Adding some utilities to test unmapped reads

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@887 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/utils/sam/ArtificialSamUtils.java

@@ -3,8 +3,10 @@ package org.broadinstitute.sting.utils.sam;
 import net.sf.samtools.*;
 
 import java.io.File;
-import java.util.ArrayList;
+import java.util.*;
-import java.util.List;
+
+import org.broadinstitute.sting.gatk.iterators.StingSAMIterator;
+import org.broadinstitute.sting.gatk.Reads;
 
 /**
  *
@@ -33,8 +35,11 @@ import java.util.List;
  * A collection of utilities for making sam and bam files.  Mostly these are for creating sam and bam files for testing purposes.
  */
 public class ArtificialSamUtils {
+    public static final int DEFAULT_READ_LENGTH = 50;
+
     /**
      * create an artificial sam file
+     *
      * @param filename            the filename to write to
      * @param numberOfChromosomes the number of chromosomes
      * @param startingChromosome  where to start counting
@@ -58,6 +63,7 @@ public class ArtificialSamUtils {
 
     /**
      * create an artificial sam file
+     *
      * @param filename            the filename to write to
      * @param numberOfChromosomes the number of chromosomes
      * @param startingChromosome  where to start counting
@@ -81,6 +87,7 @@ public class ArtificialSamUtils {
 
     /**
      * Creates an artificial sam header, matching the parameters, chromosomes which will be labeled chr1, chr2, etc
+     *
      * @param numberOfChromosomes the number of chromosomes to create
      * @param startingChromosome  the starting number for the chromosome (most likely set to 1)
      * @param chromosomeSize      the length of each chromosome
@@ -92,7 +99,7 @@ public class ArtificialSamUtils {
         // make up some sequence records
         for (int x = startingChromosome; x < startingChromosome + numberOfChromosomes; x++) {
             SAMSequenceRecord rec = new SAMSequenceRecord("chr" + (x), chromosomeSize /* size */);
-            rec.setSequenceLength(1000);
+            rec.setSequenceLength(chromosomeSize);
             dict.addSequence(rec);
         }
         header.setSequenceDictionary(dict);
@@ -101,6 +108,7 @@ public class ArtificialSamUtils {
 
     /**
      * Create an artificial read based on the parameters.  The cigar string will be *M, where * is the length of the read
+     *
      * @param header         the SAM header to associate the read with
      * @param name           the name of the read
      * @param refIndex       the reference index, i.e. what chromosome to associate it with
@@ -120,4 +128,98 @@ public class ArtificialSamUtils {
         return record;
     }
 
+
+    /**
+     * create an iterator containing the specified read piles
+     * @param startingChr the chromosome (reference ID) to start from
+     * @param endingChr the id to end with
+     * @param readCount the number of reads per chromosome
+     * @return StingSAMIterator representing the specified amount of fake data
+     */
+    public static StingSAMIterator unmappedReadIterator(int startingChr, int endingChr, int readCount ) {
+        SAMFileHeader header = createArtificialSamHeader((endingChr - startingChr) + 1, startingChr, readCount + DEFAULT_READ_LENGTH);
+        Map<Integer, Integer> map = new HashMap<Integer, Integer>();
+        for (int x = startingChr; x < endingChr; x++) {
+            map.put(x,readCount);
+        }
+        return new ArtificialSAMIterator(startingChr,  endingChr,  readCount,header);
+    }
+
+
+}
+
+/**
+ * this fake iterator allows us to look at how specific piles of reads are handled
+ */
+class ArtificialSAMIterator implements StingSAMIterator {
+
+
+    private int currentChromo = 0;
+    private int currentRead = 0;
+    private int readCount = 0;
+    private boolean done = false;
+    // the next record
+    private SAMRecord next = null;
+    SAMFileHeader header = null;
+
+    // the passed in parameters
+    final int sChr;
+    final int eChr;
+    final int rCount;
+
+    /**
+     * create the fake iterator, given the mapping of chromosomes and read counts
+     * @param startingChr the starting chromosome
+     * @param endingChr the ending chromosome
+     * @param readCount the number of reads in each chromosome
+     * @param header the associated header
+     */
+    ArtificialSAMIterator(int startingChr, int endingChr, int readCount, SAMFileHeader header) {
+        sChr = startingChr;
+        eChr = endingChr;
+        rCount = readCount;
+        this.header = header;
+        this.currentChromo = startingChr;
+    }
+
+    public Reads getSourceInfo() {
+        throw new UnsupportedOperationException("We don't support this");
+    }
+
+    public void close() {
+        // done
+        currentChromo = Integer.MAX_VALUE;
+    }
+
+    public boolean hasNext() {
+        if (currentRead >= rCount) {
+            currentChromo++;
+            currentRead = 0;
+        }
+        if (currentChromo >= eChr) {
+            return false;
+        }
+        int alignment = 0;
+        if (currentChromo >= 0) {
+            alignment = currentRead;
+        } else {
+            alignment = 0;
+        }
+
+        this.next = ArtificialSamUtils.createArtificialRead(this.header,String.valueOf(readCount), currentChromo, alignment, 50);
+        currentRead++;
+        return true;
+    }
+
+    public SAMRecord next() {
+        return next;
+    }
+
+    public void remove() {
+        throw new UnsupportedOperationException("You've tried to remove on a StingSAMIterator (unsupported), not to mention that this is a fake iterator.");
+    }
+
+    public Iterator<SAMRecord> iterator() {
+        return this;
+    }
 }