Improved integration with GLIS DOI

This issue improves the reporting of DOI relationships to GLIS DOI Registration service.

Background

DOI Relationships explains that:

Descriptor M04 method: Method tells us how the the PGRFA was obtained

GLIS currently provides 6 different methods that can be briefly described as follows. Please note that, as genetic transformations may be involved as well as just transfers or collection, we will use the terms "Predecessor" to identify the PGRFA that existed before the event and "Successor" the PGRFA that exists after the event.

Each method involves zero, one or more Predecessors as follows

Method	Code	Number of Predecessors
Acquisition	acqu	0 (e.g. for collection) or 1 (e.g. for transfers)
In-house copy	ihcp	1
In-house variant	ihva	1
Novel distinct PGRFA	nodi	1 or more
Observation - Natural	obna	0
Inherited	obin	0

Whenever a new PGRFA is registered in GLIS, using the appropriate method and providing the Predecessors, a relationship among such Predecessors and the Successor PGRFA (and therefore among the corresponding DOIs) is established.

Please note that the last two Methods do not really establish a relationship as there is no Predecessor and no Successor; likewise for Acquisition without a Predecessor, as in the case of collection from the field or acquisition from in situ conditions.

From DOI XML:

Descriptor R02 progdoi: DOI of the progenitor PGRFA as registered in GLIS. Please note that the number of allowed DOIs depends on the method M04.

The progdoi must be a valid DOI.

GGCE now uses acqu or obin method for accession DOI, but does not report progenitor DOIs.

Acquisition

This has two options: transfer and collection. In GGCE these are both recorded as AccessionSource. If there is a transfer (accession source of type DONATED) then it's a transfer, but the doi and accessionNumber of the accession at the donor institute are recorded as accession names. If we use a fixed name type for donor's DOI, we can register it as progdoi.

If there's a COLLECTED source, then there is no DOI for the source, except that one could potentially be registered as an in-situ PGRFA Observation - Natural obna in GLIS.

Regardless, we can use acqu if there exists an accession source COLLECTED or DONATED.

In-house copy and variant

Accessions in GGCE do not have a parent relationship (unlike parent inventory). But following the same logic as above, we could introduce a new accession source types COPY and VARIANT where this relationship is recorded. The logic for progdoi is then the same as above.

Novel distinct PGRFA

This represents breeding material and we use accession source DEVELOPED to indicate that the accession is a result of breeding.

The issue here is that PGRFA could be DEVELOPED and also DONATED. DONATED takes precedence and we would report the accession as acquired acqu instead of nodi. Only if there is no DONATED record we can assume that the accession was developed in house and use nodi.

Inherited

All other cases use obin.

Progenitor DOI

The DOIs that are reported as progdoi should be recorded as accession names of type PROGDOI. This gives us a fixed name type to use.

There can be 0 or more names of type PROGDOI for an accession and depending on accession sources we can:

• if there are 0 PROGDOI names, we use obin
• if there is 1, we can use acqu, ihcp, ihva or nodi
• if there are more, we can use nodi

GGCE should control how many PROGDOI names can be registered depending on the effective accession source: for DEVELOPED we allow many, for others we allow only one.

Changes to DOIs

A DOI of an accession or inventory must never be modified. The DOI could potentially be cleared, but this is problematic as GLIS does not allow for removals.

Updating GLIS by re-submitting a record should update progdoi and method according to the rules described above.

DOIs for inventories

We will need to add support for DOIs for inventories. Inventories must not use the same DOI as their accession! Inventory may get its own DOI and it have type ihva (a more generic version of ihcp).

If the inventory has a parent inventory (anywhere in the tree) that has a DOI, then that DOI is used as progdoi. In all other cases, the DOI of the accession is used.

An inventory should only get a DOI if the accession has a DOI. I think this is a reasonable constraint for GGCE.