Create sample.xml objects to send to ENA using PHAG4E's metadata fiel…

…d mapping. Keep state in sample_table. Send slack notifications if submission fails.

ena-submission/Snakefile

@@ -104,6 +104,21 @@ rule create_project:
         project_created=touch("results/project_created"),
     params:
         log_level=LOG_LEVEL,
+    shell:
+        """
+        python {input.script} \
+            --config-file {input.config} \
+            --log-level {params.log_level} \
+        """
+
+rule create_sample:
+    input:
+        script="scripts/create_sample.py",
+        config="results/config.yaml",
+    output:
+        sample_created=touch("results/sample_created"),
+    params:
+        log_level=LOG_LEVEL,
     shell:
         """
         python {input.script} \

ena-submission/config/defaults.yaml

@@ -10,3 +10,46 @@ ena_submission_url: https://wwwdev.ebi.ac.uk/ena/submit/drop-box/submit # TODO(h
 github_username: fake_username
 github_pat: fake_pat
 github_url: https://api.github.com/repos/pathoplexus/ena-submission/contents/test/approved_ena_submission_list.json?ref=main # TODO(https://github.com/loculus-project/loculus/issues/2425): update in production
+metadata_mapping:
+  'subject exposure':
+    loculus_fields: [exposure_event]
+  'type exposure':
+    loculus_fields: [exposure_event]
+  hospitalisation:
+    loculus_fields: [host_health_state]
+    function: match
+    args: [Hospital]
+  'illness symptoms':
+    loculus_fields: [signs_and_symptoms]
+  'collection date':
+    loculus_fields: [sample_collection_date]
+  'geographic location (country and/or sea)':
+    loculus_fields: [geo_loc_country]
+  'geographic location (region and locality)':
+    loculus_fields: [geo_loc_admin_1]
+  'sample capture status':
+    loculus_fields: [purpose_of_sampling]
+  'host disease outcome':
+    loculus_fields: [host_health_outcome]
+  'host common name':
+    loculus_fields: [host_name_common]
+  'host age':
+    loculus_fields: [host_age]
+  'host health state':
+    loculus_fields: [host_health_state]
+  'host sex':
+    loculus_fields: [host_gender]
+  'host scientific name':
+    loculus_fields: [host_name_scientific]
+  'isolate':
+    loculus_fields: [specimen_collector_sample_id]
+  'collecting institution':
+    loculus_fields: [sequenced_by_organization, author_affiliations]
+  'receipt date':
+    loculus_fields: [received date]
+  'isolation source host-associated':
+    loculus_fields: [anatomical material, anatomical part, body product]
+  'isolation source non-host-associated':
+    loculus_fields: [environmental site, environmental material]
+  'authors':
+    loculus_fields: [authors]

ena-submission/flyway/sql/V1__Initial_Schema.sql

@@ -10,6 +10,7 @@ CREATE TABLE submission_table (
     finished_at timestamp,
     metadata jsonb,
     unaligned_nucleotide_sequences jsonb,
+    center_name text,
     external_metadata jsonb,
     primary key (accession, version)
 );
@@ -22,6 +23,7 @@ CREATE TABLE project_table (
     status text not null,
     started_at timestamp not null,
     finished_at timestamp,
+    center_name text,
     result jsonb,
     primary key (group_id, organism)
 );

ena-submission/scripts/create_project.py

@@ -143,7 +143,10 @@ def submission_table_start(db_config):
         )
         if len(corresponding_project) == 1:
             if corresponding_project[0]["status"] == str(Status.SUBMITTED):
-                update_values = {"status_all": StatusAll.SUBMITTED_PROJECT}
+                update_values = {
+                    "status_all": StatusAll.SUBMITTED_PROJECT,
+                    "center_name": corresponding_project[0]["center_name"],
+                }
                 update_db_where_conditions(
                     db_config,
                     table_name="submission_table",
@@ -203,7 +206,10 @@ def submission_table_update(db_config):
         if len(corresponding_project) == 1 and corresponding_project[0]["status"] == str(
             Status.SUBMITTED
         ):
-            update_values = {"status_all": StatusAll.SUBMITTED_PROJECT}
+            update_values = {
+                "status_all": StatusAll.SUBMITTED_PROJECT,
+                "center_name": corresponding_project[0]["center_name"],
+            }
             update_db_where_conditions(
                 db_config,
                 table_name="submission_table",