This is not much help for the current issue (large files have proven
unwieldy in the past -- see all the bug fixes addressed to cpptraj/ptraj to
fix reading of such large files).
My suggestion is to use NetCDF files exclusively for trajectory files. The
advantages:
binary file format (so it is much, much smaller than ASCII mdcrd files),
files are difficult to completely corrupt (even if one or two frames are),
file reading is probably ~1 order of magnitude faster, and the coordinates
are stored with higher precision.
The disadvantage: data can only be accessed via an API, so you *need* to
use cpptraj/ptraj/VMD (or write your own program using the NetCDF API).
HTH,
Jason
On Tue, Jul 10, 2012 at 6:15 PM, Gordon S Freeman <gfreeman.wisc.edu> wrote:
> Thanks Dan,
>
> By exploded I mean the coordinates of the atoms in the simulation are
> randomly placed what looks like a cubic box with bonds criss-crossing the
> box. So "exploded" in the sense that the configurations that come out are
> reminiscent of a simulation in which the energy has exploded and the
> molecule has gone all to heck.
>
> The script you sent me produces the same result. Does the type of periodic
> box come into play for ptraj? i.e. do I need to tell ptraj that the box is
> octahedral rather than cubic? If the trajectory was smaller (it is 17GB) I
> would attach it for someone else to take a crack at...
>
> Any other thoughts on how to deal with this issue or strategies to try? In
> case its helpful in explaining what I mean by "the frames look exploded," I
> have attached a few vmd snapshots from early in the simulation (looks fine)
> and later in the simulation (looks not so fine)... the raw .mdcrd output
> looks good (like image_1.png) throughout the simulation while the ptraj
> output looks distorted and "explodey."
>
> Thanks a million,
>
> Gordon
>
>
>
>
>
> On 07/10/2012 09:24 AM, Daniel Roe wrote:
>
>> Hi,
>>
>> Sorry for the delay in replying.
>>
>> So just to be clear, the script:
>>
>> trajin md2.mdcrd 1000 19000 5
>>> strip :WAT
>>> strip :Na+
>>> trajout output.pdb pdb append
>>>
>> Still gives you something that looks like it exploded? By this I
>> assumes you mean the two strands of DNA have separated? I think this
>> is just an imaging issue. Try this:
>>
>> trajin md2.mdcrd 1000 19000 5
>> strip :WAT,Na+
>> center :1-20 mass origin
>> image origin center familiar
>> trajout output.pdb pdb append
>>
>> If you use cpptraj you can omit the 'append' keyword from the
>> 'trajout' command since MODEL/ENDMDL keywords are added automatically
>> for multiple frame PDB output files.
>>
>> -Dan
>>
>> On Thu, Jul 5, 2012 at 1:23 PM, Gordon S Freeman<gfreeman.wisc.edu>
>> wrote:
>>
>>> Dear Amber Users,
>>>
>>> I have .mdcrd and .prmtop files that I am trying to use to generate a
>>> .pdb output file that x3DNA plays nicely with. The trajectory is a 20 bp
>>> DNA solvated in tleap by the solvateoct command. If I visualize the
>>> .mdcrd file, everything looks just fine. I use the following ptraj
>>> script to try to generate the .pdb file I need for subsequent analysis:
>>>
>>>
>>> The resulting .pdb file looks great for the first few thousand
>>> simulation frames (i.e. frames 1000 to almost 5000 from the simulation
>>> convert just fine) while the latter 75% of the frames produce .pdb
>>> structures that look like a cubic periodic box that has exploded. The
>>> energies and .mdcrd file look fine so an explosion is clearly not the
>>> explanation. Am I missing something in my ptraj input file above to
>>> properly deal with the non-cubic periodic box the simulation was run
>>> with? Does anyone have a script they have successfully used to do this
>>> conversion? I could easily generate the pdb using VMD and the
>>> correct-looking mdcrd file and then write a script to add the MODEL and
>>> ENDMDL flags 3DNA looks for but I'd rather get my ptraj script working
>>> properly.
>>>
>>> FYI, using the same script above without the lines "unwrap :1-40\ center
>>> :1-40 mass origin\ image origin center familiar\" also produces the same
>>> strange output.
>>>
>>> Thanks in advance for this and all the other help this listserve and the
>>> developers have provided over the years.
>>>
>>> -Gordon
>>>
>>> ______________________________**_________________
>>> AMBER mailing list
>>> AMBER.ambermd.org
>>> http://lists.ambermd.org/**mailman/listinfo/amber<http://lists.ambermd.org/mailman/listinfo/amber>
>>>
>>
>>
>>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
>
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Jul 12 2012 - 13:00:02 PDT
