The bioinformatics pipelines and tools used in our experiments involved sophisticated computational steps and diverse programming languages. While some of these pipelines were presented as scripts of original command lines, others were workflows packaged in the workflow description language (WDL). For pipelines that can be executed from the command line, the user can add information for each stage of the pipeline along with the command to be executed to the configuration JSON file. Hummingbird will then parse the configuration file, extract each stage of the pipeline and accordingly profile the individual stages. However, if multiple commands (possibly utilizing different software tools) are provided in a chained or composite manner, Hummingbird will execute it as a single unit or step. In case of pipelines in the form of WDL files, Hummingbird uses Cromwell for parsing the workflow and subsequent execution of the stages. Currently, Hummingbird does not have the capability to parse other workflow file formats available and extract individual stages of the pipeline. This is out of the scope for Hummingbird as it would involve development of multiple workflow file format parsers or workflow execution engines.
Here is a template JSON file which the users can follow: https://github.com/StanfordBioinformatics/Hummingbird/blob/master/Hummingbird/conf/examples/sample.conf.json