These are my protocols n they are mine but they can be yours too
I organise files by task, in a whole bunch of directories, grouped by task. As I work I document exactly what I did in a file named PROTOCOL01.txt.
As mistakes are made or I have to redo things for some reason, I go back to that directory and create a new protocol where I document what I did differently. That way I have knowledge of what happened at each revision, and all output files are prefixed with the number of the protocol that created them.
Use new_protocol.sh to create a protocol of the next available number. The script will prompt you for information needed to print a pretty header. I've included an example output here too.
My main script, submit_new, will set up and run a new simulation.
I make one folder for each task, and subfolders for subtasks. For example, if I am equilibrating a bunch of mutants, I have a folder called [protein name] with subfolders minimization, equilibration, etc, and in the equilibration folder there will be folders with the name of each mutant.
This way I can use common scripts and name everything the same as long as it stays in its folder.
In the deepest folder there will be a bunch of files and one or more scripts named PROTOCOL##.sh. This script is run to create all of the output files.
If there are multiple protocol scripts, the one with the largest number describes the current revision of that protocol, and the previous ones are kept for historical reasons. All output files will be prefixed by the number of the protocol that generated them.
The protocol script documents everything that was done in a simulation, step by step, through a combination of comments and commands. Executing it should produce exactly the same output each time (w floating point differences).
Files that are taken from one step to another are represented by symlinks in the directory where they are an input file, pointing to the file that was created as output by a different task. That way if the dependent task is redone, the input file is always up to date.
Please talk to me if any of this is unclear.
Lab members, check out my directory in $PI_HOME on Sherlock to see how this
is all laid out in practice.
Protocol files take a long time to write, but generally they are very reuseable once they are created, especially the simulation scripts. You can find the ones I use in the simulation/input_files directory, grouped by task.
There is also a sample protocol file for creating a homology model.