Merge pull request #3 from Ferlab-Ste-Justine/feat/CQDG_257_alt_id

CQDG:257 alt id
Ferlab-Ste-Justine · Dec 7, 2023 · 70b8606 · 70b8606
2 parents 4bc33e5 + e873385
commit 70b8606
Show file tree

Hide file tree

Showing 4 changed files with 13 additions and 3 deletions.
diff --git a/src/main/resources/test.xlsx b/src/main/resources/test.xlsx
diff --git a/src/main/scala/bio/ferlab/ontology/OntologyTerm.scala b/src/main/scala/bio/ferlab/ontology/OntologyTerm.scala
@@ -4,7 +4,8 @@ case class OntologyTerm(
                          id: String,
                          name: String,
                          parents: Seq[OntologyTerm] = Nil,
-                         is_leaf: Boolean = false
+                         is_leaf: Boolean = false,
+                         alternateIds: Seq[String] = Nil
                        ) {
   override def toString: String = s"$name ($id)"
 }
diff --git a/src/main/scala/bio/ferlab/transform/DownloadTransformer.scala b/src/main/scala/bio/ferlab/transform/DownloadTransformer.scala
@@ -15,6 +15,7 @@ object DownloadTransformer {
   val patternId = "id: ([A-Z]+:[0-9]+)".r
   val patternName = "name: (.*)".r
   val patternIsA = "is_a: ([A-Z]+:[0-9]+) (\\{.*})? ?! (.*)".r
+  val patternAltId = "^alt_id: (HP:[0-9]+|MONDO:[0-9]+)$".r
 
   def using[A](r: BufferedSource)(f: BufferedSource => A): A =
     try {
@@ -45,6 +46,12 @@ object DownloadTransformer {
           val headOntoCopy = headOnto.copy(parents = headOnto.parents :+ OntologyTerm(id, name, Nil))
           headOntoCopy :: current.tail
         }
+        else if (line.matches(patternAltId.regex)) {
+          val patternAltId(altId) = line
+          val headOnto = current.head
+          val headOntoCopy = headOnto.copy(alternateIds = headOnto.alternateIds :+ altId)
+          headOntoCopy :: current.tail
+        }
         else {
           current
         }

diff --git a/src/main/scala/bio/ferlab/transform/WriteJson.scala b/src/main/scala/bio/ferlab/transform/WriteJson.scala
@@ -9,7 +9,8 @@ case class OntologyTermOutput (
                                 name: String,
                                 parents: Seq[String] = Nil,
                                 ancestors: Seq[BasicOntologyTermOutput] = Nil,
-                                is_leaf: Boolean =false
+                                is_leaf: Boolean =false,
+                                alt_ids: Seq[String] = Nil
                               ) {}
 
 case class BasicOntologyTermOutput (
@@ -31,7 +32,8 @@ object WriteJson {
         k.name,
         k.parents.map(_.toString),
         v._1.map(i => BasicOntologyTermOutput(i.id, i.name, i.parents.map(_.toString))).toSeq,
-        v._2
+        v._2,
+        k.alternateIds
       )}.toSeq.toDF().write.mode("overwrite").json(outputDir)
   }