C++ BamMultiReader::GetNextAlignmentCore方法代码示例

bool GetNextAlignment(BamAlignment& al, BamMultiReader& reader, int32_t refID)
{
    bool good = reader.GetNextAlignmentCore(al);
    if(not good or al.RefID != refID)
        return false;
    else if(not al.IsMapped() or al.IsDuplicate() or al.IsFailedQC())
        return GetNextAlignment(al, reader, refID);
    else
        return true;
}

// merges sorted temp BAM files into single sorted output BAM file
bool SortTool::SortToolPrivate::MergeSortedRuns(void) {
  
    // open up multi reader for all of our temp files
    // this might get broken up if we do a multi-pass system later ??
    BamMultiReader multiReader;
    if ( !multiReader.Open(m_tempFilenames) ) {
        cerr << "bamtools sort ERROR: could not open BamMultiReader for merging temp files... Aborting."
             << endl;
        return false;
    }

    // set sort order for merge
    if ( m_settings->IsSortingByName )
        multiReader.SetSortOrder(BamMultiReader::SortedByReadName);
    else
        multiReader.SetSortOrder(BamMultiReader::SortedByPosition);
    
    // open writer for our completely sorted output BAM file
    BamWriter mergedWriter;
    if ( !mergedWriter.Open(m_settings->OutputBamFilename, m_headerText, m_references) ) {
        cerr << "bamtools sort ERROR: could not open " << m_settings->OutputBamFilename
             << " for writing... Aborting." << endl;
        multiReader.Close();
        return false;
    }
    
    // while data available in temp files
    BamAlignment al;
    while ( multiReader.GetNextAlignmentCore(al) )
        mergedWriter.SaveAlignment(al);
  
    // close readers
    multiReader.Close();
    mergedWriter.Close();
    
    // delete all temp files
    vector<string>::const_iterator tempIter = m_tempFilenames.begin();
    vector<string>::const_iterator tempEnd  = m_tempFilenames.end();
    for ( ; tempIter != tempEnd; ++tempIter ) {
        const string& tempFilename = (*tempIter);
        remove(tempFilename.c_str());
    }
  
    return true;
}

bool StatsTool::StatsToolPrivate::Run() {
  
    // opens the BAM files without checking for indexes 
    BamMultiReader reader;
    if ( !reader.Open(settings->InputFiles, false, true) ) {
        cerr << "Could not open input BAM file(s)... quitting." << endl;
        reader.Close();
        return false;
    }
    
    // plow through file, keeping track of stats
    BamAlignment al;
    while ( reader.GetNextAlignmentCore(al) )
        ProcessAlignment(al);
    
    // print stats
    PrintStats();
    
    // clean and exit
    reader.Close();
    return true; 
}

//.........这里部分代码省略.........

    // make sure index data is available
    if ( !reader.HasIndexes() ) {
        cerr << "bamtools random ERROR: could not load index data for all input BAM file(s)... Aborting." << endl;
        reader.Close();
        return false;
    }

    // get BamReader metadata
    const string headerText = reader.GetHeaderText();
    const RefVector references = reader.GetReferenceData();
    if ( references.empty() ) {
        cerr << "bamtools random ERROR: no reference data available... Aborting." << endl;
        reader.Close();
        return false;
    }

    // determine compression mode for BamWriter
    bool writeUncompressed = ( m_settings->OutputFilename == Options::StandardOut() &&
                              !m_settings->IsForceCompression );
    BamWriter::CompressionMode compressionMode = BamWriter::Compressed;
    if ( writeUncompressed ) compressionMode = BamWriter::Uncompressed;

    // open BamWriter
    BamWriter writer;
    writer.SetCompressionMode(compressionMode);
    if ( !writer.Open(m_settings->OutputFilename, headerText, references) ) {
        cerr << "bamtools random ERROR: could not open " << m_settings->OutputFilename
             << " for writing... Aborting." << endl;
        reader.Close();
        return false;
    }

