I've given up on that approach to nested workflows - too inefficient,
too fragile. If I have lots of time (and a grad student to help out)
I've got some ideas about having a deep look at the tool runner side of
Galaxy with an eye to efficiently implementing the mapping of input data
to (sub)workflows. Right now the process of spawning a new Galaxy task
for e.g. each line in a tabular file is just too heavyweight, thus the
sub-workflow got implemented in Python.
On 24/05/2012 04:30, Jeremy Goecks wrote:
> I'm not ignoring you. However, there are others on the Galaxy team that are
> more familiar with the API and can provider better answers. I expect they'll
> chime in soon to address your questions.
> On May 20, 2012, at 5:39 PM, Peter van Heusden wrote:
>> Hi Jeremy
>> I'm need this for something I'm implementing at the moment, and how I'm
>> thinking about it is making a tool that uses the API to call a workflow.
>> There are a few problems though, correct me if I'm wrong:
>> 1) In order to make an input history item available to the called
>> workflow, the tool needs to somehow know about history items, but the
>> tool xml passes in parameters as data files. This could probably be
>> remedied by providing a type="history_item" parameter to <param> that
>> would provide the id associated with the history item. In the interem,
>> just to test things, I'm passing in parameters as a history:history_item
>> string (yeah I know, ugly!).
>> 2) My particular tool needs to take a history item, splits it into
>> partitions, and call a workflow with each of those partitions. For this
>> to work, the partition needs to be uploaded as a new history item, but
>> that is currently not possible. The other possibility is to create a
>> tool that does the split, have it in a single-tool workflow (because
>> workflows can be called from the API in such a way that their output
>> goes to a new history, whereas I don't see that in the tool interface)
>> and then iterate through the history that contains the split data,
>> calling the analysis workflow on each item.
>> P.S. for my particular problem - call a bunch of tools, once for each
>> row in a file of tabular data - it would be WAY easier to just write
>> everything in a Python script, but I'm trying to see what is do-able
>> within Galaxy.
>> On 20/05/2012 16:55, Jeremy Goecks wrote:
>>>> Is there any way we can speed up the implementation of this issue?
>>> Community contributions and always encouraged and welcomed. Partial
>>> solutions are fine, and self-contained contributions are likely to be
>>> included more quickly because they are easier to review.
>> Please keep all replies on the list by using "reply all"
>> in your mail client. To manage your subscriptions to this
>> and other Galaxy lists, please use the interface at:
Please keep all replies on the list by using "reply all"
in your mail client. To manage your subscriptions to this
and other Galaxy lists, please use the interface at: