Re: [galaxy-dev] directories as inputs/dataset file extensions
On Apr 4, 2012, at 11:35 AM, Langhorst, Brad wrote:
Given that it describes where the rest of the files are, maybe it would work to pass only the CONTENTS.json file. Then you can just build your references in a directory structure on a shared file system and keep absolute paths in the CONTENTS.json file.
If you don't want to modify the tool to accept only the json file, you could write a simple wrapper to read the json and symlink everything into place before calling your tool.
I think this is the simplest solution for now.
This is somewhat akin to the reference area galaxy uses to store reference sequences (and their various indices, formats, annotations etc). It might be possible to modify galaxy's reference infrastructure to present these pre-built reference packages in the same way a user chooses hg19, or E. coli K12. The parse JSON option looks easier to me
This seems like something I should look into, at least. These don't come with Galaxy, do they?
I'm not aware of other galaxy tools that pass "directories" in this way, but they may exist.
I haven't found anyone who seems to be doing the same thing, at least.
I don't know if you intended to go off-list, feel free to go back on if you want to.
My mistake; I resent my e-mail back on the list as well, but it's pending moderation at the moment. ~Aaron
On Apr 4, 2012, at 2:44 PM, Aaron Gallagher wrote:
On Apr 4, 2012, at 11:35 AM, Langhorst, Brad wrote:
This is somewhat akin to the reference area galaxy uses to store reference sequences (and their various indices, formats, annotations etc). It might be possible to modify galaxy's reference infrastructure to present these pre-built reference packages in the same way a user chooses hg19, or E. coli K12. The parse JSON option looks easier to me
This seems like something I should look into, at least. These don't come with Galaxy, do they?
No, one installs those references independently. see http://wiki.g2.bx.psu.edu/Admin/NGS%20Local%20Setup specifically the section on reference genomes . The more I think about this, the more i think it might make sense to explore setting up you own loc file - might not be that tough. I've never done that, so maybe somebody else who knows more about this will comment on that. Brad -- Brad Langhorst langhorst@neb.com 978-380-7564
participants (2)
-
Aaron Gallagher
-
Langhorst, Brad