FIX: Reintroduce sparkPar parameter

* The sparkPar parameter was commented out
* Led to non-deterministic test results due to machine-dependent partitioning
* Updated test case to print normalised importances

Author: NickEdwards7502

src/main/scala/au/csiro/variantspark/api/VSContext.scala

@@ -38,10 +38,10 @@ class VSContext(val spark: SparkSession) extends SqlContextHolder {
     * @param inputFile path to file or directory with VCF files to load
     * @return FeatureSource loaded from the VCF file
     */
-  def importVCF(inputFile: String, imputationStrategy: String = "none"): FeatureSource = {
+  def importVCF(inputFile: String, imputationStrategy: String = "none",
+      sparkPar: Int = 0): FeatureSource = {
     val vcfSource =
-      VCFSource(sc, inputFile)
-    //  VCFSource(sc.textFile(inputFile, if (sparkPar > 0) sparkPar else sc.defaultParallelism))
+      VCFSource(sc.textFile(inputFile, if (sparkPar > 0) sparkPar else sc.defaultParallelism))
     VCFFeatureSource(vcfSource, imputationStrategy)
   }
 

src/test/scala/au/csiro/variantspark/api/ImportanceApiTest.scala

@@ -19,7 +19,8 @@ class ImportanceApiTest extends SparkTest {
     val params = RandomForestParams(seed = 17L)
     val rfModel = RFModelTrainer.trainModel(features, label, params, 200, 50)
     val impAnalysis = new ImportanceAnalysis(sqlContext, features, rfModel)
-    val top10Variables = impAnalysis.importantVariables(10)
+    val top10Variables = impAnalysis.importantVariables(10, normalized = true)
+    top10Variables.foreach(println _)
     assertEquals(10, top10Variables.size)
     assertEquals("22_16050678_C_T", top10Variables.head._1)
   }