augur/data/schema-annotations.json

@@ -7,10 +7,14 @@
             "type": "object",
             "allOf": [{ "$ref": "#/$defs/startend" }],
             "properties": {
+                "start": {
+                    "enum": [1],
+                    "$comment": "nuc must begin at 1"
+                },
                 "strand": {
                     "type": "string",
                     "not": {"enum":["-"]},
-                    "$comment": "Auspice assumes +ve strand for 'nuc', but will not proceed if you specify '-'"
+                    "$comment": "The strand for 'nuc' is not needed, as it should be +ve for all genomes. Auspice will not proceed if you specify '-'"
                 }
             },
             "additionalProperties": true,
@@ -32,9 +36,9 @@
                     "$comment": "Shown in on-hover infobox & influences default CDS colors"
                 },
                 "strand": {
-                    "description": "Is the CDS on the positive ('+') or negative ('-') strand.",
-                    "$comment": "Auspice assumes positive strand unless strand is '-'",
-                    "type": "string"
+                    "description": "Strand of the CDS",
+                    "type": "string",
+                    "enum": ["-", "+"]
                 },
                 "color": {
                     "type": "string",

tests/test_validate.py

@@ -4,7 +4,10 @@
 from augur.validate import (
     validate_collection_config_fields,
     validate_collection_display_defaults,
-    validate_measurements_config
+    validate_measurements_config,
+    load_json_schema,
+    validate_json,
+    ValidateError
 )
 
 
@@ -88,3 +91,71 @@ def test_validate_measurements_config_invalid_default_collection(self, example_m
         }
         assert not validate_measurements_config(measurements)
         assert capsys.readouterr().err == "ERROR: The default collection key 'invalid_collection' does not match any of the collections' keys.\n"
+
+
+@pytest.fixture
+def genome_annotation_schema():
+    return load_json_schema("schema-annotations.json")
+
+class TestValidateGenomeAnnotations():
+    def test_negative_strand_nuc(self, capsys, genome_annotation_schema):
+        d = {"nuc": {"start": 1, "end": 200, "strand": "-"}}
+        with pytest.raises(ValidateError):
+            validate_json(d, genome_annotation_schema, "<test-json>")
+        capsys.readouterr() # suppress validation error printing
+
+    def test_nuc_not_starting_at_one(self, capsys, genome_annotation_schema):
+        d = {"nuc": {"start": 100, "end": 200, "strand": "+"}}
+        with pytest.raises(ValidateError):
+            validate_json(d, genome_annotation_schema, "<test-json>")
+        capsys.readouterr() # suppress validation error printing
+
+    def test_missing_nuc(self, capsys, genome_annotation_schema):
+        d = {"cds": {"start": 100, "end": 200, "strand": "+"}}
+        with pytest.raises(ValidateError):
+            validate_json(d, genome_annotation_schema, "<test-json>")
+        capsys.readouterr() # suppress validation error printing
+
+    def test_missing_properties(self, capsys, genome_annotation_schema):
+        d = {"nuc": {"start": 1, "end": 100}, "cds": {"start": 20, "strand": "+"}}
+        with pytest.raises(ValidateError):
+            validate_json(d, genome_annotation_schema, "<test-json>")
+        capsys.readouterr() # suppress validation error printing
+
+    def test_not_stranded_cds(self, capsys, genome_annotation_schema):
+        # Strand . is for features that are not stranded (as per GFF spec), and thus they're not CDSs
+        d = {"nuc": {"start": 1, "end": 100}, "cds": {"start": 18, "end": 20, "strand": "."}}
+        with pytest.raises(ValidateError):
+            validate_json(d, genome_annotation_schema, "<test-json>")
+        capsys.readouterr() # suppress validation error printing
+
+    def test_negative_coordinates(self, capsys, genome_annotation_schema):
+        d = {"nuc": {"start": 1, "end": 100}, "cds": {"start": -2, "end": 10, "strand": "+"}}
+        with pytest.raises(ValidateError):
+            validate_json(d, genome_annotation_schema, "<test-json>")
+        capsys.readouterr() # suppress validation error printing
+
+    def test_valid_genome(self, capsys, genome_annotation_schema):
+        d = {"nuc": {"start": 1, "end": 100}, "cds": {"start": 20,  "end": 28, "strand": "+"}}
+        validate_json(d, genome_annotation_schema, "<test-json>")
+        capsys.readouterr() # suppress validation error printing
+
+    def test_valid_segmented_genome(self, capsys, genome_annotation_schema):
+        d = {"nuc": {"start": 1, "end": 100},
+             "cds": {"segments": [{"start": 20,  "end": 28}], "strand": "+"}}
+        validate_json(d, genome_annotation_schema, "<test-json>")
+        capsys.readouterr() # suppress validation error printing
+
+    def test_invalid_segmented_genome(self, capsys, genome_annotation_schema):
+        d = {"nuc": {"start": 1, "end": 100},
+             "cds": {"segments": [{"start": 20,  "end": 28}, {"start": 27}], "strand": "+"}}
+        with pytest.raises(ValidateError):
+            validate_json(d, genome_annotation_schema, "<test-json>")
+        capsys.readouterr() # suppress validation error printing
+
+    def test_string_coordinates(self, capsys, genome_annotation_schema):
+        d = {"nuc": {"start": 1, "end": 100},
+             "cds": {"segments": [{"start": 20,  "end": 28}, {"start": "27", "end": "29"}], "strand": "+"}}
+        with pytest.raises(ValidateError):
+            validate_json(d, genome_annotation_schema, "<test-json>")
+        capsys.readouterr() # suppress validation error printing