Running Your First Job
Assuming you are running a fresh instance of Matlab, first begin by opening EEGLAB.
After being presented with the first GUI, navigate to File > Batch > Run history template batch:
This new GUI combines both context config and batch config into one interface.
For a job we must specify the following five things:
- Context configuration
- Batch configuration
- History and script files
- Data files
Instructions on how to complete each step in order and brief explanations are found below.
1. Context Configuration
As specified in the context tutorial page, this config specifies personal and cluster information.
If the area contains the correct information already, proceed to the following section.
Begin by clicking the Load context configuration button on the GUI. You should be presented with a file chooser. Either navigate to your personally saved configuration or navigate to
derivatives/lossless/code/config and select
contextconfig.cfg. This will load some default options assuming the graham cluster's specifications. From this point you will need to modify the options as they only contain sample paths. For an in depth explanation of what needs to be modified, see the tutorial.
2. Batch Configuration
The next step is to load the configuration files that specify how the scripts will interact with the cluster and its scheduler. First, click Load batch config. A small GUI should be presented. Click Batch configuration file to open a file chooser. Either navigate to your own personalized configuration files, or navigate to
derivatives/lossless/code/config/remote-sbatch/. Select all files beginning with the c prefix and press open. Selection of multiple files can typically be done by holding down the ctrl key. The lower text area of the small GUI should now have entries in it corresponding to the selected filed. Press ok to load these configurations into Matlab.
The batch configuration area should now be filled with text.
NOTE: Users will need to modify the 'submit_options' field. Instead of
--account=def-bucanl, it should be updated to be
--account=groupname for all batch configurations, where
groupname is the name of the desired group. Alternatively, this field may be left empty if the user submitting the job(s) is only a member of one group. In this case, the user's default group will be used.
NOTE: These configurations assume your testing session is of average length. Increasing various parameters can cause a major increase in queue time. See the batch configuration page on how to customize memory and time parameters safely.
3. History and script files
These files specify what the pipeline actually does and how to accomplish it. For an in depth breakdown of what each section of the pipeline does, see here.
To select the pipeline's history files, click History file and navigate to
derivatives/lossless/code/scripts/ and select all files beginning with s and press open (currently, s01 - s05). Selection of multiple files can be done by holding down the ctrl key.
4. Data files
As this is the BIDS lossless pipeline, if your data is stored BIDS compliant, begin importing by clicking the Bids Import button. Navigate to whichever stage of BIDS compliant subject structure you would like the pipeline to run on. Enter the directory so that all subject folders are visible in the file chooser. Click open in the file chooser. All subject files should now be imported into the text field. If not all subject files are currently desired, manually editing the text box and removing the line specifying participant files is the suggested method.
path field after BIDS importing should remain blank. Instead, the
file field should now include the file path along with the filename.
If you wish to import data files not in BIDS format, you may select the Data files import button and navigate to your files this way. Note that the path field may no longer be empty and as such should be validated to be correct.
At the bottom of GUI is a drop down menu allowing for the choice of remote connection type.
If system is chosen, you will need to enter your password into the Matlab command window. Note that the password will be hidden from view and is secure. This option will also take advantage of RSA keys.
If ssh2 is picked, a small GUI allowing password entry will appear. Enter your password and click ok.
Do not select none.
Submission is then finally done by pressing Ok.
The Matlab Command Window will detail the process of submitting the job and display any errors that occur.
Below is an example of a completely filled in GUI with three participants files: