docs for tool_inputs parameter to run_tool() about bioblend HOT 9 CLOSED

Using the state_inputs it works.
Thanks a lot!

from bioblend.

Certainly there is no top-down desire to push people toward workflows over tools - like most of the API development it is entirely organic. More people have made use of the workflow API - it has had more eyes, more community development, etc... than the tools client and API. There has been an open Trello card with the feature request for sometime - https://trello.com/c/67HsrgVm - I will try to get to it. In the mean time checking out the API tests for tools might elucidate things slightly - https://bitbucket.org/galaxy/galaxy-central/src/5f89fecb2f0703ce237cc185b2ccbb2bc0fd2658/test/api/test_tools.py?at=default. If you really want to get into the weeds the tool test framework targets the API so it has more details - https://bitbucket.org/galaxy/galaxy-central/src/5f89fecb2f0703ce237cc185b2ccbb2bc0fd2658/test/base/interactor.py?at=default#cl-143.

from bioblend.

Just to be spell things out at a high-level - inputs should just be a dict with the input parameters as keys and the values as values. Parameters nested inside of conditionals should have the form cond_name|input_name, repeat parameters should likewise be nested but with an index appended to the repeat name repeat_name_0|input_name , repeat_name_1|input_name, etc... Values are fairly straight forward except for dataset inputs - these should take the form - {'src': 'hda', 'id': '<encoded_dataset_id>'}. There are many more details - some that I don't even understand or haven't worked through. Like I said though, I will try to find some time to document these in the next couple weeks.

from bioblend.

Hello,

I'm having some problems with the tool_inputs to run tools, especially setting the dataset.
Here is a simple example with fastq_info:

inputs = {"input_1": "bbd44e69cb8906b550b38de0ec289fcd", 'r': 'true'}
gi.tools.run_tool("b47d6ac96387d407" , gi.tools.show_tool("fastq_info")["id"], tool_inputs=inputs)

I checked the inputs with the show_tool method, and I'm able to change inputs whose argument is not None (like 'r', in the example above). However I cannot set the dataset, or other inputs whose argument is None.
How should I solve this?

Thanks in advance

from bioblend.

@ecunha1996 I guess that this should work for you:

inputs = {"input_1": {src: "hda", id: "bbd44e69cb8906b550b38de0ec289fcd"}, "r": "true"}

from bioblend.

@nsoranzo I tried that but it keeps ignoring the dataset I select. I ran the tool directly in the galaxy portal, and than checked the inputs with the show_job method, and looked like this:

'inputs': {'fastq_1': {'id': 'bbd44e69cb8906b5b717152b0718556f',
'src': 'hda',
'uuid': 'aaf14c77-4631-4406-9b88-bb30bbe6770f'}}

Instead of using "input_1" I tried to use "fastq_1", but the problem remains.

from bioblend.

@ecunha1996 I think show_tool() is the wrong thing to check. The state_inputs returned by build() should be correct. That gives

{"input_option": {"input_type": "single", "fastq_1": {"values": [{"id": "bbd44e69cb8906b5b717152b0718556f", "src": "hda"}]}}, "q": "true", "r": "false"}

I hope that works for you? This should be better documented for sure. :/

from bioblend.

I wrote a sentence in the docs directing users to the build() method: 984421d

Not sure if that is enough to close this issue?

from bioblend.

Yes, that's a good start, thanks @simonbray !

from bioblend.