Hi Again Galaxy team. I'm attempting to use the new workflow API, which I understand is still in development. I created a test workflow with a single input, and was able to use 'example_watch_folder.py' to successfully execute it as a history and get the output to a specified location. My question is how I can modify the script to accept multiple inputs (i.e. how do I define which files in the input folder I want to be each input) and if there's a way to specify runtime parameters. For instance, the workflow I want to execute has a filter step on a tabular input item as one of the later steps which needs to be defined at runtime. How would I specify this in the 'watch_folder.py' parameters? or is this not possible yet? FYI, I know perl but not python, so ultimately I want to wrap the python scripts into a larger perl script to execute recursive workflows. Thanks for all your help. Galaxy is an amazing tool and the workflow API is a fantastic improvement. Simon Simon Lank Research Specialist O'Connor Lab, WNPRC 555 Science Dr. Madison WI (608) 265-3389
My question is how I can modify the script to accept multiple inputs (i.e. how do I define which files in the input folder I want to be each input) and if there's a way to specify runtime parameters. For instance, the workflow I want to execute has a filter step on a tabular input item as one of the later steps which needs to be defined at runtime. How would I specify this in the 'watch_folder.py' parameters? or is this not possible yet?
For accepting multiple inputs, I'm assuming you'd differentiate your input files by a particular naming scheme, or location. Using the example_watch_folder.py as an example, you'd need to identify and upload both files (the portion creating the libset variable, you'd want to create multiple here) and finally modify the portion where the 'ds_map' is created. That dictionary is a mapping of step id's (Input Dataset steps) to the uploaded library id's from libset. As far as specifying other runtime parameters, support for this is planned but currently it isn't possible and the workflow needs to be fully predefined outside of the input datasets. Hope this helps. Thanks for using Galaxy, apologies for the delayed response, Dannon
participants (2)
-
Dannon Baker
-
Simon Lank