IGVF-3614-encode-measurement

src/igvfd/schemas/changelogs/measurement_set.md

@@ -2,6 +2,7 @@
 
 ### Minor changes since schema version 49
 
+* Update `dbxrefs` regex to add `ENCODE` as a namespace.
 * Extend `preferred_assay_titles` enum list to include `Parse TAP-seq`.
 
 ### Schema version 49

src/igvfd/schemas/changelogs/sequence_file.md

@@ -2,6 +2,7 @@
 
 ### Minor changes since schema version 16
 
+* Update `dbxrefs` regex to add `ENCODE` as a namespace.
 * Extend `collections` enum list to include `Cross-Disorder CNVs`.
 * Extend `collections` enum list to include `Coronary Artery Disease`.
 * Extend `collections` enum list to include `PD single cell multiomics`.

src/igvfd/schemas/measurement_set.json

@@ -432,6 +432,19 @@
                 "appscript": "10X Chromium Single Cell 5 prime v3",
                 "igvf_utils": "10X Chromium Single Cell 5 prime v3"
             }
+        },
+        "dbxrefs": {
+            "comment": "Supported external resources include: GEO, MaveDB, ENCODE.",
+            "items": {
+                "title": "External identifier",
+                "description": "Identifier from an external resource that may have 1-to-1 or 1-to-many relationships with IGVF file sets.",
+                "type": "string",
+                "pattern": "^(GEO:GSE\\d+|urn:mavedb:\\d{8}-[0a-z]+-\\d+|ENCODE:ENCSR\\d{3}[A-Z]{3})$"
+            },
+            "submissionExample": {
+                "appscript": "[\"GEO:GSE5169185\", \"urn:mavedb:00001250-a-1\", \"ENCODE:ENCSR000AAA\"]",
+                "igvf_utils": "GEO:GSE5169185, urn:mavedb:00001250-a-1, ENCODE:ENCSR000AAA"
+            }
         }
     },
     "fuzzy_searchable_fields": [

src/igvfd/schemas/sequence_file.json

@@ -176,17 +176,17 @@
             }
         },
         "dbxrefs": {
-            "comment": "Supported external resources include: Sequence Read Archive (SRA).",
+            "comment": "Supported external resources include: Sequence Read Archive (SRA), ENCODE.",
             "items": {
                 "title": "External identifier",
                 "description": "Identifier from an external resource that may have 1-to-1 or 1-to-many relationships with IGVF file objects.",
                 "type": "string",
                 "minItems": 1,
-                "pattern": "^(SRA:(SRR|SRX)\\d+)$"
+                "pattern": "^(SRA:(SRR|SRX)\\d+|ENCODE:ENCFF\\d{3}[A-Z]{3})$"
             },
             "submissionExample": {
-                "appscript": "[\"SRA:SRR1234\", \"SRA:SRX1234\"]",
-                "igvf_utils": "SRA:SRR1234, SRA:SRX1234"
+                "appscript": "[\"SRA:SRR1234\", \"SRA:SRX1234\", \"ENCODE:ENCFF000AAA\"]",
+                "igvf_utils": "SRA:SRR1234, SRA:SRX1234, ENCODE:ENCFF000AAA"
             }
         },
         "flowcell_id": {

src/igvfd/tests/data/inserts/measurement_set.json

@@ -24,7 +24,8 @@
             "https://www.protocols.io/private/test-protocols-url-12345"
         ],
         "dbxrefs": [
-            "GEO:GSE187549"
+            "GEO:GSE187549",
+            "ENCODE:ENCSR000AAA"
         ],
         "auxiliary_sets": [
             "IGVFDS0001AUXI"

src/igvfd/tests/data/inserts/sequence_file.json

@@ -48,7 +48,8 @@
         "file_size": 23940124,
         "upload_status": "validated",
         "dbxrefs": [
-            "SRA:SRR0000015"
+            "SRA:SRR0000015",
+            "ENCODE:ENCFF000AAA"
         ],
         "file_set": "igvf:basic_analysis_set",
         "sequencing_run": 12,

src/igvfd/tests/test_schema_measurement_set.py

@@ -18,3 +18,23 @@ def test_measurement_set_protocols_regex(measurement_set, testapp):
         measurement_set['@id'],
         {'protocols': ['https://www.protocols.io/view/123/ABC']})
     assert res.status_code == 200
+
+
+def test_measurement_set_dbxrefs_regex(measurement_set, testapp):
+    res = testapp.patch_json(
+        measurement_set['@id'],
+        {'dbxrefs': ['not_a_valid_dbxref']},
+        expect_errors=True
+    )
+    assert res.status_code == 422
+    res = testapp.patch_json(
+        measurement_set['@id'],
+        {'dbxrefs': ['ENCODE:ENCFF000AAA']},
+        expect_errors=True
+    )
+    assert res.status_code == 422
+    res = testapp.patch_json(
+        measurement_set['@id'],
+        {'dbxrefs': ['GEO:GSE187549', 'urn:mavedb:00001250-a-1', 'ENCODE:ENCSR000AAA']}
+    )
+    assert res.status_code == 200

src/igvfd/tests/test_schema_sequence_file.py

@@ -85,6 +85,19 @@ def test_sequence_file_dbxrefs_regex(testapp, sequence_file):
     )
     assert res.status_code == 200
 
+    res = testapp.patch_json(
+        sequence_file['@id'],
+        {'dbxrefs': ['ENCODE:ENCSR000AAA']},
+        expect_errors=True
+    )
+    assert res.status_code == 422
+
+    res = testapp.patch_json(
+        sequence_file['@id'],
+        {'dbxrefs': ['ENCODE:ENCFF000AAA']}
+    )
+    assert res.status_code == 200
+
 
 def test_sequence_file_sequencing_run_uniqueness(
     testapp,