    // if user specified a REGION constraint, attempt to parse REGION string
    BamRegion region;
    if ( m_settings->HasRegion && !Utilities::ParseRegionString(m_settings->Region, reader, region) ) {
        cerr << "bamtools random ERROR: could not parse REGION: " << m_settings->Region << endl;
        cerr << "Check that REGION is in valid format (see documentation) and that the coordinates are valid"
             << endl;
        reader.Close();
        writer.Close();
        return false;
    }

    // seed our random number generator
    srand( time(NULL) );

    // grab random alignments
    BamAlignment al;
    unsigned int i = 0;
    while ( i < m_settings->AlignmentCount ) {

        int randomRefId    = 0;
        int randomPosition = 0;

        // use REGION constraints to select random refId & position
        if ( m_settings->HasRegion ) {

            // select a random refId
            randomRefId = getRandomInt(region.LeftRefID, region.RightRefID);

            // select a random position based on randomRefId
            const int lowerBoundPosition = ( (randomRefId == region.LeftRefID)
                                             ? region.LeftPosition
                                             : 0 );
            const int upperBoundPosition = ( (randomRefId == region.RightRefID)
                                             ? region.RightPosition
                                             : (references.at(randomRefId).RefLength - 1) );
            randomPosition = getRandomInt(lowerBoundPosition, upperBoundPosition);
        }

        // otherwise select from all possible random refId & position
        else {

            // select random refId
            randomRefId = getRandomInt(0, (int)references.size() - 1);

            // select random position based on randomRefId
            const int lowerBoundPosition = 0;
            const int upperBoundPosition = references.at(randomRefId).RefLength - 1;
            randomPosition = getRandomInt(lowerBoundPosition, upperBoundPosition);
        }

        // if jump & read successful, save first alignment that overlaps random refId & position
        if ( reader.Jump(randomRefId, randomPosition) ) {
            while ( reader.GetNextAlignmentCore(al) ) {
                if ( al.RefID == randomRefId && al.Position >= randomPosition ) {
                    writer.SaveAlignment(al);
                    ++i;
                    break;
                }
            }
        }
    }

    // cleanup & exit
    reader.Close();
    writer.Close();
    return true;
}

int CountTool::Run(int argc, char* argv[]) { 

    // parse command line arguments
    Options::Parse(argc, argv, 1);

    // if no '-in' args supplied, default to stdin
    if ( !m_settings->HasInput ) 
        m_settings->InputFiles.push_back(Options::StandardIn());
    
    // open reader without index
    BamMultiReader reader;
    if (!reader.Open(m_settings->InputFiles, false, true)) {
        cerr << "ERROR: Could not open input BAM file(s)... Aborting." << endl;
        return 1;
    }

    // alignment counter
    BamAlignment al;
    int alignmentCount(0);
    
    // if no region specified, count entire file 
    if ( !m_settings->HasRegion ) {
        while ( reader.GetNextAlignmentCore(al) ) 
            ++alignmentCount;
    }
    
    // otherwise attempt to use region as constraint
    else {
        
        // if region string parses OK
        BamRegion region;
        if ( Utilities::ParseRegionString(m_settings->Region, reader, region) ) {

            // attempt to re-open reader with index files
            reader.Close();
            bool openedOK = reader.Open(m_settings->InputFiles, true, true );
            
            // if error
            if ( !openedOK ) {
                cerr << "ERROR: Could not open input BAM file(s)... Aborting." << endl;
                return 1;
            }
            
            // if index data available, we can use SetRegion
            if ( reader.IsIndexLoaded() ) {
              
                // attempt to use SetRegion(), if failed report error
                if ( !reader.SetRegion(region.LeftRefID, region.LeftPosition, region.RightRefID, region.RightPosition) ) {
                    cerr << "ERROR: Region requested, but could not set BamReader region to REGION: " << m_settings->Region << " Aborting." << endl;
                    reader.Close();
                    return 1;
                } 
              
                // everything checks out, just iterate through specified region, counting alignments
                while ( reader.GetNextAlignmentCore(al) )
                    ++alignmentCount;
            } 
            
            // no index data available, we have to iterate through until we
            // find overlapping alignments
            else {
                while( reader.GetNextAlignmentCore(al) ) {
                    if ( (al.RefID >= region.LeftRefID)  && ( (al.Position + al.Length) >= region.LeftPosition ) &&
                          (al.RefID <= region.RightRefID) && ( al.Position <= region.RightPosition) ) 
                    {
                        ++alignmentCount;
                    }
                }
            }
        } 
        
