Merge pull request #368 from AdamaJava/qprofiler2_longread

Qprofiler2 longread
AdamaJava · Jul 25, 2024 · c7b5eae · c7b5eae
2 parents defe66e + 37e1480
commit c7b5eae
Show file tree

Hide file tree

Showing 18 changed files with 870 additions and 119 deletions.
diff --git a/qprofiler2/src/org/qcmg/qprofiler2/Options.java b/qprofiler2/src/org/qcmg/qprofiler2/Options.java
@@ -33,7 +33,8 @@ final class Options {
 	private static final String OUTPUT_FILE_DESCRIPTION = Messages.getMessage(msgResource,"OUTPUT_FILE_DESCRIPTION");
 	private static final String INPUT_FILE_DESCRIPTION = Messages.getMessage(msgResource,"INPUT_FILE_DESCRIPTION");
 	private static final String INDEX_FILE_DESCRIPTION = Messages.getMessage(msgResource,"INDEX_FILE_DESCRIPTION");
-	private static final String VALIDATION_STRINGENCY_OPTION_DESCRIPTION = Messages.getMessage(msgResource,"VALIDATION_STRINGENCY_DESCRIPTION");	
+	private static final String VALIDATION_STRINGENCY_OPTION_DESCRIPTION = Messages.getMessage(msgResource,"VALIDATION_STRINGENCY_DESCRIPTION");
+	private static final String LONG_READ_OPTION_DESCRIPTION = Messages.getMessage(msgResource, "LONG_READ_OPTION_DESCRIPTION");
 	private static final String FULL_BAMHEADER_OPTION_DESCRIPTION = Messages.getMessage(msgResource, "FULL_BAMHEADER_OPTION_DESCRIPTION");
 
 	// vcf mode
@@ -68,7 +69,8 @@ final class Options {
 		parser.accepts("bam-validation", VALIDATION_STRINGENCY_OPTION_DESCRIPTION).withRequiredArg().ofType(String.class);
 		parser.accepts("bam-full-header", FULL_BAMHEADER_OPTION_DESCRIPTION);
 		parser.accepts("vcf-format-field", FORMAT_OPTION_DESCRIPTION).withRequiredArg().ofType(String.class).withValuesSeparatedBy(',');
-		parser.accepts("bam-index", INDEX_FILE_DESCRIPTION).withRequiredArg().ofType(String.class).withValuesSeparatedBy(',');		
+		parser.accepts("bam-index", INDEX_FILE_DESCRIPTION).withRequiredArg().ofType(String.class).withValuesSeparatedBy(',');
+		parser.accepts("long-read", LONG_READ_OPTION_DESCRIPTION);
 		parser.posixlyCorrect(true);
 		options = parser.parse(args);
 
@@ -120,6 +122,10 @@ boolean hasFullBamHeaderOption() {
 		return options.has("bam-full-header") ; 
 	}
 
+	boolean hasLongReadBamOption() {
+		return options.has("long-read") ;
+	}
+
 	boolean hasVersionOption() {
 		return options.has("version"); 
 	}

diff --git a/qprofiler2/src/org/qcmg/qprofiler2/QProfiler2.java b/qprofiler2/src/org/qcmg/qprofiler2/QProfiler2.java
@@ -60,6 +60,7 @@ public class QProfiler2 {
 	private String logFile;
 	private String validation;
 	private boolean  isFullBamHeader;
+	private boolean  isLongReadBam;
 
 
 	/*
@@ -209,10 +210,13 @@ private Summarizer getSummarizer(ProfileType key) {
 				}
 				break;
 			case BAM:
+				if (isLongReadBam) {
+					logger.info("Processing long read BAM file");
+				}
 				if (noOfConsumerThreads > 0) {
-					summarizer = new BamSummarizerMT(noOfProducerThreads, noOfConsumerThreads, maxRecords,  validation,  isFullBamHeader);
+					summarizer = new BamSummarizerMT(noOfProducerThreads, noOfConsumerThreads, maxRecords,  validation,  isFullBamHeader, isLongReadBam);
 				} else {
-					summarizer = new BamSummarizer( maxRecords, validation, isFullBamHeader);
+					summarizer = new BamSummarizer( maxRecords, validation, isFullBamHeader, isLongReadBam);
 				}
 				break;
 			case XML:
@@ -299,6 +303,8 @@ public int setup(String args[]) throws Exception {
 			}
 
 			this.isFullBamHeader = options.hasFullBamHeaderOption();
+
+			this.isLongReadBam = options.hasLongReadBamOption();
 
