Added a metadata table - one-to-many relationship apporach

[waterflow80]

Description

closes #13

In this PR, we've separated the seqcol metadata from the seqcol entities, by storing them into a new entity/table, following a
one to many relationship:

classDiagram
    class SeqColLevelOneEntity
    SeqColLevelOneEntity: - digest
    SeqColLevelOneEntity: - seqColLevel1Object
    class SeqColMetadata
    SeqColMetadata: - seqColDigest
    SeqColMetadata: - sourceIdentifier
    SeqColMetadata: - sourceUrl
    SeqColMetadata: - namingConvention
    SeqColMetadata: - timestamp
    SeqColLevelOneEntity "1" --> "1..*" SeqColMetadata

Loading

Here's a sample content of the seqcol_md table:

Changes

• Created the entity SeqColMetadataEntity with many-to-one relationship to the seqcolLevel1
• Added the SeqColMetadataDTO class to format the metadata
• Set the timestamp automatically using the @Temporal and the @CreationTimestamp annotations
• Check for the existence of the asm_accession before downloading.
• Added a new endpoint /eva/webservices/seqcol/metadata/{digest} to fetch the metadata of a specific seqcol in the following format:

[
    {
        "sourceId": "GCA_000146045.2",
        "sourceUrl": null,
        "namingConvention": "UCSC",
        "timestamp": "2024-04-12T21:19:13.686+00:00"
    }
]

Todo

• Write some test cases to ensure the efficiency of the new model.
• Consider passing the metadata object as argument to some construction methods, instead of creating the metadata inside of these methods. Like here for example.
• Complete the swagger docs for the new metadata endpoint.

Discussion

Currently, the service only checks whether the given asm_accession is already ingested (exists in the metadata table), if so it doesn't proceed with the download.
I assume we should make it able to check if all the metadata information is the same in order to abort the download, however we still don't have the required mechanisms to test that. (because we only have an ingestion endpoint with asm_accesion In).

After we complete that, we should complete the logic of updating the metadata of a defined seqcol.

Note: the new seqcol_md table has only a simple integer as primary key, so all other attribute values can be duplicated in other entries, and I think this will fulfill all the required use cases, described here by @apriltuesday.

We may eventually have to check for all attributes equality in the code to avoid full entry duplication.

[waterflow80]

@nitin-ebi @apriltuesday
Just wanna confirm that this is right implementation, because it didn't run otherwise, it returned an error:

org.hibernate.PropertyValueException: not-null property references a null or transient value : uk.ac.ebi.eva.evaseqcol.entities.SeqColMetadataEntity.seqColLevelOne

Originally, I thought that the initialization would be made by JPA automatically.

[apriltuesday]

Sorry we missed this question when we met yesterday... Others might be more familiar with Hibernate/JPA, but I think this seems right - it's similar to this question where the answers suggest you do have to set the parent explicitly.

[waterflow80]

Clear,
Thanks April.