Fig 1.
A screenshot of a Porcupine workflow.
The editor is divided into four panels, each of them targeted at facilitating a more understandable and reproducible analysis. The workflow editor (1) provides a visual overview of one’s analysis. The functions are all listed in the node editor (2), where the parameters for all functions can be orderly stored. This may include links to important parameters that are listed in the parameter editor (3), such that an overview of the main analysis settings can be easily viewed and modified. Readily executable analysis code is generated in the code window (4).
Fig 2.
An example of simple workflow.
In three steps, this pipeline loads data, processes it, and writes it to disk. This is achieved by connecting the input and output fields from subsequent nodes in the pipeline. The constructed workflow is then transformed in readily executable (Nipype) analysis code.
Fig 3.
An example of a more complicated and realistic fMRI preprocessing pipeline.
Once the code is generated, this can in turn be transformed into a Nipype graph visualisation. Whereas this is usually the end point for a pipeline in Nipype, we here propose to use a visualisation as a starting point of one’s analysis.