search

brainstem-org / brainstem_python_api_tools

Python toolset for interacting with the REST API of BrainSTEM
1 stars 1 forks source link

Path to data files #4

Open luiztauffer opened 1 year ago

luiztauffer commented 1 year ago

The path to the data files in globus seems to be incomplete, e.g. 

'data_protocols_json': [{'path': 'https://app.globus.org/file-manager?origin_id=188a6110-96db-11eb-b7a9-f57b2d55370d&origin_path=%2F',`

this will take me to the projects page in Globus, but not to the files for that particular experimental session. Is that a bug? What would be the recommended way to find the data files for any specific experimental session, given that I know the dataset id?

petersenpeter commented 1 year ago

Hi Luiz, is this a link from a project or a dataset?

luiztauffer commented 1 year ago

hi @petersenpeter the data comes from requesting the project associated with a dataset, now I realized that indeed the globus link I pasted is probably correct, it is the root folder of a project with many experiments. Is there a way to map the globus link to the files related to a specific BrainSTEM dataset?

petersenpeter commented 1 year ago

The user can link to a specific folder or file in globus for either a project or a dataset. For datasets, the idea is that the user links to a folder or file, and with further development use their API to initialize a data transfer. But at this point, we can only provide URLs to data repositories.

URLs can also be made to a Datadiset, e.g. BrainSTEM Project URL to a dandiset: https://dandiarchive.org/dandiset/000059 BrainSTEM dataset URL (Peter-MS22-180623-170141-concat) to dataset inside the project: https://api.dandiarchive.org/api/assets/d8ac6e2a-26e6-457d-9ac1-97d3421078ea/download?content_disposition=inline

luiztauffer commented 1 year ago

That’s really cool that you have already recordings pointing to DANDI sets! What we’re trying to accomplish now is somehow the other way around:

I guess we could skip the ones that are already pointing to dandisets =) Then we’re left with the problem of mapping each BrainSTEM subject/dataset to the correct files in globus I believe we’re meeting tomorrow with Rodrigo to discuss those ideas