Merge pull request #828 from bids-standard/bep018

Bep018
bids-standard/bids-examples#178

Author: rwblair

bids-validator/bids_validator/rules/fixed_top_level_names.json

@@ -3,6 +3,7 @@
     "/README",
     "/CHANGES",
     "/dataset_description.json",
+    "/genetic_info.json",
     "/participants.tsv",
     "/participants.json",
     "/phasediff.json",

bids-validator/package.json

@@ -31,6 +31,7 @@
     "ignore": "^4.0.2",
     "is-utf8": "^0.2.1",
     "jshint": "^2.9.6",
+    "lerna": "^3.15.0",
     "minimatch": "3.0.4",
     "nifti-js": "^1.0.1",
     "p-limit": "^2.1.0",

bids-validator/tests/json.spec.js

@@ -271,4 +271,39 @@ describe('JSON', function() {
       assert.deepEqual(issues, [])
     })
   })
+
+  var genetic_info_file = {
+    name: 'genetic_info.json',
+    relativePath: '/genetic_info.json',
+  }
+
+  it('sample genetic_info.json should parse', function() {
+    var jsonObj = {
+      GeneticLevel: 'Genetic',
+      AnalyticalApproach: ['SNP Genotypes'],
+      SampleOrigin: 'brain',
+      TissueOrigin: 'gray matter',
+      CellType: 'neuron',
+      BrainLocation: '[-30 -15 10]',
+    }
+    jsonDict[genetic_info_file.relativePath] = jsonObj
+    validate.JSON(genetic_info_file, jsonDict, function(issues) {
+      assert.deepEqual(issues, [])
+    })
+  })
+
+  it('genetic_info.json should use limited vocabulary for sample origin', function() {
+    var jsonObj = {
+      GeneticLevel: 'Genetic',
+      AnalyticalApproach: ['SNP Genotypes'],
+      SampleOrigin: 'not_from_around_here',
+      TissueOrigin: 'gray matter',
+      CellType: 'neuron',
+      BrainLocation: '[-30 -15 10]',
+    }
+    jsonDict[genetic_info_file.relativePath] = jsonObj
+    validate.JSON(genetic_info_file, jsonDict, function(issues) {
+      assert(issues.length === 1 && issues[0].code == 55)
+    })
+  })
 })

bids-validator/utils/issues/list.js

@@ -703,4 +703,10 @@ export default {
     reason:
       'Status column in channels.tsv files must contain only one of two values: good or bad. Per the BIDS spec: (https://bids-specification.readthedocs.io/en/stable/04-modality-specific-files/04-intracranial-electroencephalography.html#channels-description-_channelstsv).'
   },
+  128: {
+    key: 'NO_GENETIC_DATABASE',
+    severity: 'error',
+    reason:
+      'A genetic_info.json file is present but no Database field present in Genetics object in dataset_description.json.',
+  },
 }

bids-validator/validators/bids/__tests__/checkDatasetDescription.spec.js

@@ -58,4 +58,32 @@ describe('checkDatasetDescription', () => {
       )
     })
   })
+  describe('checkGeneticDatabaseField', () => {
+    it('returns code 128 when there is no Genetics.Dataset with a genetic_info.json present', () => {
+       const invalidJsonContentsDict = {
+        '/dataset_description.json': { },
+        '/genetic_info.json': { }
+      }
+      let issues = checkDatasetDescription(invalidJsonContentsDict)
+      assert(
+        issues.findIndex(issue => issue.code === 128) > -1,
+        'issues include a code 128',
+      )
+    })
+    it('does not return code 128 when GeneticDataset field and genetic_info.json present', () => {
+       const validJsonContentsDict = {
+        '/dataset_description.json': {
+          Authors: ['Benny', 'the Jets'],
+          Genetics: {Dataset: 'GeneticGeneticDataset'},
+        },
+        '/genetic_info.json': { }
+      }
+      let issues = checkDatasetDescription(validJsonContentsDict)
+      assert(
+        issues.findIndex(issue => issue.code === 128) === -1,
+        'issues does not include a code 128'
+      )
+    })
+
+  })
 })

bids-validator/validators/bids/checkDatasetDescription.js

@@ -6,6 +6,10 @@ const checkDatasetDescription = jsonContentsDict => {
   const hasDatasetDescription = jsonFilePaths.some(path => {
     return path == '/dataset_description.json'
   })
+  const hasGeneticInfo = jsonFilePaths.some(path => {
+    return path === '/genetic_info.json'
+  })
+
   if (!hasDatasetDescription) {
     issues.push(new Issue({ code: 57 }))
   } else {
@@ -14,6 +18,11 @@ const checkDatasetDescription = jsonContentsDict => {
     // check to ensure that the dataset description Authors are
     // properly formatted
     issues = issues.concat(checkAuthorField(datasetDescription.Authors))
+
+    // if genetic info json present ensure mandatory GeneticDataset present
+    if (hasGeneticInfo && !(('Genetics' in datasetDescription) && ('Dataset' in datasetDescription.Genetics))) {
+      issues.push(new Issue({ code: 128 }));
+    }
   }
   return issues
 }

bids-validator/validators/json/json.js

@@ -107,6 +107,10 @@ const selectSchema = file => {
       file.name.endsWith('coordsystem.json')
     ) {
       schema = require('./schemas/coordsystem_eeg.json')
+    } else if (
+        file.name.endsWith('genetic_info.json')
+    ) {
+      schema = require('./schemas/genetic_info.json')
     }
   }
   return schema

bids-validator/validators/json/schemas/dataset_description.json

@@ -43,6 +43,26 @@
     },
     "DatasetDOI": {
       "type": "string"
+    },
+    "Genetics": {
+      "type": "object",
+      "properties": {
+        "Database": {
+          "items": {
+            "type": "string"
+          },
+          "type": ["array", "string"]
+        },
+        "Dataset": {
+          "type": "string"
+        },
+        "Descriptors": {
+          "items": {
+            "type": "string"
+          },
+          "type": ["array", "string"]
+        }
+      }
     }
   },
   "required": ["Name", "BIDSVersion"],

bids-validator/validators/json/schemas/genetic_info.json

@@ -0,0 +1,50 @@
+{
+  "type": "object",
+  "properties": {
+    "GeneticLevel": {
+      "type": "string",
+      "enum": [
+        "Genetic",
+        "Genomic",
+        "Epigenomic",
+        "Transcriptomic",
+        "Metabolomic",
+        "Proteomic"
+      ]
+    },
+    "AnalyticalApproach": {
+      "items": {
+        "type": "string"
+      },
+      "type": "array"
+    },
+    "SampleOrigin": {
+      "type": "string",
+      "enum": [
+        "blood",
+        "saliva",
+        "brain",
+        "csf",
+        "breast milk",
+        "bile",
+        "amniotic fluid",
+        "other biospecimen"
+      ]
+    },
+    "TissueOrigin": {
+      "type": "string",
+      "enum": [
+        "gray matter",
+        "white matter",
+        "csf",
+        "meninges",
+        "macrovascular",
+        "microvascular"
+      ]
+    },
+    "BrainLocation": { "type": "string" },
+    "CellType": { "type": "string" }
+  },
+  "required": ["GeneticLevel", "SampleOrigin"],
+  "additionalProperties": false
+}