        // error parsing REGION string
        else {
            cerr << "ERROR: Could not parse REGION - " << m_settings->Region << endl;
            cerr << "Be sure REGION is in valid format (see README) and that coordinates are valid for selected references" << endl;
            reader.Close();
            return 1;
        }
    }
    
    // print results 
    cout << alignmentCount << endl;
    
    // clean & exit
    reader.Close();
    return 0;
}

//.........这里部分代码省略.........
				if( autoCreateBamIndex ) {
					cerr << "Warning: Did not locate BAM index (BAI) file" << plural << ", creating bamtools version..." << endl;
					// to avoid bug use new instance of BamMultiReader
					BamMultiReader bamReader2;
					if( !bamReader2.Open(cmdArgs) || !bamReader2.CreateIndexes() ) {
						cerr << "WARNING: Failed to create BAM index file" << plural << "." << endl;
						bamReaderSetRegions = false;
					} else {
						if( cmdArgs.size() == 1 ) {
							cerr << "Successfully created BAM index file: " << BbcUtils::fileName(cmdArgs[0]) << ".bai" << endl;
						} else {
							cerr << "Successfully created BAM index files." << endl;
						}
						// re-locate indexes with first reader - could not seem to locate BTI files created!
						if( !bamReader.LocateIndexes() ) {
							cerr << "WARNING: Failed to locate BAM index file" << plural << " just created!" << endl;
							bamReaderSetRegions = false;
						}
					}
				} else {
					cerr << "Warning: BAM index file" << plural << " not located for targeted BAM access." << endl;
					bamReaderSetRegions = false;
				}
			}
			// cancel region filtering if there are no regions to iterate (unexpected)
			if( !regions->GetNextRegion( trgContig, trgSrtPos, trgEndPos ) ) {
				onlyOnTargetReads = bamReaderSetRegions = false;
			}
			if( bamReaderSetRegions ) {
				bamReader.Jump( trgContig, trgSrtPos-maxReadLen );
			}
		}
		BamAlignment aln;
		while( bamReader.GetNextAlignmentCore(aln) ) {
			// appears to be an undocumented behavior here
			if( aln.RefID < 0 ) continue;
			// skip filtered reads by flag, length or mapping quality
			if( aln.AlignmentFlag & skipFlag ) continue;
			if( aln.MapQuality < minMapQuality ) continue;
			int32_t endPos = aln.GetEndPosition();
			if( minAlignLength > 0 ) {
				if( endPos - aln.Position < minAlignLength ) continue;
			}
			// screen for on-target reads
			if( onlyOnTargetReads ) {
				// find next region overlapping or beyond of current read
				bool moreRegions = true;
				bool setRegion = false;
				while( aln.RefID > trgContig || (aln.RefID == trgContig && aln.Position > trgEndPos) ) {
					if( !regions->GetNextRegion( trgContig, trgSrtPos, trgEndPos ) ) {
						moreRegions = false;
						break;
					}
					setRegion = bamReaderSetRegions;
				}
				if( !moreRegions ) {
					// prevent further on-target checks and exit early if not using sumStats
					onlyOnTargetReads = false;
					if( trackAllReads ) {
						// force tracking of off-target reads
						regions->TrackReadsOnRegion(aln,endPos);
						if( readTracker ) readTracker->Write(aln,endPos);
						continue;
					}
					break;
				}

bool MergeTool::MergeToolPrivate::Run(void) {

    // set to default input if none provided
    if ( !m_settings->HasInputBamFilename )
        m_settings->InputFiles.push_back(Options::StandardIn());

    // opens the BAM files (by default without checking for indexes)
    BamMultiReader reader;
    if ( !reader.Open(m_settings->InputFiles) ) {
        cerr << "bamtools merge ERROR: could not open input BAM file(s)... Aborting." << endl;
        return false;
    }

    // retrieve header & reference dictionary info
    std::string mergedHeader = reader.GetHeaderText();
    RefVector references = reader.GetReferenceData();

