gatk-3.8/java/src/org/broadinstitute/sting/alignment/bwa/BWTFiles.java

package org.broadinstitute.sting.alignment.bwa;

import org.broadinstitute.sting.utils.StingException;
import org.broadinstitute.sting.alignment.reference.packing.PackUtils;
import org.broadinstitute.sting.alignment.reference.bwt.BWT;
import org.broadinstitute.sting.alignment.reference.bwt.BWTWriter;
import org.broadinstitute.sting.alignment.reference.bwt.SuffixArray;
import org.broadinstitute.sting.alignment.reference.bwt.SuffixArrayWriter;
import org.broadinstitute.sting.alignment.reference.bwt.ANNWriter;
import org.broadinstitute.sting.alignment.reference.bwt.AMBWriter;

import java.io.File;
import java.io.IOException;

import net.sf.samtools.SAMSequenceDictionary;
import net.sf.samtools.SAMSequenceRecord;

/**
 * Support files for BWT.
 *
 * @author mhanna
 * @version 0.1
 */
public class BWTFiles {
    /**
     * ANN (?) file name.
     */
    public final File annFile;

    /**
     * AMB (?) file name.
     */
    public final File ambFile;

    /**
     * Packed reference sequence file.
     */
    public final File pacFile;

    /**
     * Forward BWT file.
     */
    public final File forwardBWTFile;

    /**
     * Forward suffix array file.
     */
    public final File forwardSAFile;

    /**
     * Reverse BWT file.
     */
    public final File reverseBWTFile;

    /**
     * Reverse suffix array file.
     */
    public final File reverseSAFile;

    /**
     * Where these files autogenerated on the fly?
     */
    private final boolean autogenerated;

    /**
     * Create a new BWA configuration file using the given prefix.
     * @param prefix Prefix to use when creating the configuration.  Must not be null.
     */
    public BWTFiles(String prefix) {
        if(prefix == null)
            throw new StingException("Prefix must not be null.");
        annFile = new File(prefix + ".ann");
        ambFile = new File(prefix + ".amb");
        pacFile = new File(prefix + ".pac");
        forwardBWTFile = new File(prefix + ".bwt");
        forwardSAFile = new File(prefix + ".sa");
        reverseBWTFile = new File(prefix + ".rbwt");
        reverseSAFile = new File(prefix + ".rsa");
        autogenerated = false;
    }

    /**
     * Hand-create a new BWTFiles object, specifying a unique file object for each type.
     * @param annFile ANN (alternate dictionary) file.
     * @param ambFile AMB (holes) files.
     * @param pacFile Packed representation of the forward reference sequence.
     * @param forwardBWTFile BWT representation of the forward reference sequence.
     * @param forwardSAFile SA representation of the forward reference sequence.
     * @param reverseBWTFile BWT representation of the reversed reference sequence.
     * @param reverseSAFile SA representation of the reversed reference sequence.
     */
    private BWTFiles(File annFile,
                     File ambFile,
                     File pacFile,
                     File forwardBWTFile,
                     File forwardSAFile,
                     File reverseBWTFile,
                     File reverseSAFile) {
        this.annFile = annFile;
        this.ambFile = ambFile;
        this.pacFile = pacFile;
        this.forwardBWTFile = forwardBWTFile;
        this.forwardSAFile = forwardSAFile;
        this.reverseBWTFile = reverseBWTFile;
        this.reverseSAFile = reverseSAFile;        
        autogenerated = true;
    }

    /**
     * Close out this files object, in the process deleting any temporary filse
     * that were created.
     */
    public void close() {
        if(autogenerated) {
            boolean success = true;
            success = annFile.delete();
            success &= ambFile.delete();
            success &= pacFile.delete();
            success &= forwardBWTFile.delete();
            success &= forwardSAFile.delete();
            success &= reverseBWTFile.delete();
            success &= reverseSAFile.delete();

            if(!success)
                throw new StingException("Unable to clean up autogenerated representation");
        }
    }

    /**
     * Create a new set of BWT files from the given reference sequence.
     * @param referenceSequence Sequence from which to build metadata.
     * @return A new object representing encoded representations of each sequence.
     */
    public static BWTFiles createFromReferenceSequence(byte[] referenceSequence) {
        File annFile,ambFile,pacFile,bwtFile,saFile,rpacFile,rbwtFile,rsaFile;
        try {
            // Write the ann and amb for this reference sequence.
            annFile = File.createTempFile("bwt","ann");
            ambFile = File.createTempFile("bwt","amb");

            SAMSequenceDictionary dictionary = new SAMSequenceDictionary();
            dictionary.addSequence(new SAMSequenceRecord("autogenerated",referenceSequence.length));

            ANNWriter annWriter = new ANNWriter(annFile);
            annWriter.write(dictionary);
            annWriter.close();

            AMBWriter ambWriter = new AMBWriter(ambFile);
            ambWriter.writeEmpty(dictionary);
            ambWriter.close();

            // Write the encoded files for the forward version of this reference sequence.
            pacFile = File.createTempFile("bwt","pac");
            bwtFile = File.createTempFile("bwt","bwt");
            saFile = File.createTempFile("bwt","sa");

            writeEncodedReferenceSequence(referenceSequence,pacFile,bwtFile,saFile);

            // Write the encoded files for the reverse version of this reference sequence.
            byte[] reverseReferenceSequence = new byte[referenceSequence.length];
            System.arraycopy(referenceSequence,0,reverseReferenceSequence,0,referenceSequence.length);

            rpacFile = File.createTempFile("bwt","rpac");
            rbwtFile = File.createTempFile("bwt","rbwt");
            rsaFile = File.createTempFile("bwt","rsa");

            writeEncodedReferenceSequence(reverseReferenceSequence,rpacFile,rbwtFile,rsaFile);
        }
        catch(IOException ex) {
            throw new StingException("Unable to write autogenerated reference sequence to temporary files");
        }

        // Make sure that, at the very least, all temporary files are deleted on exit.
        annFile.deleteOnExit();
        ambFile.deleteOnExit();
        pacFile.deleteOnExit();
        bwtFile.deleteOnExit();
        saFile.deleteOnExit();
        rpacFile.deleteOnExit();
        rbwtFile.deleteOnExit();
        rsaFile.deleteOnExit();

        return new BWTFiles(annFile,ambFile,pacFile,bwtFile,saFile,rbwtFile,rsaFile);
    }

    /**
     * Write the encoded form of the reference sequence.  In the case of BWA, the encoded reference
     * sequence is the reference itself in PAC format, the BWT, and the suffix array.
     * @param referenceSequence The reference sequence to encode.
     * @param pacFile Target for the PAC-encoded reference.
     * @param bwtFile Target for the BWT representation of the reference.
     * @param suffixArrayFile Target for the suffix array encoding of the reference.
     * @throws java.io.IOException In case of issues writing to the file.
     */
    private static void writeEncodedReferenceSequence(byte[] referenceSequence,
                                               File pacFile,
                                               File bwtFile,
                                               File suffixArrayFile) throws IOException {
        PackUtils.writeReferenceSequence(pacFile,referenceSequence);

        BWT bwt = BWT.createFromReferenceSequence(referenceSequence);
        BWTWriter bwtWriter = new BWTWriter(bwtFile);
        bwtWriter.write(bwt);
        bwtWriter.close();

        SuffixArray suffixArray = SuffixArray.createFromReferenceSequence(referenceSequence);
        SuffixArrayWriter suffixArrayWriter = new SuffixArrayWriter(suffixArrayFile);
        suffixArrayWriter.write(suffixArray);
        suffixArrayWriter.close();
    }    
}