Hi all,

I was recently wondering how it might be possible for an output from tool A to be used as a parameter when calling tool B (without modifying tool B to take a data configuration file or anything like that).

To clarify, I'm thinking about simple parameters like a single integer/float/string like "median insert size" or "average coverage" which map to a single input parameter in a downstream tool.

Workflow parameters basic $variable support is similar to what I'm thinking about here, see: https://wiki.galaxyproject.org/DevNewsBriefs/2011_01_31

Perhaps one day in addition to tool outputs being files, they can include setting $variables (local to the history), which could then be used within downstream tools? I can see this has downsides with tracking the providence of each variable though... thus:

Another idea would be extra filetypes for each parameter type (Unix philosophy - make everything into a file), e.g. integer_parameter, float_parameter, ... which "Tool A" could define as an output, and write to. Galaxy would then need to offer these history entries to the user when a later tool (e.g. "Tool B") asked for a parameter of that type (e.g. an integer, or a float).

(This idea could be extended further, with data files using JSON or something to hold a structured set of parameter values which Galaxy could then offer to fill later tool's inputs.)

Crazy, or is there some mileage in this idea?

Peter