    // determine compression mode for BamWriter
    bool writeUncompressed = ( m_settings->OutputFilename == Options::StandardOut() &&
                               !m_settings->IsForceCompression );
    BamWriter::CompressionMode compressionMode = BamWriter::Compressed;
    if ( writeUncompressed ) compressionMode = BamWriter::Uncompressed;

    // open BamWriter
    BamWriter writer;
    writer.SetCompressionMode(compressionMode);
    if ( !writer.Open(m_settings->OutputFilename, mergedHeader, references) ) {
        cerr << "bamtools merge ERROR: could not open "
             << m_settings->OutputFilename << " for writing." << endl;
        reader.Close();
        return false;
    }

    // if no region specified, store entire contents of file(s)
    if ( !m_settings->HasRegion ) {
        BamAlignment al;
        while ( reader.GetNextAlignmentCore(al) )
            writer.SaveAlignment(al);
    }

    // otherwise attempt to use region as constraint
    else {

        // if region string parses OK
        BamRegion region;
        if ( Utilities::ParseRegionString(m_settings->Region, reader, region) ) {

            // attempt to find index files
            reader.LocateIndexes();

            // if index data available for all BAM files, we can use SetRegion
            if ( reader.HasIndexes() ) {

                // attempt to use SetRegion(), if failed report error
                if ( !reader.SetRegion(region.LeftRefID,
                                       region.LeftPosition,
                                       region.RightRefID,
                                       region.RightPosition) )
                {
                    cerr << "bamtools merge ERROR: set region failed. Check that REGION describes a valid range"
                         << endl;
                    reader.Close();
                    return false;
                }

                // everything checks out, just iterate through specified region, storing alignments
                BamAlignment al;
                while ( reader.GetNextAlignmentCore(al) )
                    writer.SaveAlignment(al);
            }

            // no index data available, we have to iterate through until we
            // find overlapping alignments
            else {
                BamAlignment al;
                while ( reader.GetNextAlignmentCore(al) ) {
                    if ( (al.RefID >= region.LeftRefID)  && ( (al.Position + al.Length) >= region.LeftPosition ) &&
                         (al.RefID <= region.RightRefID) && ( al.Position <= region.RightPosition) )
                    {
                        writer.SaveAlignment(al);
                    }
                }
            }
        }

        // error parsing REGION string
        else {
            cerr << "bamtools merge ERROR: could not parse REGION - " << m_settings->Region << endl;
            cerr << "Check that REGION is in valid format (see documentation) and that the coordinates are valid"
                 << endl;
            reader.Close();
            writer.Close();
            return false;
        }
    }

    // clean & exit
    reader.Close();
    writer.Close();
    return true;
//.........这里部分代码省略.........

int FileReader::runInternal()
{
    ogeNameThread("am_FileReader");

    if(!format_specified)
        format = deduceFileFormat();

    if(format == FORMAT_BAM)
    {
        BamMultiReader reader;
        
        if(!reader.Open(filenames)) {
            cerr << "Error opening BAM files." << endl;
            reader.Close();
            return -1;
        }
        
        header = reader.GetHeader();
        references = reader.GetReferenceData();
        open = true;
        
        BamAlignment * al;
        
        while(true)
        {
            if(load_string_data)
                al = reader.GetNextAlignment();
            else
                al = reader.GetNextAlignmentCore();

            if(!al)
                break;
            
            putOutputAlignment(al);
        }
        
        reader.Close();
    } else if(format == FORMAT_SAM) {
        
        vector<SamReader> readers;
        
        SamHeader first_header;

        // before doing any reading, open the files to
        // verify they are the right format, etc.
        for(int i = 0; i < filenames.size(); i++) {
            SamReader reader;
            
            if(!reader.Open(filenames[i])) {
                cerr << "Error opening SAM file: " << filenames[i] << endl;
                return -1;
            }

            if(filenames.size() > 1 && i == 0)
                first_header = header;
            
