> AMBER mailing list
>>>> AMBER.ambermd.org
>>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>
>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber



-- 
-------------------------
Daniel R. Roe
Laboratory of Computational Biology
National Institutes of Health, NHLBI
5635 Fishers Ln, Rm T900
Rockville MD, 20852
https://www.lobos.nih.gov/lcb
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Dec 07 2016 - 13:30:02 PST
Custom Search