C++ SamFile::setSortedValidation方法代码示例

//.........这里部分代码省略.........
        }
        else
        {
            mySecondFile = ifopen(secondOut, "w", myCompression);
        }
    
        if(myUnpairedFile == NULL)
        {
            std::cerr << "Failed to open " << unpairedOut
                      << " so can't convert bam2FastQ.\n";
            return(-1);
        }
        if(myFirstFile == NULL)
        {
            std::cerr << "Failed to open " << firstOut
                      << " so can't convert bam2FastQ.\n";
            return(-1);
        }
        if(mySecondFile == NULL)
        {
            std::cerr << "Failed to open " << secondOut
                      << " so can't convert bam2FastQ.\n";
            return(-1);
        }
    }

    if((readName) || (strcmp(mySamHeader.getSortOrder(), "queryname") == 0))
    {
        readName = true;
    }
    else
    {
        // defaulting to coordinate sorted.
        samIn.setSortedValidation(SamFile::COORDINATE);
    }

    // Setup the '=' translation if the reference was specified.
    if(!refFile.IsEmpty())
    {
        GenomeSequence* refPtr = new GenomeSequence(refFile);
        samIn.SetReadSequenceTranslation(SamRecord::BASES);
        samIn.SetReference(refPtr);
    }

    SamRecord* recordPtr;
    int16_t samFlag;

    SamStatus::Status returnStatus = SamStatus::SUCCESS;
    while(returnStatus == SamStatus::SUCCESS)
    {
        recordPtr = myPool.getRecord();
        if(recordPtr == NULL)
        {
            // Failed to allocate a new record.
            throw(std::runtime_error("Failed to allocate a new SAM/BAM record"));
        }
        if(!samIn.ReadRecord(mySamHeader, *recordPtr))
        {
            // Failed to read a record.
            returnStatus = samIn.GetStatus();
            continue;
        }

        // Have a record.  Check to see if it is a pair or unpaired read.
        samFlag = recordPtr->getFlag();
        if(SamFlag::isPaired(samFlag))

//.........这里部分代码省略.........
    }

    if(logFile.IsEmpty())
    {
        logFile = outFile + ".log";
    }

    if(myDoRecab)
    {
        int status = myRecab.processRecabParam();
        if(status != 0)
        {
            inputParameters.Status();
            return(status);
        }
    }

    if(params)
    {
        inputParameters.Status();
    }

    Logger::gLogger = new Logger(logFile.c_str(), verboseFlag);

    /* -------------------------------------------------------------------
     * The arguments are processed.  Prepare the input BAM file,
     * instantiate dedup_LowMem, and construct the read group library map
     * ------------------------------------------------------------------*/

    SamFile samIn;

    samIn.OpenForRead(inFile.c_str());
    // If the file isn't sorted it will throw an exception.
    samIn.setSortedValidation(SamFile::COORDINATE);

    SamFileHeader header;
    samIn.ReadHeader(header);

    buildReadGroupLibraryMap(header);

    lastReference = -1;
    lastCoordinate = -1;

    // for keeping some basic statistics
    uint32_t recordCount = 0;
    uint32_t pairedCount = 0;
    uint32_t properPairCount = 0;
    uint32_t unmappedCount = 0;
    uint32_t reverseCount = 0;
    uint32_t qualCheckFailCount = 0;
    uint32_t secondaryCount = 0;
    uint32_t supplementaryCount = 0;
    uint32_t excludedCount = 0;

    // Now we start reading records
    SamRecord* recordPtr;
    SamStatus::Status returnStatus = SamStatus::SUCCESS;
    while(returnStatus == SamStatus::SUCCESS)
    {
        recordPtr = mySamPool.getRecord();
        if(recordPtr == NULL)
        {
            std::cerr << "Failed to allocate enough records\n";
            return(-1);
        }
        if(!samIn.ReadRecord(header, *recordPtr))

void testIndex(BamIndex& bamIndex)
{
    assert(bamIndex.getNumMappedReads(1) == 2);
    assert(bamIndex.getNumUnMappedReads(1) == 0);
    assert(bamIndex.getNumMappedReads(0) == 4);
    assert(bamIndex.getNumUnMappedReads(0) == 1);
    assert(bamIndex.getNumMappedReads(23) == -1);
    assert(bamIndex.getNumUnMappedReads(23) == -1);
    assert(bamIndex.getNumMappedReads(-1) == 0);
    assert(bamIndex.getNumUnMappedReads(-1) == 2);
    assert(bamIndex.getNumMappedReads(-2) == -1);
    assert(bamIndex.getNumUnMappedReads(-2) == -1);
    assert(bamIndex.getNumMappedReads(22) == 0);
    assert(bamIndex.getNumUnMappedReads(22) == 0);

