make analyte.primary_biosample_id required when adding RNA-seq or long read tables?

[bheavner]

• if primary_biosample_id is optional - would not it mean that not everyone will use it?

Yes, but that’s okay - the primary use case would be if multiple analytes were extracted and experiments run from a single biosample (e.g. if DNA and RNA were extracted from whole blood). In this case, a primary_biosample_id would allow linking the DNA and RNA data to the same biosample; however, in the majority of cases, RCs would only conduct one analysis (DNA sequencing) from a biosample. This is particularly true for V1 of the data model, which focuses on supporting short-read genetic sequencing data. As the model moves towards V2, this field could become required (in which case we would likely use the analyte.analyte_id for any missing values).