Complete, end-user control over how and where Galaxy reads and writes files would be particularly useful for dealing with very large files and for integrating Galaxy output into complex, ad hoc external workflows.
I have three suggestions:
1. Rename the "Upload File" tool as "Data Source" and allow users to select whether the specified file should be uploaded to "Galaxy" or read in situ.
2. Create a new tool called "Default Data Destination" with which users can specify the default path where all data output is to be written.
3. Add a "Data Destination" field to the page of every tool that writes output to disk, so users can specify, on a tool-by-tool basis, where they want the output to be written. a. If the user has specified a default location via the "Default Data Destination" tool, the field would be pre-filled with this path; b. if not, the field would be blank (or contain "to Galaxy" in disabled text), and Galaxy will write to its write folder, c. unless the user explicitly specifies a new write path in this field for this tool.
I am aware of the various, current ways one can specify file handling strategies within Galaxy, but these seem to be too static, too global, and too "subliminal" for the kind of work at hand; something finer- grained, more pervasive, and more explicit seems to be called for.
[If implemented, it is clear that these changes would add fragility to the Galaxy history mechanism, but this could be dealt with at multiple levels, including in the UI by issuing warnings and/or offering choices.]
Thanks for all your great work on this fine platform for tool integration. (In the words of one of your very knowledgeable "competitors" on seeing Galaxy for the first time, "This is slick!" )
Dean A. Snyder Senior Programmer/Analyst Center for Inherited Disease Research (CIDR) Johns Hopkins School of Medicine Bayview Research Campus 333 Cassell Dr, Triad Bldg, Suite 2000 Baltimore, MD 21224 www.cidr.jhmi.edu cell:717 668-3048 office:410-550-4629