Merge pull request #369 from opentargets/vh-add-tss

feat: add tss column to target index
opentargets · Jan 14, 2025 · b027543 · b027543
2 parents 82093f8 + a7641e2
commit b027543
Showing 1 changed file with 10 additions and 0 deletions.
diff --git a/src/main/scala/io/opentargets/etl/backend/target/Target.scala b/src/main/scala/io/opentargets/etl/backend/target/Target.scala
@@ -166,6 +166,7 @@ object Target extends LazyLogging {
       .transform(addTargetSafety(inputDataFrames, ensemblIdLookupDf))
       .transform(addReactome(reactome))
       .transform(removeDuplicatedSynonyms)
+      .transform(addTss)
 
     val targetEssentialityDF = targetsDF
       .transform(addGeneEssentiality(inputDataFrames("geneEssentiality").data, ensemblIdLookupDf))
@@ -176,6 +177,15 @@ object Target extends LazyLogging {
     )
   }
 
+  private def addTss(dataFrame: DataFrame): DataFrame = {
+    logger.info("Adding tss column to target dataframe")
+    dataFrame.withColumn(
+      "tss",
+      when(col("canonicalTranscript.strand") === 1, col("canonicalTranscript.start"))
+        .when(col("canonicalTranscript.strand") === -1, col("canonicalTranscript.end"))
+    )
+  }
+
   /** for all alternative names or symbols a target can take is worth cleaning them up from
     * duplicated entries
     */