 			// setup the ExecutorService thread pool
 			// the size of the pool is the smaller of the no of files, and the NO_OF_PROCESSORS variable

diff --git a/qprofiler2/src/org/qcmg/qprofiler2/bam/BamSummarizer.java b/qprofiler2/src/org/qcmg/qprofiler2/bam/BamSummarizer.java
@@ -34,24 +34,26 @@ public class BamSummarizer implements Summarizer {
 	private int maxRecords;
 	private String validation;
 	private boolean isFullBamHeader;
+	private boolean isLongReadBam;
 
 	private static final QLogger logger = QLoggerFactory.getLogger(BamSummarizer.class);	
 	public BamSummarizer() {}	 // default constructor	
-	public BamSummarizer( int maxRecords, String validation, boolean isFullBamHeader) {
+	public BamSummarizer( int maxRecords, String validation, boolean isFullBamHeader, boolean isLongReadBam) {
 		this.maxRecords = maxRecords;
 		this.validation = validation;
 		this.isFullBamHeader = isFullBamHeader;
+		this.isLongReadBam = isLongReadBam;
 	}	
 
-	public static BamSummaryReport createReport(SAMFileHeader header, String file, int maxRecords, boolean isFullBamHeader) throws IOException {
+	public static BamSummaryReport createReport(SAMFileHeader header, String file, int maxRecords, boolean isFullBamHeader, boolean isLongReadBam) throws IOException {
 
 		// create the SummaryReport
 
 		SAMSequenceDictionary samSeqDict  = header.getSequenceDictionary();
         // Natural order
         List<String> readGroupIds = header.getReadGroups().stream().map(SAMReadGroupRecord::getId).sorted(Comparator.comparing(String::toString)).collect(toList());
 
-        BamSummaryReport bamSummaryReport = new BamSummaryReport( maxRecords, isFullBamHeader );
+        BamSummaryReport bamSummaryReport = new BamSummaryReport( maxRecords, isFullBamHeader, isLongReadBam );
 		bamSummaryReport.setBamHeader(header, isFullBamHeader);		
 		bamSummaryReport.setSamSequenceDictionary(samSeqDict);
 		bamSummaryReport.setReadGroups(readGroupIds);		
@@ -68,7 +70,7 @@ public SummaryReport summarize(String input, String index) throws IOException {
 		SamReader reader = SAMFileReaderFactory.createSAMFileReaderAsStream(input, index, vs);
 
 		// create the SummaryReport		
-        BamSummaryReport bamSummaryReport = createReport(reader.getFileHeader(), input,  maxRecords, isFullBamHeader);
+        BamSummaryReport bamSummaryReport = createReport(reader.getFileHeader(), input,  maxRecords, isFullBamHeader, isLongReadBam);
 
 		boolean logLevelEnabled = logger.isLevelEnabled(QLevel.DEBUG);		
 		long currentRecordCount = 0;

diff --git a/qprofiler2/src/org/qcmg/qprofiler2/bam/BamSummarizerMT.java b/qprofiler2/src/org/qcmg/qprofiler2/bam/BamSummarizerMT.java
@@ -47,16 +47,18 @@ public class BamSummarizerMT implements Summarizer {
 	private final int maxRecords;
 	private final String validation;
 	private final boolean isFullBamHeader;
+	private final boolean isLongReadBam;
 	ValidationStringency vs;
 	private static final String UNMAPPED_READS = "Unmapped";
 
-	public BamSummarizerMT(int noOfProducerThreads, int noOfThreads, int maxNoOfRecords, String validation,boolean isFullBamHeader) {
+	public BamSummarizerMT(int noOfProducerThreads, int noOfThreads, int maxNoOfRecords, String validation,boolean isFullBamHeader, boolean isLongReadBam) {
 		super();
 		this.noOfProducerThreads = noOfProducerThreads;
 		this.noOfConsumerThreads = noOfThreads;
 		this.maxRecords = maxNoOfRecords;
 		this.validation = validation;
 		this.isFullBamHeader =  isFullBamHeader;
+		this.isLongReadBam = isLongReadBam;
 	}
 
 	@Override
@@ -103,7 +105,7 @@ public int compare(SAMSequenceRecord o1, SAMSequenceRecord o2) {
 
 		final long start = System.currentTimeMillis();		
 		final File file = new File(input);
-		final BamSummaryReport bamSummaryReport =  BamSummarizer.createReport(header, input,  maxRecords, isFullBamHeader );				 		
+		final BamSummaryReport bamSummaryReport =  BamSummarizer.createReport(header, input,  maxRecords, isFullBamHeader, isLongReadBam );
 		logger.info("will create " + noOfConsumerThreads + " consumer threads");
 
 		final CountDownLatch pLatch = new CountDownLatch(noOfProducerThreads);