(Investigate) Support for handover method for data access

[blankdots]

Description

Investigate how to implement support for ga4gh-beacon/specification#114

DoD (Definition of Done)

Figure out if Handover method support is possible and how, add PoC (Proof of Concept).

Testing

Peer review, if implemented unit tests

[MalinAhlberg]

The outcome of our discussions so far:

Quick&easy setup / a proof of concept implementation that could be improved later on:
​

• a Beacon instance keeps an url to a DRS in its config
• for each answer, the Beacon will construct dataset handovers containing all necessary information,
  for example datasetname+region+ref/alt (eg swegen-22-46615880-46615880-A-G)
  ​

"datasetHandover": [
      {
        "handoverType": {
          "id": "Custom",
          "label": "Variants"
        },
        "description": "retrieve data of the variants matched by the query",
        "url": "https://drs_from_config?dataset=swegen&chrom=19&46615715&end=46615880"
      }
  ]

​

• a list of id, description and labels could also be kept in the config
  ​
• Example of what information you could get if you follow the links:
  • a browsing page https://swefreq.nbis.se/dataset/SweGen/browser/region/22-46615715-46615880
  • a page with information and contact data about the data owner
    ​
• Drawbacks:
  • not very flexible, all datasets get the same set of handovers
    ​
• Questions:
  • does the Beacon Network expect some specific format/information? For example, a fixed set of id and label, a maximum number of handovers (@teemukataja )..?
  • what to put in the beaconhandover?

[blankdots]

Drawbacks:
not very flexible, all datasets get the same set of handovers

@viklund & @MalinAhlberg how about making use of the metadata for a dataset (https://github.com/CSCfi/beacon-python/blob/master/data/example_metadata.json) to specify what handovers each dataset could have. This does imply changes to the database maybe a new column and potentially an additional table.

This issue would need changes to the JSON schema to validate the new Handover things.
Not clear to me how this would work with the Permissions model for beacon Public, Registered, Controlled.

what to put in the beaconhandover?

That is also unclear to me and this is the closest I found as a clear explanation:

e.g. variant data, phenotypic information, low-level sequencing results)

https://beacon-project.io/roadmap/handover.html
And I have read the comments on the issue, and it sounds to me like some sort of HATEOAS HAL.

[blankdots]

Well I asked just to be sure: ga4gh-beacon/specification#230 (comment)

[MalinAhlberg]

@blankdots, thanks for the feeback!

We have started implementing a version of this on our fork of the repo, just to get started:
https://github.com/NBISweden/beacon-python/tree/feature/add_handover
Hopefully we will have a demo version running this week. It will be smart (no changes to the metadata, no any specific connection to permissions), but we will get back to those parts...

what to put in the beaconhandover?

For now, we just have links to all accessible datasets, eg https://swefreq.nbis.se/dataset/SweGen

[blankdots]

@MalinAhlberg ok, once you see fit it would be nice if you could bring the changes here with a PR.

[blankdots]

For tracking purposes, started this issue: ga4gh-beacon/specification#276