Improvements to make this work with uncompressed fastq files. Pulled the fastq parser out into it's own SAMFileReader-like entity.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1520 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/playground/tools/FastqToBam.java

@@ -6,106 +6,30 @@ import net.sf.picard.cmdline.Option;
 import net.sf.samtools.*;
 
 import java.io.*;
-import java.util.zip.GZIPInputStream;
-import java.util.Iterator;
 
-import org.broadinstitute.sting.utils.StingException;
+import org.broadinstitute.sting.utils.fastq.FastqReader;
-
+import org.broadinstitute.sting.utils.fastq.FastqRecord;
-class FastqRecord {
-    private String seqHeader;
-    private String seqLine;
-    private String qualHeader;
-    private String qualLine;
-
-    private String accessionName;
-    private String readName;
-    private String runName;
-
-    public FastqRecord(BufferedReader in) {
-        try {
-            if (in.ready()) {
-                seqHeader = in.readLine();
-                seqLine = in.readLine();
-                qualHeader = in.readLine();
-                qualLine = in.readLine();
-
-                String[] seqHeaderPieces = seqHeader.split("\\s+");
-                accessionName = seqHeaderPieces[0];
-                readName = seqHeaderPieces[1];
-
-                String[] readNamePieces = readName.split(":");
-                runName = readNamePieces[0];
-            }
-        } catch (IOException e) {
-            throw new StingException("Could not read from fastq file.");
-        }
-    }
-
-    public String getReadName() { return readName; }
-    public String getRunName() { return runName; }
-    public String getReadString() { return seqLine; }
-    public String getBaseQualityString() { return qualLine; }
-
-    public String toString() {
-        return String.format("%s %s : %s : %s", accessionName, readName, seqLine, qualLine);
-    }
-}
-
-class FastqReader implements Iterator<FastqRecord>, Iterable<FastqRecord> {
-    private BufferedReader in;
-    private FastqRecord nextRecord;
-    private String runName;
-
-    public FastqReader(File file) {
-        try {
-            in = new BufferedReader(new InputStreamReader(new GZIPInputStream(new FileInputStream(file))));
-
-            nextRecord = new FastqRecord(in);
-            runName = nextRecord.getRunName();
-        } catch (IOException e) {
-            throw new StingException("IO problem");
-        }
-    }
-
-    public String getRunName() { return runName; }
-
-    public boolean hasNext() { return nextRecord != null; }
-
-    public FastqRecord next() {
-        FastqRecord rec = nextRecord;
-
-        try {
-            if (in.ready()) {
-                nextRecord = new FastqRecord(in);
-            } else {
-                nextRecord = null;
-            }
-        } catch (IOException e) {
-            throw new StingException("IO problem");
-        }
-        
-        return rec;
-    }
-
-    public void remove() { throw new UnsupportedOperationException("Unsupported operation"); }
-
-    public Iterator<FastqRecord> iterator() { return this; }
-}
 
 public class FastqToBam extends CommandLineProgram {
     @Usage(programVersion="1.0") public String USAGE = "Extracts read sequences and qualities from the input fastq file and writes them into the output file in unaligned BAM format.";
 
     @Option(shortName="I1", doc="Input file (fastq.gz) to extract reads from (single-end fastq or, if paired, first end of the pair fastq).", optional=false) public File IN1 = null;
     @Option(shortName="I2", doc="Input file (fastq.gz) to extract reads from (if paired, second end of the pair fastq).", optional=true) public File IN2 = null;
-    @Option(shortName="O", doc="Output file (bam).", optional=false) public File OUT = null;
+    @Option(shortName="O",  doc="Output file (bam).", optional=false) public File OUT = null;
+    @Option(shortName="RB", doc="Run barcode", optional=false) public String RUN_BARCODE;
     @Option(shortName="RG", doc="Read group name", optional=false) public String READ_GROUP_NAME;
     @Option(shortName="SM", doc="Sample name", optional=false) public String SAMPLE_NAME;
-    @Option(shortName="V", doc="Verbose mode", optional=true) public Boolean VERBOSE = false;
+    @Option(shortName="V",  doc="Verbose mode", optional=true) public Boolean VERBOSE = false;
 
     public static void main(final String[] argv) {
         System.exit(new FastqToBam().instanceMain(argv));
     }
 
+    private String getReadName(String fqrHeader) {
+        String[] headerPieces = fqrHeader.split("\\s+");
+        return headerPieces[0];
+    }
+
     protected int doWork() {
         FastqReader end1 = new FastqReader(IN1);
         FastqReader end2 = (IN2 == null) ? null : new FastqReader(IN2);
@@ -124,12 +48,14 @@ public class FastqToBam extends CommandLineProgram {
             FastqRecord fqr1 = end1.next();
             FastqRecord fqr2 = (end2 == null) ? null : end2.next();
 
-            if (fqr2 != null && !fqr1.getReadName().equalsIgnoreCase(fqr2.getReadName())) {
+            String fqr1Name = getReadName(fqr1.getReadHeader());
+
+            //if (fqr2 != null && !fqr1Name.equalsIgnoreCase(fqr2Name)) {
                 //throw new StingException(String.format("In paired mode, but end 1 read name (%s) does not match end 2 read name (%s)", fqr1.getReadName(), fqr2.getReadName()));
-            }
+            //}
 
