refactor: all fields can be missing except the variant itself

opentargets · Oct 1, 2018 · 6eb0114 · 6eb0114
1 parent c594000
commit 6eb0114
Show file tree

Hide file tree

Showing 2 changed files with 7 additions and 6 deletions.
diff --git a/src/main/scala/ot/geckopipe/Main.scala b/src/main/scala/ot/geckopipe/Main.scala
@@ -64,13 +64,14 @@ class Commands(val ss: SparkSession, val sampleFactor: Double, val c: Configurat
     val vIdx = vIdxBuilder.load
     val nearests = vIdxBuilder.loadNearestGenes.map( df => {
       logger.info("generate variant index LUT with nearest genes (prot-cod and not prot-cod")
-      vIdx.table.join(df, VariantIndex.variantColumnNames, "left_outer")
-        .write
-        .json(c.output.stripSuffix("/").concat("/variant-index-lut/"))
+      val joint = vIdx.table.join(df, VariantIndex.variantColumnNames, "left_outer")
+
+      joint.show(10, false)
+      joint.write.json(c.output.stripSuffix("/").concat("/variant-index-lut/"))
     })
 
     nearests match {
-      case scala.util.Success(lut) => logger.info("generated variant index LUT")
+      case scala.util.Success(_) => logger.info("generated variant index LUT")
       case scala.util.Failure(ex) => logger.error(ex.getMessage)
     }
   }

diff --git a/src/main/scala/ot/geckopipe/index/VariantIndex.scala b/src/main/scala/ot/geckopipe/index/VariantIndex.scala
@@ -32,9 +32,9 @@ object VariantIndex {
 
   val nearestGenesSchema = StructType(
     StructField("varid", StringType, false) ::
-      StructField("gene_id_prot_coding", StringType, false) ::
+      StructField("gene_id_prot_coding", StringType) ::
       StructField("gene_id_prot_coding_distance", LongType) ::
-      StructField("gene_id", StringType, false) ::
+      StructField("gene_id", StringType) ::
       StructField("gene_id_distance", LongType) :: Nil)
 
   /** this class build based on the Configuration it creates a VariantIndex */