[gmx-users] Analyzing a trajectory split over multiple files

[Suman Chakrabarty]

On Wed, Nov 26, 2008 at 2:19 PM, Omer Markovitch <omermar at gmail.com> wrote:
> On Tue, Nov 25, 2008 at 01:47, Suman Chakrabarty <suman at sscu.iisc.ernet.in>
> wrote:
>>
>> Dear all,
>>
>> I have a very long trajectory split over a large number of files. What
>> would be the most efficient way to use the analysis programs over them?
>> Do they support multiple input for trajectory? Or I need to combine all
>> files into a single one anyway?
>
> You should analyze each individual part on its own, and output the raw data
> of the result, then average the parts together.
>
> For example - say you are interested in coordination number of molecule X:
> For each part, output both the total number of X's found, plus the total
> number of X neighbors found. The coordination number is then the sum of
> neig. divided by the sum of X's, from ALL parts.
> It might be an issue if each part has different time length, then averaging
> the coordination number of each part might give more statistical weigh to
> short parts then they should have.
>
> In other words, you should first collect the data of each part together, and
> only then avg..
>
> --Omer.


Thanks for your response Omer. But I am currently more interested in
observing the time evolution of certain properties, e.g. radius of
gyration etc. and their histogram over the full trajectory. So I must
combine either trajectory files or the output analysis data.

It would have been much easier if there was some way to concatenate
the multiple .xvg files easily. Also, the analysis programs should be
able to export the output data in raw 2 (or more) column format
without all the formatting statements as an option.


Regards,
Suman.