Agenda

  • Patient/donor sample tracking (Bill Shirey - HuBMAP)
    • Use of tracking software

Call Notes & Action Items

  • Lab creates donor ID and records any internal ID they may be using
  • Initial data may not be as curated/complete but HuBMAP leadership has made the decision to accept all preliminary data
  • Modeling organs because we begin with whole organ as a start (OPO organ provider organizations)
  • Does UUID have any specific length, e.g. after 1000? - never more than 10,000 donors per tissue center
  • Local uniqueness with IDs that would not happen if they were lab-based, lab IDs entered so they can be cross-referenced. No ID clashes yet
  • No PHI included for Donors

Chuck Borromeo on Neo4J:

  • Graph db behind the scenes linking all data in hierarchy, find one sample generated by Bill and show linkage back to TMC generating data
  • Backend to store data

  • Left to right: TEST TMC, the various TMS (Vanderbilt, Florida, etc.), connected to Registered Donors, to the organ, and from the organ, a small hierarchy of downstream specimens can be created. Metadata is associated with each module

  • Similar to HCA model with slight differences, some differences due to not knowing what provenance is needed. HCA metadata has been looked at for sequencing, use and contribute back - issues welcome on GitHub repo, can always set up ad hoc calls

  • Everything is visible to centers in web interface, search and find registered samples and see the details, hierarchical representation not there yet but in an upcoming release
  • Create and edit capabilities on anything input from your lab
  • If someone downloaded a dataset and did their own analysis, someone versioned the data for any reason, anyone who downloaded it will need to get notified
  • QA steps? Editing on lab end, review on back end
  • Two states of data: staged and published
  • Adding new fields is easier, adding new/editing graphs takes a lot of work