    // Get the chunks for reference id 1.
    Chunk testChunk;
    SortedChunkList chunkList;
    assert(bamIndex.getChunksForRegion(1, -1, -1, chunkList) == true);
    assert(!chunkList.empty());
    testChunk = chunkList.pop();
    assert(chunkList.empty());
    assert(testChunk.chunk_beg == 0x4e7);
    assert(testChunk.chunk_end == 0x599);

    // Get the chunks for reference id 0.
    assert(bamIndex.getChunksForRegion(0, -1, -1, chunkList) == true);
    assert(!chunkList.empty());
    testChunk = chunkList.pop();
    assert(chunkList.empty());
    assert(testChunk.chunk_beg == 0x360);
    assert(testChunk.chunk_end == 0x4e7);


    // Get the chunks for reference id 2.
    assert(bamIndex.getChunksForRegion(2, -1, -1, chunkList) == true);
    assert(!chunkList.empty());
    testChunk = chunkList.pop();
    assert(chunkList.empty());
    assert(testChunk.chunk_beg == 0x599);
    assert(testChunk.chunk_end == 0x5ea);

    // Get the chunks for reference id 3.
    // There isn't one for this ref id, but still successfully read the file,
    // so it should return true, but the list should be empty.
    assert(bamIndex.getChunksForRegion(3, -1, -1, chunkList) == true);
    assert(chunkList.empty());

    // Test reading an indexed bam file.
    SamFile inFile;
    assert(inFile.OpenForRead("testFiles/sortedBam.bam"));
    inFile.setSortedValidation(SamFile::COORDINATE);
    assert(inFile.ReadBamIndex("testFiles/sortedBam.bam.bai"));
    SamFileHeader samHeader;
    assert(inFile.ReadHeader(samHeader));
    SamRecord samRecord;

    // Test getting num mapped/unmapped reads.
    assert(inFile.getNumMappedReadsFromIndex(1) == 2);
    assert(inFile.getNumUnMappedReadsFromIndex(1) == 0);
    assert(inFile.getNumMappedReadsFromIndex(0) == 4);
    assert(inFile.getNumUnMappedReadsFromIndex(0) == 1);
    assert(inFile.getNumMappedReadsFromIndex(23) == -1);
    assert(inFile.getNumUnMappedReadsFromIndex(23) == -1);
    assert(inFile.getNumMappedReadsFromIndex(-1) == 0);
    assert(inFile.getNumUnMappedReadsFromIndex(-1) == 2);
    assert(inFile.getNumMappedReadsFromIndex(-2) == -1);
    assert(inFile.getNumUnMappedReadsFromIndex(-2) == -1);
    assert(inFile.getNumMappedReadsFromIndex(22) == 0);
    assert(inFile.getNumUnMappedReadsFromIndex(22) == 0);

    assert(inFile.getNumMappedReadsFromIndex("2", samHeader) == 2);
    assert(inFile.getNumUnMappedReadsFromIndex("2", samHeader) == 0);
    assert(inFile.getNumMappedReadsFromIndex("1", samHeader) == 4);
    assert(inFile.getNumUnMappedReadsFromIndex("1", samHeader) == 1);
    assert(inFile.getNumMappedReadsFromIndex("22", samHeader) == 0);
    assert(inFile.getNumUnMappedReadsFromIndex("22", samHeader) == 0);
    assert(inFile.getNumMappedReadsFromIndex("", samHeader) == 0);
    assert(inFile.getNumUnMappedReadsFromIndex("*", samHeader) == 2);
    assert(inFile.getNumMappedReadsFromIndex("unknown", samHeader) == -1);
    assert(inFile.getNumUnMappedReadsFromIndex("unknown", samHeader) == -1);
    assert(inFile.getNumMappedReadsFromIndex("X", samHeader) == 0);
    assert(inFile.getNumUnMappedReadsFromIndex("X", samHeader) == 0);

    // Section -1 = Ref *: 2 records (8 & 10 from testSam.sam that is reflected
    // in the validation.
    assert(inFile.SetReadSection(-1));
    assert(inFile.ReadRecord(samHeader, samRecord));
    validateRead8(samRecord);
    assert(inFile.ReadRecord(samHeader, samRecord));
    validateRead10(samRecord);
    assert(inFile.ReadRecord(samHeader, samRecord) == false);

    // Section 2 = Ref 3: 1 records (9 from testSam.sam that is reflected
    // in the validation.
    assert(inFile.SetReadSection(2));
    assert(inFile.ReadRecord(samHeader, samRecord));
    validateRead9(samRecord);
    assert(inFile.ReadRecord(samHeader, samRecord) == false);

//.........这里部分代码省略.........