feat(configs): Ena metadata updates (#2423)

* Rename bioproject_accessions to bioproject_accession

* sra_run_accession is per sequence not per segment

ena-submission/config/config.yaml

@@ -58,7 +58,7 @@ organisms:
         name: insdc_accession_full_S
         type: string
       - externalMetadataUpdater: ena
-        name: bioproject_accessions
+        name: bioproject_accession
         type: string
       - externalMetadataUpdater: ena
         name: biosample_accession
@@ -73,13 +73,7 @@ organisms:
         name: ncbi_virus_tax_id
         type: int
       - externalMetadataUpdater: ena
-        name: sra_run_accession_L
-        type: string
-      - externalMetadataUpdater: ena
-        name: sra_run_accession_M
-        type: string
-      - externalMetadataUpdater: ena
-        name: sra_run_accession_S
+        name: sra_run_accession
         type: string
   ebola-sudan:
     ingest:
@@ -105,7 +99,7 @@ organisms:
         name: insdc_accession_full
         type: string
       - externalMetadataUpdater: ena
-        name: bioproject_accessions
+        name: bioproject_accession
         type: string
       - externalMetadataUpdater: ena
         name: biosample_accession

ena-submission/test/ena_submission_list.json

@@ -100,9 +100,7 @@
             "total_ambiguous_nucs_L": 0,
             "total_ambiguous_nucs_M": 0,
             "total_ambiguous_nucs_S": 0,
-            "bioproject_accessions_L": null,
-            "bioproject_accessions_M": null,
-            "bioproject_accessions_S": null,
+            "bioproject_accession": null,
             "food_product_properties": null,
             "host_vaccination_status": null,
             "quality_control_details": null,

ena-submission/test/external_metadata_test.ndjson

@@ -1 +1 @@
-{"accession": "LOC1", "version": 1, "externalMetadata": {"ncbi_release_date": null,"ncbi_update_date": null,"ncbi_submitter_country": null,"insdc_accession_base": null,"insdc_version": null,"insdc_accession_full": null,"bioproject_accessions": null,"biosample_accession": null,"ncbi_sourcedb": null,"ncbi_virus_name": null,"ncbi_virus_tax_id": null,"sra_run_accession": null}}
+{"accession": "LOC1", "version": 1, "externalMetadata": {"ncbi_release_date": null,"ncbi_update_date": null,"ncbi_submitter_country": null,"insdc_accession_base": null,"insdc_version": null,"insdc_accession_full": null,"bioproject_accession": null,"biosample_accession": null,"ncbi_sourcedb": null,"ncbi_virus_name": null,"ncbi_virus_tax_id": null,"sra_run_accession": null}}

ingest/config/defaults.yaml

@@ -29,7 +29,7 @@ keep:
 # Used to deduplicate samples
 # Used for segmented viruses only
 shared_fields:
-  - bioproject_accessions
+  - bioproject_accession
   - biosample_accession
   - geo_loc_country
   - geo_loc_admin_1

ingest/tests/config_cchf/config.yaml

@@ -14,7 +14,7 @@ nucleotide_sequences:
   - S
 organism: cchf
 rename:
-  bioprojects: bioproject_accessions
+  bioprojects: bioproject_accession
   country: geo_loc_country
   division: geo_loc_admin_1
   genbank_accession: insdc_accession_full

kubernetes/loculus/values.yaml

@@ -456,7 +456,7 @@ defaultOrganismConfig: &defaultOrganismConfig
         noInput: true
         perSegment: true
         oneHeader: true
-      - name: bioproject_accessions
+      - name: bioproject_accession
         customDisplay:
           type: link
           url: "https://www.ncbi.nlm.nih.gov/bioproject/__value__"
@@ -941,7 +941,6 @@ defaultOrganismConfig: &defaultOrganismConfig
         header: "INSDC"
         ingest: ncbi_sra_accessions
         noInput: true
-        perSegment: true
       - name: total_snps
         type: int
         header: "Alignment states and QC metrics"