             SAMRecord sr1 = new SAMRecord(sfh);
-            sr1.setReadName(fqr1.getReadName());
+            sr1.setReadName(RUN_BARCODE + ":" + fqr1Name);
             sr1.setReadString(fqr1.getReadString());
             sr1.setBaseQualityString(fqr1.getBaseQualityString());
             sr1.setReadUmappedFlag(true);
@@ -144,17 +70,17 @@ public class FastqToBam extends CommandLineProgram {
                 sr1.setSecondOfPairFlag(false);
                 sr1.setMateUnmappedFlag(true);
 
+                String fqr2Name = getReadName(fqr2.getReadHeader());
                 sr2 = new SAMRecord(sfh);
-
+                sr2.setReadName(RUN_BARCODE + ":" + fqr2Name);
-                sr2.setReadName(fqr2.getReadName());
                 sr2.setReadString(fqr2.getReadString());
                 sr2.setBaseQualityString(fqr2.getBaseQualityString());
                 sr2.setReadUmappedFlag(true);
                 sr2.setReadPairedFlag(true);
+                sr2.setAttribute("RG", READ_GROUP_NAME);
                 sr2.setFirstOfPairFlag(false);
                 sr2.setSecondOfPairFlag(false);
                 sr2.setMateUnmappedFlag(true);
-                sr2.setAttribute("RG", READ_GROUP_NAME);
             }
 
             sfw.addAlignment(sr1);
@@ -164,8 +90,6 @@ public class FastqToBam extends CommandLineProgram {
             readsSeen++;
 
             if (VERBOSE) {
-                //System.out.println(fqr1);
-                //if (fqr2 != null) { System.out.println(fqr2); }
                 System.out.println(sr1.format());
                 if (fqr2 != null) { System.out.println(sr2.format()); }
             }

java/src/org/broadinstitute/sting/utils/fastq/FastqReader.java

@@ -0,0 +1,72 @@
+package org.broadinstitute.sting.utils.fastq;
+
+import org.broadinstitute.sting.utils.StingException;
+
+import java.util.Iterator;
+import java.util.zip.GZIPInputStream;
+import java.io.*;
+
+public class FastqReader implements Iterator<FastqRecord>, Iterable<FastqRecord>, Closeable {
+    private File fastqFile;
+    private BufferedReader in;
+    private FastqRecord nextRecord;
+
+    public FastqReader(File file) {
+        fastqFile = file;
+
+        try {
+            if (fastqFile.getName().endsWith(".gz")) {
+                in = new BufferedReader(new InputStreamReader(new GZIPInputStream(new FileInputStream(fastqFile))));
+            } else {
+                in = new BufferedReader(new InputStreamReader(new FileInputStream(fastqFile)));
+            }
+
+            nextRecord = readNextRecord();
+        } catch (IOException e) {
+            throw new StingException(String.format("Error opening '%s'", fastqFile.getAbsolutePath()));
+        }
+    }
+
+    private FastqRecord readNextRecord() {
+        try {
+            String seqHeader = in.readLine();
+            String seqLine = in.readLine();
+            String qualHeader = in.readLine();
+            String qualLine = in.readLine();
+
+            return new FastqRecord(seqHeader, seqLine, qualHeader, qualLine);
+        } catch (IOException e) {
+            throw new StingException(String.format("Error reading '%s'", fastqFile.getAbsolutePath()));
+        }
+    }
+
+    public boolean hasNext() { return nextRecord != null; }
+
+    public FastqRecord next() {
+        FastqRecord rec = nextRecord;
+
+        try {
+            if (in.ready()) {
+                nextRecord = readNextRecord();
+            } else {
+                nextRecord = null;
+            }
+        } catch (IOException e) {
+            throw new StingException("IO problem");
+        }
+
+        return rec;
+    }
+
+    public void remove() { throw new UnsupportedOperationException("Unsupported operation"); }
+
+    public Iterator<FastqRecord> iterator() { return this; }
+
+    public void close() {
+        try {
+            in.close();
+        } catch (IOException e) {
+            
+        }
+    }
+}

java/src/org/broadinstitute/sting/utils/fastq/FastqRecord.java

@@ -0,0 +1,45 @@
+package org.broadinstitute.sting.utils.fastq;
+
+import org.broadinstitute.sting.utils.StingException;
+
+import java.io.BufferedReader;
+import java.io.IOException;
+
+public class FastqRecord {
+    private String seqHeader;
+    private String seqLine;
+    private String qualHeader;
+    private String qualLine;
+
+    public FastqRecord(String seqHeader, String seqLine, String qualHeader, String qualLine) {
+        setReadHeader(seqHeader);
+        setReadString(seqLine);
+        setBaseQualityHeader(qualHeader);
+        setBaseQualityString(qualLine);
+    }
+
+    public void setReadHeader(String seqHeader) {
+        this.seqHeader = seqHeader.replaceFirst("@", "");
+    }
+
+    public void setReadString(String seqLine) { this.seqLine = seqLine; }
+
+    public void setBaseQualityHeader(String qualHeader) {
+        this.qualHeader = qualHeader.replaceFirst("\\+", "");
+    }
+
+    public void setBaseQualityString(String qualLine) { this.qualLine = qualLine; }
+
+    public String getReadHeader() { return seqHeader; }
+    public String getReadString() { return seqLine; }
+    public String getBaseQualityHeader() { return qualHeader; }
+    public String getBaseQualityString() { return qualLine; }
+
+    public String format() {
+        return String.format("@%s\n%s\n+%s\n%s", seqHeader, seqLine, qualHeader, qualLine);
+    }
+
+    public String toString() {
+        return String.format("%s : %s %s", seqHeader, seqLine, qualLine);
+    }
+}