            // TODO: We can probably find a better way to deal with multiple SAM file headers,
            // but for now we should disallow different headers to avoid issues.
            if(i > 0 && header.ToString() != first_header.ToString())
                cerr << "Warning! SAM input files have different headers." << endl;
            
            reader.Close();
        }

        for(int i = 0; i < filenames.size(); i++) {
            SamReader reader;
            
            if(!reader.Open(filenames[i])) {
                cerr << "Error opening SAM file: " << filenames[i] << endl;
                return -1;
            }
            
            header = reader.GetHeader();
            references = reader.GetReferenceData();
            open = true;
            
            if(filenames.size() > 1 && i == 0)
                first_header = header;

            BamAlignment * al = NULL;
            while(true)
            {
                al = reader.GetNextAlignment();
                
                if(NULL == al)
                    break;
                
                putOutputAlignment(al);
            }

            reader.Close();
        }
    } else {
        cerr << "FileReader couldn't detect file format. Aborting." << endl;
        exit(-1);
        return -1;
    }

    return 0;
}

int MergeTool::Run(int argc, char* argv[]) {

    // parse command line arguments
    Options::Parse(argc, argv, 1);

    // set to default input if none provided
    if ( !m_settings->HasInputBamFilename )
        m_settings->InputFiles.push_back(Options::StandardIn());

    // opens the BAM files (by default without checking for indexes)
    BamMultiReader reader;
    if ( !reader.Open(m_settings->InputFiles, false, true) ) {
        cerr << "ERROR: Could not open input BAM file(s)... Aborting." << endl;
        return 1;
    }

    // retrieve header & reference dictionary info
    std::string mergedHeader = reader.GetHeaderText();
    RefVector references = reader.GetReferenceData();

    // open writer
    BamWriter writer;
    bool writeUncompressed = ( m_settings->OutputFilename == Options::StandardOut() && !m_settings->IsForceCompression );
    if ( !writer.Open(m_settings->OutputFilename, mergedHeader, references, writeUncompressed) ) {
        cerr << "ERROR: Could not open BAM file " << m_settings->OutputFilename << " for writing... Aborting." << endl;
        reader.Close();
        return 1;
    }

    // if no region specified, store entire contents of file(s)
    if ( !m_settings->HasRegion ) {
        BamAlignment al;
        while ( reader.GetNextAlignmentCore(al) )
            writer.SaveAlignment(al);
    }

    // otherwise attempt to use region as constraint
    else {

        // if region string parses OK
        BamRegion region;
        if ( Utilities::ParseRegionString(m_settings->Region, reader, region) ) {

            // attempt to re-open reader with index files
            reader.Close();
            bool openedOK = reader.Open(m_settings->InputFiles, true, true );

            // if error
            if ( !openedOK ) {
                cerr << "ERROR: Could not open input BAM file(s)... Aborting." << endl;
                return 1;
            }

            // if index data available, we can use SetRegion
            if ( reader.IsIndexLoaded() ) {

                // attempt to use SetRegion(), if failed report error
                if ( !reader.SetRegion(region.LeftRefID, region.LeftPosition, region.RightRefID, region.RightPosition) ) {
                    cerr << "ERROR: Region requested, but could not set BamReader region to REGION: " << m_settings->Region << " Aborting." << endl;
                    reader.Close();
                    return 1;
                }

                // everything checks out, just iterate through specified region, storing alignments
                BamAlignment al;
                while ( reader.GetNextAlignmentCore(al) )
                    writer.SaveAlignment(al);
            }

            // no index data available, we have to iterate through until we
            // find overlapping alignments
            else {
                BamAlignment al;
                while ( reader.GetNextAlignmentCore(al) ) {
                    if ( (al.RefID >= region.LeftRefID)  && ( (al.Position + al.Length) >= region.LeftPosition ) &&
                            (al.RefID <= region.RightRefID) && ( al.Position <= region.RightPosition) )
                    {
                        writer.SaveAlignment(al);
                    }
                }
            }
        }

        // error parsing REGION string
        else {
            cerr << "ERROR: Could not parse REGION - " << m_settings->Region << endl;
            cerr << "Be sure REGION is in valid format (see README) and that coordinates are valid for selected references" << endl;
            reader.Close();
            writer.Close();
            return 1;
        }
    }

    // clean & exit
    reader.Close();
    writer.Close();
    return 0;
}