From 095dacd154241ecbc3a552b001934fc4a9eecad6 Mon Sep 17 00:00:00 2001
From: kiran <kiran@348d0f76-0448-11de-a6fe-93d51630548a>
Date: Tue, 12 May 2009 19:46:50 +0000
Subject: [PATCH] Experimental refactoring.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@675 348d0f76-0448-11de-a6fe-93d51630548a
---
 .../fourbasecaller/AddFourProbsToSAM.java     | 96 +++++++++++++++++++
 1 file changed, 96 insertions(+)
 create mode 100755 java/src/org/broadinstitute/sting/playground/fourbasecaller/AddFourProbsToSAM.java

diff --git a/java/src/org/broadinstitute/sting/playground/fourbasecaller/AddFourProbsToSAM.java b/java/src/org/broadinstitute/sting/playground/fourbasecaller/AddFourProbsToSAM.java
new file mode 100755
index 000000000..c9b0e47db
--- /dev/null
+++ b/java/src/org/broadinstitute/sting/playground/fourbasecaller/AddFourProbsToSAM.java
@@ -0,0 +1,96 @@
+package org.broadinstitute.sting.playground.fourbasecaller;
+
+import org.broadinstitute.sting.utils.cmdLine.CommandLineProgram;
+import org.broadinstitute.sting.utils.cmdLine.Argument;
+import org.broadinstitute.sting.utils.QualityUtils;
+
+import java.io.File;
+import java.util.HashMap;
+
+import net.sf.samtools.*;
+
+public class AddFourProbsToSAM extends CommandLineProgram {
+    public static AddFourProbsToSAM Instance = null;
+
+    public File UNALIGNED_SAM;
+    public File ALIGNED_SAM;
+    public File FINAL_SAM;
+    public int END;
+    public Boolean DEBUG = false;
+
+    public static void main(String[] argv) {
+        Instance = new AddFourProbsToSAM();
+        start(Instance, argv);
+    }
+
+    protected void setupArgs() {
+        //m_parser.addRequiredArg("unaligned_sam", "U", "Unaligned SAM file", "UNALIGNED_SAM");
+        //m_parser.addRequiredArg("aligned_sam",   "A", "Aligned SAM file",   "ALIGNED_SAM");
+        //m_parser.addRequiredArg("final_sam",     "F", "Final SAM file",     "FINAL_SAM");
+        //m_parser.addRequiredArg("end",           "E", "Pair end (0 - all, 1 - first, 2 - second)", "END");
+        //m_parser.addOptionalFlag("debug",        "D", "Turn on debugging output", "DEBUG");
+    }
+
+    protected int execute() {
+        int processed;
+
+        SAMFileReader alignedSf = new SAMFileReader(ALIGNED_SAM);
+        alignedSf.setValidationStringency(SAMFileReader.ValidationStringency.SILENT);
+
+        // First, hash the aligned records (because there are less of them than unaligned reads)
+        System.err.println("Hashing aligned records...");
+
+        HashMap<String, SAMRecord> records = new HashMap<String, SAMRecord>(10000000);
+        processed = 0;
+        for (SAMRecord alignedSr : alignedSf) {
+            if (END == 0 || (END == 1 && alignedSr.getSecondOfPairFlag() == false) || (END == 2 && alignedSr.getSecondOfPairFlag() == true)) {
+                if (!alignedSr.getReadUnmappedFlag()) {
+                    records.put(alignedSr.getReadName(), alignedSr);
+
+                    if (processed % 100000 == 0) { System.err.print("\tProcessed " + processed + " records.\r"); }
+                    processed++;
+                }
+            }
+        }
+
+        // Now, iterate over the unaligned SAM file and stick the four-base probs in.
+        System.err.println("\nInterating over unaligned records...");
+
+        SAMFileReader unalignedSf = new SAMFileReader(UNALIGNED_SAM);
+        unalignedSf.setValidationStringency(SAMFileReader.ValidationStringency.SILENT);
+
+        SAMFileHeader swhead = alignedSf.getFileHeader();
+        swhead.setSortOrder(SAMFileHeader.SortOrder.unsorted);
+        SAMFileWriter sw = new SAMFileWriterFactory().makeSAMOrBAMWriter(swhead, true, FINAL_SAM);
+
+        processed = 0;
+        for (SAMRecord unalignedSr : unalignedSf) {
+            if (records.containsKey(unalignedSr.getReadName())) {
+                SAMRecord alignedSr = records.get(unalignedSr.getReadName());
+
+                byte[] sq = (byte[]) unalignedSr.getAttribute("SQ");
+                if (alignedSr.getReadNegativeStrandFlag()) {
+                    sq = QualityUtils.reverseComplementCompressedQualityArray(sq);
+                }
+                
+                alignedSr.setAttribute("SQ", sq);
+                alignedSr.setAttribute("KB", unalignedSr.getReadBases());
+                alignedSr.setAttribute("KQ", unalignedSr.getBaseQualities());
+                sw.addAlignment(alignedSr);
+
+                if (DEBUG) {
+                    System.out.println(alignedSr.format());
+                }
+
+                if (processed % 100000 == 0) { System.err.print("\tProcessed " + processed + " records.\r"); }
+                processed++;
+            }
+        }
+
+        sw.close();
+        alignedSf.close();
+        unalignedSf.close();
+
+        return 0;
+    }
+}