Fixes for shattering, added JUnit test case

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@332 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/gatk/dataSources/shards/ShardStrategy.java

@@ -1,8 +1,12 @@
 package org.broadinstitute.sting.gatk.dataSources.shards;
 
+
 import net.sf.samtools.SAMSequenceDictionary;
+import org.apache.log4j.Logger;
+import org.broadinstitute.sting.utils.FastaSequenceFile2;
 import org.broadinstitute.sting.utils.GenomeLoc;
 
+import java.io.File;
 import java.util.Iterator;
 /**
  *
@@ -47,6 +51,9 @@ public abstract class ShardStrategy implements Iterator<Shard>, Iterable<Shard>
     // do we have another contig?
     private boolean nextContig = false;
 
+    /** our log, which we want to capture anything from this class */
+    private static Logger logger = Logger.getLogger(ShardStrategy.class);
+
 
     /**
      * the constructor, taking a seq dictionary to parse out contigs
@@ -110,7 +117,7 @@ public abstract class ShardStrategy implements Iterator<Shard>, Iterable<Shard>
     public Shard next() {
         // lets get some background info on the problem
         long length = dic.getSequence(seqLoc).getSequenceLength();
-        long proposedSize = nextShardSize();
+        long proposedSize = nextShardSize() - 1;
         long nextStart = mLoc.getStop() + 1;
         // can we fit it into the current seq size?
         if (nextStart + proposedSize < length) {
@@ -120,12 +127,14 @@ public abstract class ShardStrategy implements Iterator<Shard>, Iterable<Shard>
         }
         // else we can't make it in the current location, we have to stitch one together
         else {
-            lastGenomeLocSize = nextStart + proposedSize - length;
+            long overflow = nextStart + proposedSize - length;
-
+            logger.debug("Overflow = " + overflow + " length: " + length);
-
+            lastGenomeLocSize = lastGenomeLocSize - overflow;
             // move to the next contig
+            // the next sequence should start at the begining of the next contig
+            Shard ret = Shard.toShard(new GenomeLoc(dic.getSequence(seqLoc).getSequenceName(), nextStart, nextStart + lastGenomeLocSize));
             jumpContig();
-            return Shard.toShard(new GenomeLoc(dic.getSequence(seqLoc).getSequenceName(), nextStart, lastGenomeLocSize));
+            return ret;
         }
 
     }
@@ -133,14 +142,16 @@ public abstract class ShardStrategy implements Iterator<Shard>, Iterable<Shard>
     /** jump to the next contig */
     private void jumpContig() {
         ++seqLoc;
-        if (dic.getSequences().size() <= seqLoc) {
+
+        if (!(seqLoc < dic.getSequences().size())) {
             nextContig = false;
             return;
         }
-
+        logger.debug("Next contig, name = " + dic.getSequence(seqLoc).getSequenceName());
-        // the next sequence should start at the begining of the next contig
         mLoc = new GenomeLoc(dic.getSequence(seqLoc).getSequenceName(), 0, 0);
 
+
+
     }
 
     /**
@@ -168,4 +179,8 @@ public abstract class ShardStrategy implements Iterator<Shard>, Iterable<Shard>
     }
 
 
+    public static void main (String[] strs) {
+        
+    }
+
 }

java/src/org/broadinstitute/sting/gatk/dataSources/shards/ShardStrategyFactory.java

@@ -1,6 +1,7 @@
 package org.broadinstitute.sting.gatk.dataSources.shards;
 
 import net.sf.samtools.SAMSequenceDictionary;
+import org.apache.log4j.Logger;
 
 /**
  *
@@ -34,6 +35,10 @@ public class ShardStrategyFactory {
         LINEAR, EXPONENTIAL
     }
 
+    /** our log, which we want to capture anything from this class */
+    private static Logger logger = Logger.getLogger(ShardStrategyFactory.class);
+
+
     /**
      * get a new shatter strategy
      *

java/test/org/broadinstitute/sting/gatk/dataSources/shards/ShardStrategyFactoryTest.java

@@ -0,0 +1,101 @@
+package org.broadinstitute.sting.gatk.dataSources.shards;
+
+import static junit.framework.Assert.fail;
+import org.broadinstitute.sting.utils.FastaSequenceFile2;
+import org.broadinstitute.sting.utils.GenomeLoc;
+import org.junit.*;
+
+import java.io.File;
+
+/**
+ *
+ * User: aaron
+ * Date: Apr 8, 2009
+ * Time: 11:31:04 AM
+ *
+ * The Broad Institute
+ * SOFTWARE COPYRIGHT NOTICE AGREEMENT 
+ * This software and its documentation are copyright 2009 by the
+ * Broad Institute/Massachusetts Institute of Technology. All rights are reserved.
+ *
+ * This software is supplied without any warranty or guaranteed support whatsoever. Neither
+ * the Broad Institute nor MIT can be responsible for its use, misuse, or functionality.
+ *
+ */
+
+
+/**
+ * @author aaron
+ * @version 1.0
+ * @date Apr 8, 2009
+ * <p/>
+ * Class ShardFactoryTest
+ * <p/>
+ * A descriptions should go here. Blame aaron if it's missing.
+ */
+public class ShardStrategyFactoryTest {
+
+    FastaSequenceFile2 seq = null;
+
+    /**
+     * This function (because of the @BeforeClass tag) gets called only once ever,
+     * before any tests are run
+     */
+    @BeforeClass
+    public static void doBeforeAnyTests() {
+
+    }
+
+    /**
+     * Tears down the test fixture after each call.
+     * <p/>
+     * Called after every test case method.
+     */
+    @AfterClass
+    public static void doAfterAllTests() {
+
+    }
+
+    /**
+     * This function does the setup of our parser, before each method call.
+     * <p/>
+     * Called before every test case method.
+     */
+    @Before
+    public void doForEachTest() {
+        seq = new FastaSequenceFile2(new File("/seq/references/Homo_sapiens_assembly18/v0/Homo_sapiens_assembly18.fasta"));
+    }
+
+    /**
+     * Tears down the test fixture after each call.
+     * <p/>
+     * Called after every test case method.
+     */
+    @After
+    public void undoForEachTest() {
+
+    }
+
+    /** Tests that we got a string parameter in correctly */
+    @Test
+    public void testFullGenomeCycle() {
+        ShardStrategy strategy = ShardStrategyFactory.shatter(ShardStrategyFactory.SHATTER_STRATEGY.LINEAR, seq.getSequenceDictionary(), 100000);
+        int shardCount = 0;
+        try {
+
+            for (Shard s : strategy) {
+                GenomeLoc l = s.getGenomeLoc();
+                //logger.debug("Shard start: " + l.getStart() + " stop " + l.getStop() + " contig " + l.getContig());
+                shardCount++;
+            }
+
+            // check to make sure we got apple shards
+            //logger.debug("shardCount : " + shardCount + " seq size = " + seq.getSequenceDictionary().size());
+
+        } catch (Exception e) {
+            e.printStackTrace();
+            fail("We Shouldn't of seen an exception! : " + e.getMessage() + "; shard count " +  shardCount);
+        }
+    }
+
+}