GPL-559 Automatically batch lighthouse sentinel samples

[rl15]

User story GPL-559 | As genomics POs (Emma, Rich C & James) we would like Lighthouse sentinel samples to be automatically batched for cherry picking

Who are the primary contacts for this story Emma G Rich C James M Danni W

*Acceptance criteria To be considered successful the solution must allow:

• [ ] Extend GUI delivered in GPL-558 to allow selection and deselection of plates (and so the samples on the plate) - see mockup
• [ ] Indicate with feedback that plate has been removed from a batch*1
• [ ] Create cherry pick batch in sequencescape from selected plates
• [ ] feedback to user batch number when successfully created.

*1 There maybe a better way to feedback that a plate and so its samples have been excluded from the mega batch than a table underneath (EG big red cross next to item /greyed out?). However this is working design for now.

Dependencies This story has dependency on GPL-558: https://github.com/sanger/lighthouse-ui/issues/2 And: https://github.com/sanger/sequencescape/issues/2837

Additional context See sequence flow diagram: https://ssg-confluence.internal.sanger.ac.uk/x/jqsKBg

[rl15]

Mockup (mailed to SSR Tuesday, 14 July 2020 at 11:20): 20200714_GPL-559_Batch_Sentinal_CP.pdf

[JamesGlover]

This is a bit of a diversion from the current process for generating batches. Its one of the reasons I had hoped to get cherrypicking re-written at some point.

Hidden requirements:

• Sequencescape batches are collections of requests. So there will be a need to build requests.
• Requests are usually built via a submission/order. Given the vast majority of requests in Sequencescape (and all cherrypick requests) have a submission, this may be risky to bypass.
• Requests need some additional information: Study, Project, Request Type

Infrastructure in place:

• Submission templates provide API endpoint (currently just V1) to automatically build submissions
• This could automatically build the requests as needed, although its a little harder to add and remove requests after built.
• Submissions templates can have different builder classes (eg. To automatically build a batch)

• Its also possible fro requests to register a callback with the submission, which gets executed after the submission is built (example below) however this would require a different request class for lighthouse picks compared to normal cherrypick requests.

  class Request::AutoMultiplexing < Request::Multiplexing
  after_create :register_transfer_callback
  
  # Triggers immediate transfer into the tubes if the source asset already
  # exists. This allows multiplexing requests to be made on plates at the
  # end of library prep, after the plate is qc_complete.
  # If no asset is present then we haven't got to that stage yet and transfer
  # will be triggered as part of the standard workflow.
  def register_transfer_callback
  # We go via order as we need to get a particular instance of submission
  return if asset.blank?
  
  order.submission.register_callback(:once) do
    Transfer::FromPlateToTubeByMultiplex.create!(
      source: asset.plate,
      user: order.user
    )
  end
  end
  end

Other things to consider:

• Submission building is mostly asynchronous, which would be undesirable if you want immediate

[JamesGlover]

I'd suggest adding a 'Batched cherrypick' submission template, and then trigger that via the API. This would have the advantage that it would also expose the functionality within Sequencescape in general. That way we'd need:

• SubmissionTemplate should generate a 'BatchedLinearSubmission'
• BatchedLinear submission use standard submission behaviour, but batch up all requests of the first request type automatically.
• Expose the submission template endpoints via the V2 API (probably)
• Add a synchronous: true flag to submission to build them synchronously.
• Submissions have_many :batches, via: requests and then expose thos batches via the API

[JamesGlover]

OOh! Better idea! New endpoint: pick-list which just takes an array of receptacles, and study information, and does all this under the hood. That way if we get around to substantially changing the way cherrypicking works we can do so in a nicely isolate manner. From a suggestion to @stevieing

Add a new pick-list endpoint, which allows a ‘post’ containing a list of receptacles, and probably also allow a study/project as well. This will generate all the necessary requests and submissions behind the scenes. The response will contain a pick-list name and link. These will provide a link to the batch.

[emrojo]

I think this feature (create a batch automatically) is mainly a change in SS so is going to be implemented in https://github.com/sanger/sequencescape/issues/2837 , I dont see what should be changed in Lighthouse/Crawler. I'll add the 'Question' flag for this reason.

[JamesGlover]

I think this feature (create a batch automatically) is mainly a change in SS so is going to be implemented in sanger/sequencescape#2837 , I dont see what should be changed in Lighthouse/Crawler. I'll add the 'Question' flag for this reason.

There is the front end component in the app @stevieing is delivering as part of GPL-558 (sanger/lighthouse-ui#2) which will mostly be the API call to Sequencescape, and displaying the returned link.

[stevieing]

Tasks

• [x] create a new endpoint in Labwhere to retrieve labwares for multiple locations
• [ ] create a new page in lighthouse ui to create a mega cherrypicking batch from multiple box barcodes
• [ ] create an api request in the lighthouse ui to create mega batch in sequencescape
• [x] update the process flow diagram in confluence to add the sentinel batch create and cherrypicking

[stevieing]

Sequencescape endpoint:

{
          'data' => {
            'type' => 'pick_lists',
            'attributes' => {
              'asynchronous' => true,
              'labware_pick_attributes' => [
                { source_labware_id: plate_1.id },
                { source_labware_barcode: plate_2.machine_barcode },
              ]
            }
          }
        }