Checking if the interval files exist before using them to calculate the minimum scatter parts.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@5143 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/utils/interval/IntervalUtils.java

@@ -157,13 +157,26 @@ public class IntervalUtils {
      * @return true if the token looks like a filename, or false otherwise.
      */
     public static boolean isIntervalFile(String str) {
+        return isIntervalFile(str, true);
+    }
+
+    /**
+     * Check if string argument was intented as a file
+     * Accepted file extensions: .bed .list, .picard, .interval_list, .intervals.
+     * @param str token to identify as a filename.
+     * @param checkExists if true throws an exception if the file doesn't exist.
+     * @return true if the token looks like a filename, or false otherwise.
+     */
+    public static boolean isIntervalFile(String str, boolean checkExists) {
         // should we define list of file extensions as a public array somewhere?
         // is regex or endsiwth better?
         File file = new File(str);
         if (str.toUpperCase().endsWith(".BED") || str.toUpperCase().endsWith(".LIST") ||
                 str.toUpperCase().endsWith(".PICARD") || str.toUpperCase().endsWith(".INTERVAL_LIST")
                 || str.toUpperCase().endsWith(".INTERVALS")) {
-            if (file.exists())
+            if (!checkExists)
+                return true;
+            else if (file.exists())
                 return true;
             else
                 throw new UserException.CouldNotReadInputFile(file, "The interval file does not exist.");

java/test/org/broadinstitute/sting/utils/interval/IntervalUtilsUnitTest.java

@@ -128,6 +128,22 @@ public class IntervalUtilsUnitTest extends BaseTest {
         Assert.assertEquals(IntervalUtils.countIntervalArguments(reference, Arrays.asList("chr1:1-2", "chr1:4-5", "chr2:1-1", "chr3:2-2"), true), 3);
     }
 
+    @Test
+    public void testIsIntervalFile() {
+        Assert.assertTrue(IntervalUtils.isIntervalFile(BaseTest.validationDataLocation + "empty_intervals.list"));
+        Assert.assertTrue(IntervalUtils.isIntervalFile(BaseTest.validationDataLocation + "empty_intervals.list", true));
+
+        List<String> extensions = Arrays.asList("bed", "interval_list", "intervals", "list", "picard");
+        for (String extension: extensions) {
+            Assert.assertTrue(IntervalUtils.isIntervalFile("test_intervals." + extension, false), "Tested interval file extension: " + extension);
+        }
+    }
+
+    @Test(expectedExceptions = UserException.CouldNotReadInputFile.class)
+    public void testMissingIntervalFile() {
+        IntervalUtils.isIntervalFile(BaseTest.validationDataLocation + "no_such_intervals.list");
+    }
+
     @Test
     public void testBasicScatter() {
         GenomeLoc chr1 = genomeLocParser.parseGenomeInterval("chr1");

scala/src/org/broadinstitute/sting/queue/extensions/gatk/IntervalScatterFunction.scala

@@ -67,8 +67,18 @@ class IntervalScatterFunction extends ScatterFunction with InProcessFunction {
       this.includeUnmapped = gatk.intervalsString.exists(interval => IntervalUtils.isUnmapped(interval))
     }
 
+    this.scatterCount = originalFunction.scatterCount
+    if (this.intervalFilesExist) {
       val maxScatterCount = IntervalUtils.countIntervalArguments(this.referenceSequence, this.intervals, this.splitByContig)
-    this.scatterCount = maxScatterCount min originalFunction.scatterCount
+      this.scatterCount = this.scatterCount min maxScatterCount
+    }
+  }
+
+  /**
+   * Returns true if all interval files exist.
+   */
+  private def intervalFilesExist = {
+    !intervals.exists(interval => IntervalUtils.isIntervalFile(interval, false) && !new File(interval).exists)
   }
 
   def initCloneInputs(cloneFunction: CloneFunction, index: Int) = {