Running a pipeline
JIPipe stores its results in a user-defined output folder. After clicking Run (located at the top right corner of a project window) JIPipe will query the output folder from you. You can also generate a random folder that will be located on your operating system’s temporary directory by clicking the button.
After confirming the settings by clicking Run now, JIPipe will start processing the pipeline.
You can have as many runs as you want. Runs will not execute in parallel, but be scheduled to wait until the current run finishes.
The JIPipe output folder is automatically organized in a standardized way, so JIPipe can open a result folder. You find the “Open analysis output …” button in the “Project” menu.
After all results are generated, the result analysis UI will be automatically opened. It allows you to navigate through the results, and import them back into ImageJ (if supported by the generated data type).
The UI has two main components:
- The slot tree shows a hierarchy of all data slots. It is organized by compartment, then algorithm, and finally data slot. By selecting any of the entries, all data associated to the selection or any n-child is displayed.
- The output data table lists the stored data of the selected slot(s) as table (if you are unfamiliar why this is, please take a look at the explanation on how JIPipe processes data)
After selecting a set of slots in the slot tree the output data table will update to only display the data of the selected slots. The table has following columns:
- (Optional) Compartment shows in which graph compartment the generating algorithm is located
- (Optional) Algorithm shows the name of the algorithm that generated the data.
- Location is the folder name where the data is located relative to the slot directory
- Data shows a string representation of the data row. The contents vary depending on the slot data type. For example, file system data display their path at this location.
- Additional columns correspond to the data annotations that were attached by various algorithms
Importing results back into JIPipe/ImageJ/…
On selecting one or multiple rows, additional UI elements are displayed below the table. They contain various actions that can be applied to the output data, such as opening the results directory, importing the data back into ImageJ or JIPipe, or any other operation.
Double-click a row to execute the default action, which is the one most commonly used.
Using third-party software for further analysis
JIPipe stores outputs in a standardized format that can be easily read with third-party software.
The output folder of always contains a file
project.jip, which is the project that generated the output.
This folder contains sub-folders named according to the graph compartment. The ID is the unique ID of the graph compartment
and can be looked up in the
project.jip file or the compartment graph editor.
Each of these compartment folders contains a set of additional sub-folders based on unique ID of the node (also visible in the graph editor).
A node folder contains a set of folders, one for each output slot. This folder may be empty if JIPipe is instructed to not save the output slot.
Each slot folder contains a metadata file
data-table.json that contains information about all stored outputs,
including the data type and annotations. The metadata is also usually available as
data-table.csv file that can be
read more easily. The folder contains sub-folders with numeric names that correspond to the table rows. The contents of
each of these folders is determined by the standardized storage format of the data type (
true-data-type in the data table).
You can open existing project folders in JIPipe. Just go to Project > Open analysis output and select the folder that contains the project.jip file.
We recommend to always copy the whole project folder. This will ensure the highest reproducibility.
Slot output folders are used beyond result analysis for importing/exporting cache states and to communicate with external applications like Python or R. Only the JSON file carries the metadata.
A description of the contents of a table row folder can be viewed in the data type compendium (help menu).