Merge pull request #644 from microbiomedata/issue-639-embargoed

Add embargoed slot, assign to Biosample, and add tests
microbiomedata · Jan 24, 2023 · 333dd1f · 333dd1f
2 parents e1f3c73 + eb62cb0
commit 333dd1f
Show file tree

Hide file tree

Showing 5 changed files with 106 additions and 0 deletions.
diff --git a/jsonschema/nmdc.schema.json b/jsonschema/nmdc.schema.json
@@ -433,6 +433,10 @@
                     "$ref": "#/$defs/QuantityValue",
                     "description": "Elevation of the sampling site is its height above a fixed reference point, most commonly the mean sea level. Elevation is mainly used when referring to points on the earth's surface, while altitude is used for points above the surface, such as an aircraft in flight or a spacecraft in orbit."
                 },
+                "embargoed": {
+                    "description": "If true, the data are embargoed and not available for public access.",
+                    "type": "boolean"
+                },
                 "emsl_biosample_identifiers": {
                     "description": "A list of identifiers for the biosample from the EMSL database.  This is used to link the biosample, as modeled by NMDC, to the biosample in the planned EMSL NEXUS database.",
                     "items": {
@@ -880,6 +884,10 @@
                     "$ref": "#/$defs/QuantityValue",
                     "description": "Temperature at which sample was stored, e.g. -80 degree Celsius"
                 },
+                "samp_taxon_id": {
+                    "description": "NCBI taxon id of the sample. May be a single taxon or mixed taxa sample. Use 'synthetic metagenome' for mock community/positive controls, or 'blank sample' for negative controls.",
+                    "type": "string"
+                },
                 "samp_vol_we_dna_ext": {
                     "$ref": "#/$defs/QuantityValue",
                     "description": "Volume (ml) or mass (g) of total collected sample processed for DNA extraction. Note: total sample collected should be entered under the term Sample Size (mixs:0000001)."

diff --git a/src/schema/nmdc.yaml b/src/schema/nmdc.yaml
@@ -255,6 +255,7 @@ classes:
         Samples are associated with checklists, which define the fields used to annotate
         the samples. Samples are always associated with a taxon.
     slots:
+      - embargoed
       - collected_from
       - type
       - img_identifiers
@@ -1007,6 +1008,17 @@ enums:
 
 slots:
 
+  embargoed:
+    description: >-
+      If true, the data are embargoed and not available for public access.
+    range: boolean
+    recommended: true
+    todos:
+      - make this required?
+      - first apply to Biosample
+      - try to apply to all Biosamples in a particular nmdc-server SubmissionMetadata?
+      - applying to a Study may nto be granular enough
+
   collected_from:
     description: The Site from which a Biosample was collected
     domain: Biosample

diff --git a/test/data/invalid_data/minimal_non_boolean_embargoed_biosample_test.json b/test/data/invalid_data/minimal_non_boolean_embargoed_biosample_test.json
@@ -0,0 +1,29 @@
+{
+  "biosample_set": [
+    {
+      "id": "nmdc:bsm-99-dtTMNb",
+      "part_of": [
+        "gold:Gs0110115"
+      ],
+      "env_broad_scale": {
+        "has_raw_value": "ENVO:00002030",
+        "term": {
+          "id": "ENVO:00002030"
+        }
+      },
+      "env_local_scale": {
+        "has_raw_value": "ENVO:00002169",
+        "term": {
+          "id": "ENVO:00002169"
+        }
+      },
+      "env_medium": {
+        "has_raw_value": "ENVO:00005792",
+        "term": {
+          "id": "ENVO:00005792"
+        }
+      },
+      "embargoed": 999
+    }
+  ]
+}
diff --git a/test/data/minimal_valid_biosample_test.json b/test/data/minimal_valid_biosample_test.json
@@ -0,0 +1,28 @@
+{
+  "biosample_set": [
+    {
+      "id": "nmdc:bsm-99-dtTMNb",
+      "part_of": [
+        "gold:Gs0110115"
+      ],
+      "env_broad_scale": {
+        "has_raw_value": "ENVO:00002030",
+        "term": {
+          "id": "ENVO:00002030"
+        }
+      },
+      "env_local_scale": {
+        "has_raw_value": "ENVO:00002169",
+        "term": {
+          "id": "ENVO:00002169"
+        }
+      },
+      "env_medium": {
+        "has_raw_value": "ENVO:00005792",
+        "term": {
+          "id": "ENVO:00005792"
+        }
+      }
+    }
+  ]
+}
diff --git a/test/data/minimal_valid_embargoed_biosample_test.json b/test/data/minimal_valid_embargoed_biosample_test.json
@@ -0,0 +1,29 @@
+{
+  "biosample_set": [
+    {
+      "id": "nmdc:bsm-99-dtTMNb",
+      "part_of": [
+        "gold:Gs0110115"
+      ],
+      "env_broad_scale": {
+        "has_raw_value": "ENVO:00002030",
+        "term": {
+          "id": "ENVO:00002030"
+        }
+      },
+      "env_local_scale": {
+        "has_raw_value": "ENVO:00002169",
+        "term": {
+          "id": "ENVO:00002169"
+        }
+      },
+      "env_medium": {
+        "has_raw_value": "ENVO:00005792",
+        "term": {
+          "id": "ENVO:00005792"
+        }
+      },
+      "embargoed": true
+    }
+  ]
+}