hubmapconsortium / ingest-pipeline

Data ingest pipeline(s) for QA/metadata etl/post-processing

HuBMAP Ingest Pipeline

About

This repository implements the internals of the HuBMAP data repository processing pipeline. This code is independent of the UI but works in response to requests from the data-ingest UI backend.

Using the devtest assay type

devtest is a mock assay for use by developers. It provides a testing tool controlled by a simple YAML file, allowing a developer to simulate execution of a full ingest pipeline without the need for real data. To do a devtest run, follow this procedure.

1. Create an input dataset, for example using the ingest UI.

• It must have a valid Source ID.
• Its datatype must be Other -> devtest

2. Insert a control file named test.yml into the top-level directory of the dataset. The file format is described below. You may include any other files in the directory, as long as test.yml exists.
3. Submit the dataset.

Ingest operations will proceed normally from that point:

1. The state of the original dataset will change from New through Processing to QA.
2. A secondary dataset will be created, and will move through Processing to QA with an adjustable delay (see below).
3. Files specified in test.yml may be moved into the dataset directory of the secondary dataset.
4. All normal metadata will be returned, including extra metadata specified in test.yml (see below).

The format for test.yml is:

{
  # the following line is required for the submission to be properly identified at assay 'devtest'
  collectiontype: devtest,
  
  # The pipeline_exec stage will delay for this many seconds before returning (default 30 seconds)
  delay_sec: 120,
  
  # If this list is present, the listed files will be copied from the submission directory to the derived dataset.
  files_to_copy: ["file_068.bov", "file_068.doubles"],
  
  # If present, the given metadata will be returned as dataset metadata for the derived dataset.
  metadata_to_return: {
    mymessage: 'hello world',
    othermessage: 'and also this'
  }
}

API

API Test
Description	Test that the API is available
HTTP Method	GET
Example URL	/api/hubmap/test
URL Parameters	None
Data Parameters	None
Success Response	Code: 200 Content: {"api_is_alive":true}
Error Responses	None

Get Process Strings
Description	Get a list of valid process identifier keys
HTTP Method	GET
Example URL	/api/hubmap/get_process_strings
URL Parameters	None
Data Parameters	None
Success Response	Code: 200 Content: {"process_strings":[...list of keys...]}
Error Responses	None

Get Version Information
Description	Get API version information
HTTP Method	GET
Example URL	/api/hubmap/version
URL Parameters	None
Data Parameters	None
Success Response	Code: 200 Content: {"api":API version, "build":build version}
Error Responses	None

Request Ingest
Description	Cause a workflow to be applied to a dataset in the LZ. The full dataset path is computed from the data parameters.
HTTP Method	POST
Example URL	/api/hubmap/request_ingest
URL Parameters	None
Data Parameters	provider : one of a known set of providers, e.g. 'Vanderbilt' submission_id : unique identifier string for the data submission process : one of a known set of process names, e.g. 'MICROSCOPY.IMS.ALL'
Success Response	Code: 200 Content:{ "ingest_id":"some_unique_string", "run_id":"some_other_unique_string" }
Error Responses	Bad Request:   Code: 400   Content strings:     "Must specify provider to request data be ingested"     "Must specify sample_id to request data be ingested"     "Must specify process to request data be ingested"     "NAME is not a known ingestion process" Unauthorized:   Code: 401   Content strings:     "You are not authorized to use this resource" Not Found:   Code: 404   Content strings:     "Resource not found"     "Dag id DAG_ID not found" Server Error:   Code: 500   Content strings:     "An unexpected problem occurred"     "The request happened twice?"     "Attempt to trigger run produced an error: ERROR"