Update gentrap v0.4 schema and parsing: reference section

src/main/resources/schemas/biopet/v0.4/gentrap.json

@@ -118,6 +118,16 @@
                     "type": "object",
                     "minItems": 1,
                     "additionalProperties": { "$ref": "#/definitions/referenceContig" }
+                  },
+
+                  "species": {
+                    "description": "Reference sequence species name",
+                    "type": "string"
+                  },
+
+                  "name": {
+                    "description": "Build name of the reference sequence. For example, hg38 for the human genome.",
+                    "type": "string"
                   }
                 }
               }
@@ -515,7 +525,7 @@
     "referenceContig": {
       "description": "Entry for a single contig in a reference sequence",
       "type": "object",
-      "required": [ "md5", "size" ],
+      "required": [ "md5", "length" ],
 
       "properties": {
 
@@ -524,8 +534,8 @@
           "type": "string"
         },
 
-        "size": {
-          "description": "Size of the contig in base pairs",
+        "length": {
+          "description": "Length of the contig in base pairs",
           "type": "integer"
         }
       }

src/main/scala/nl/lumc/sasc/sentinel/models/ReferenceRecord.scala

@@ -25,14 +25,23 @@ import org.bson.types.ObjectId
  * Representation of an alignment reference sequence.
  *
  * @param refId Database IDs.
- * @param contigMd5s MD5 checksums of all contigs / chromosomes in this reference sequence.
+ * @param contigs Record of all contigs / chromosomes in this reference sequence.
  * @param combinedMd5 MD5 checksum of the concatenated string of all contig MD5 checksums, sorted alphabetically.
  * @param name Reference sequence name.
  * @param creationTimeUtc UTC time when the reference record was created.
  */
 case class ReferenceRecord(
   @Key("_id") refId: ObjectId,
-  contigMd5s: Seq[String],
+  contigs: Seq[ReferenceContigRecord],
   combinedMd5: String,
   name: Option[String] = None,
+  species: Option[String] = None,
   creationTimeUtc: Option[Date] = None)
+
+/**
+ * Representation of a reference sequence contig / chromosome.
+ *
+ * @param md5 MD5 checksum of the sequence.
+ * @param length Length of the sequence.
+ */
+case class ReferenceContigRecord(md5: String, length: Long)

src/main/scala/nl/lumc/sasc/sentinel/processors/gentrap/GentrapV04InputProcessor.scala

@@ -49,15 +49,16 @@ class GentrapV04InputProcessor(protected val mongo: MongodbAccessObject)
 
   /** Extracts a reference record from a Gentrap summary. */
   private[processors] def extractReference(runJson: JValue): ReferenceRecord = {
-    val contigMd5s = (runJson \ "gentrap" \ "settings" \ "reference" \\ "md5")
-      .children
-      .map(_.extract[String])
-      .sorted
-    val combinedMd5 = calcMd5(contigMd5s)
+    val refJson = runJson \ "gentrap" \ "settings" \ "reference"
+    val contigs = (refJson \ "contigs")
+      .extract[Map[String, ReferenceContigRecord]]
+      .values.toSeq
     ReferenceRecord(
       refId = new ObjectId,
-      combinedMd5 = combinedMd5,
-      contigMd5s = contigMd5s,
+      combinedMd5 = calcMd5(contigs.map(_.md5).sorted),
+      contigs = contigs,
+      species = (refJson \ "species").extractOpt[String],
+      name = (refJson \ "name").extractOpt[String],
       creationTimeUtc = Option(getUtcTimeNow))
   }
 

src/test/scala/nl/lumc/sasc/sentinel/JsonLoader.scala

@@ -22,5 +22,6 @@ import org.json4s.jackson.JsonMethods.parse
 import nl.lumc.sasc.sentinel.utils.getResourceStream
 
 trait JsonLoader {
+  /** Given a schema URL, parses the file as JSON and returns its contents as a JValue object. */
   def loadJson(url: String): JValue = parse(getResourceStream(url))
 }

src/test/scala/nl/lumc/sasc/sentinel/processors/gentrap/GentrapValidationSpec.scala

@@ -16,11 +16,10 @@
  */
 package nl.lumc.sasc.sentinel.processors.gentrap
 
-import org.specs2.mutable.Specification
-import org.specs2.mock.Mockito
-
 import nl.lumc.sasc.sentinel.JsonLoader
 import nl.lumc.sasc.sentinel.db.MongodbAccessObject
+import org.specs2.mock.Mockito
+import org.specs2.mutable.Specification
 
 class GentrapValidationSpec extends Specification with JsonLoader with Mockito {