Dbgap ResearchStudy_Content extensions (experimental)

[bwalsh]

This experimental/draft PR:

• Adds gofsh reverse engineering tool to Dockerfile
• Documents a general method to import unpublished extensions with their namespace into out IG

Discussion points, Next steps

• This PR imports dbGAP's aggregate counts ResearchStudy_Content.
  • I realize and support our position of modeling aggregates as Observations on the ResearchStudy, and think that doing a deeper dive into what dbGAP has in place will accelerate consensus.
• For example, should we support DbGaPConcept(s):
  • DbGaPStudyIdentifier
  • HarmonizationUnit
  • HarmonizedVariableAccession
  • SecurityStudyConsent
  • SubjectIdentifier
  • TopMedHarmonizedVariableId
• Other extensions to evaluate include:
  • StudyConsents
  • StudyDesign
  • see dbGAP FHIR Presentation

Import existing FHIR Resources into FSH

This process is useful for importing FHIR resources that are not published i.e. not in simplifier.net

To retrieve and install the Extensions, in this case dbGAP.

# From the root directory:
# Make a work directory, retrieve source extensions
mkdir -p dbgap/input
wget https://dbgap-api.ncbi.nlm.nih.gov/fhir/x1/StructureDefinition?type=Extension -O dbgap/input/dbgap-extentions.search-bundle

# Parse the search bundle into separate resource files
python3 fhir/ncpi-fhir-ig/dbgap/bundle_writer.py 

# Run goFSH to reverse engineer the extensions
rm -rf dbgap/gofsh
fsh gofsh /src/dbgap --out /src/dbgap/gofsh

Manual steps:

• Examine outputs in fhir/ncpi-fhir-ig/dbgap/gofsh/input/fsh/extensions.
• Edit as appropriate:
  • ResearchStudy_Content.fsh: Added ^url
  • ResearchStudyContent*.fsh: Added ^url; Removed ^context.type ^context.expression
• Copy to fhir/ncpi-fhir-ig/input/fsh
• Generate IG as ususal

Results

[RobertJCarroll]

Thanks for showing these! It looks like you have filtered to some of the content count ones- how hard is it to retrieve all of the extensions?

[bwalsh]

@RobertJCarroll Yes, I focused on the aggregations. It should not be hard to incorporate others of interest.

[RobertJCarroll]

Closing this PR to come up with a strategy in #55 .