DH modularity and 1:N wish list

[griffie]

A sample can contain multiple organisms, multiple kinds of the same organism (i.e. multiple isolates), and isolates may be sequenced multiple times using different protocols or instruments. This creates a 1-to-many issue, where one sample may need to be linked to multiple organisms, isolates, library IDs, associated tests (AMR drug panels from different companies) etc.

Currently the contextual data for organisms, isolates etc from the same sample have to be entered repeatedly over and over again which creates a data entry burden for data providers.

Ideally, modularity could be created so that sample information could be entered once and linked to different isolates.
Similarly, isolate information could be entered once and linked to different libraries with different processing details/instruments.
Also similarly, libraries could be linked to multiple sequencing runs and/or associated tests.

To submit the data to LIMS or public repositories, every library or isolate or organism would need the metadata from the sample so
ideally upon export, the DH would populate that info and present each thing as a separate line in a spreadsheet.
e.g. the above situation would appear like:
sample 1 --> organism 1 --> isolate A --> library 1 --> sequence 1
sample 1 --> organism 2 --> isolate B --> library 2 --> sequence 2
sample 1 --> organism 2 --> isolate C --> library 3 --> sequence 3
sample 1 --> organism 2 --> isolate C --> library 4 --> sequence 4
sample 1 --> organism 2 --> isolate C --> library 4 --> sequence 5
*But the data provider wouldn't have to enter the different metadata multiple times.

Can we make the DH do this modular/1:N data capture and transformation (pretty please)?

[ddooley]

This is all being handled over on main DataHarmonizer development repo. But we do need the GRDI 1-n table model developed still.