However this is useful only when creating new seqCol objects and populating our database.

We should be using the Assembly accession to check that an assembly has been ingested already.

@tcezard
I think this topic is more related to the spec itself, I'm not sure if we should keep this issue on the eva-seqcol repo.

I've said above that we should be storing the Assembly accession but we should also try to think about other potential source of sequence collection and how they would impact our design.
The main goals of this issue are:

• storing an identifier of the ingested sequence collection so we do not need to retrieve the data again if it has been done before
• store enough metadata so we can provide the origin of the sequence collection

At the moment the main identifier is the assembly accession since it is what we use in the ingestion parameter so it makes sense to store this. However it won't be enough to know exactly where the sequence collection comes from. For this I think storing the source URL would be more accurate.

I also think that in the future we will want to be able to store sequence collection that are not linked to INSDC accession. These most likely will have URL and we will also need to find some form of identifier associated. To enable this we should chose generic column names such as datasource identifier rather than insdc accession

All these points makes me think we are building a set of metadata associated with the sequence collection which should be stored in a separate table than the one we already use for storing the digests and JSON objects.

There is the question of the same metadata being used for multiple collections because they have different naming convention and the possibility that the same sequence collection could exist in different source. For this, I'm thinking we might need a many to many relationship between a sequence collection metadata set and a sequence collection.
We could also chose to extract the naming convention to the metadata table which could potentially change how the relationship between the two table is defined.

Concretely there are 3 pieces of metadata that I think are very relevant for a sequence collection:

• Source identifier
• Source URL
• Naming convention

There could also be some other information we might want to store like ingestion time if we ever want to list all sequence collections that have been ingested before/after specific time.