PROFASI  Version 1.5
Generating trajectory files

What are trajectory files ?

PROFASI multi-segment trajectory files are described in more detail in PROFASI trajectory files . In short, they are files containing information about certain other files. PROFASI stores snapshots of the current state of the protein system and random number generator in a binary format during the run. The binary data is stored in files called "conf.something". When a run is restarted many times, each restart results in a new "conf.specialtag" file. The "traj" file is a file which keeps track of these multiple "conf.specialtag" files so that information can be querried from the entire history of the run together, without actually merging the binary files of the segments of the run.

What programs generate them ?

BasicMCRun, SimTempRun, SimAnnealRun, ParTempRun and WLRun generate them when they are used for a simulation. The trajectories files are called "traj" and are located inside the output directories of the simulation. The program make_traj can parse a series of "conf.specialtag" files and generate a "traj" file describing them, when the binary files come from a single multi-segment run.

Analysis of data generated by older PROFASI versions

If you had a version of PROFASI from 1.1 to 1.4, and have made many simulations, the output of the simulations does not have "traj" files. If there are "" files containing layout information about the binary conf files, you are on safe grounds. You can run "make_traj" as explained below in Generating a trajectory file from binary conf files . The program will detect and use the "" files.

If you have generated data from a PROFASI version so old that it did not even generate "", then things are somewhat more difficult. One way is to generate the missing info files by starting a simulation for the same system on the same computer where the original run was made, and killing it once it has created the traj file. Along with the traj file, the first binary data file will be written with an inline header. It's a binary file, but you can open it with "less" and read the header section which is human readable text. Copy the text between tags "PROFASI_CONF_HEADER" and "END_PROFASI_CONF_HEADER", and paste into a new file called "". That's the file you need to put into the same directory as the binary files which you want to analyze.

Why does one need traj files ?

Trajectory files are necessary to extract snapshots from older runs with extract_snapshot, reconstruct run-time history with additional measurements using extract_props, or even to continue an older run.

Aren't such things handled by the "conf" files ?

It was, in PROFASI versions prior to 1.4.8. The meta-data traj file provides a cleaner way to handle multi-segment runs. It would be possible to write a little code to make sure that extract_snapshot etc continue to work with single binary "" files as before, while handling multi-segment runs with the traj file as an additional feature. But disabling direct handling of the binary files is a more drastic way to draw attention to the preferred new way of doing things.

Generating a trajectory file from binary conf files

Do this only if you don't have the "traj" files, i.e., your runs were made with a version of PROFASI which did not write traj files during the simulations.

make_traj -o n0/traj n0/*

The above takes all files matching n0/*, parses them to find the MC cycle limits, sorts them in the order of the starting MC cycles, adjust cycle ranges to avoid overlaps, and creates a trajectory file called n0/traj. The inputs should be PROFASI binary conf files.

The binary files should preferably each have their own header sections. If they don't, the information of the header section can be provided in a single "" file in the same directory. One "" file must describe the binary files without headers in the same directory.

The names of the input files (conf...) as well as the output trajectory file (n0/traj) should be specified relative to the current working directory. make_traj can not process absolute path names.

Our reasons for introducing traj files

In the oldest versions of PROFASI, there was only one file to store system snapshots along the run, called "conf" inside the output directories. There was no information on the layout of the bytes in the files anywhere. Those files can essentially only be interpreted with the same version of PROFASI, or if they are re-written with header information.

PROFASI 1.1 introduced the additional "" files which contained detailed byte-by-byte description of the contents of the binary files. This information is now written at the start of the binary "conf" files, in a "header" section.

Another important change was motivated by our experience with running long simulations on clusters or supercomputers. When data is transferred from a compute cluster to the user's long term storage, the constantly growing size of the "conf" files lead to larger and larger quantities of data being transferred for later stages of the run. Therefore we decided to start new binary snapshot files for every restart of a run, rather than appending to the pre-existing "conf" files. With the new split files method, the data to be transferred remained roughly constant for all stages of long runs. There were now many files now, containing different parts of the information that was written to the "conf" files of older PROFASI versions. These files were called variously in different development versions of PROFASI, such as "conf.bkp_TIMESTAMP", "conf.stage_num" etc.

The segment-wise storage of configuration snapshots afforded some extra flexibility to the user, such as running different parts of the simulation on different computers without worrying about how they store the binary data. But analysing the data became more cumbersome as scripts now had to work through all the snapshot files. The "traj" file provides a nice way to organize all the information in these segment files.

Another important reason to have the traj file instead of initializing with the segment files every time is that it reduces the number of files which must be opened and closed during initialisation. The traj file contains information on MC cycle limits associated with different segments. So, only those segment files are opened which are required.

See Also
PROFASI trajectory files

PROFASI: Protein Folding and Aggregation Simulator, Version 1.5
© (2005-2016) Anders Irbäck and Sandipan Mohanty
Documentation generated on Mon Jul 18 2016 using Doxygen version 1.8.2