Standards Call 2021-06

Agenda

How to continue with Travis (see #484). Also: Can we reduce the resource usage of the R validation routines (currently take 10 min. and seems to download half of Bioconductor)...
Move on with single-cell data representation (#409)
Draft Germline schema (#530)
Draft repertoire grouping (#533)
Bus protection status

Travis: Discussed general issues around costs for using Travis as CI and how to be more efficient with these resources. Will continue in the next call. Until then:
- Jason will check for containerized solutions in R
- Artur will see what resources INESC can offer
- Scott will shut off Travis test for branches, only PRs will be tested.
Single-cell: Restarting the discussion around on-disk representation of single-cell related data (#409, the recent comments start here). Questions: To which extent do we support inferred cells (i.e., stochastic chain pairing), how many different possible representations would we allow for the same data, how do we link between objects, and do the Schema, the API and the on-disk representation have to be identical? Discussion has moved to Github, but the existence/non-existence of files should be solved via the should be solved via manifest
Germline schema: OGRDB schema has been proposed to be merged into AIRR Schema via PR #530. Further thoughts on a general schema for germline information repos based on discussions within GLDB WG can be found in this GDoc and this Miro board.
Can the proposed Person object be an ORCID?
There are some compatibility breaks with PR #502, but we should proceed with the merge. We will assess the naming and behavior of the validation functions at a later date to ensure consistency across the R and python libraries.
Brief discussion on Manifest and DataFile.