Merge pull request #379 from AdamaJava/qmotif_revisit

Qmotif revisit

Author: holmeso

docs/qmotif/qmotif_1_2.md

@@ -14,6 +14,6 @@
 import org.qcmg.motif.util.RegionCounter;
 
 interface Algorithm {
-	public String getName();
-	public boolean applyTo(final SAMRecord read, Map<ChrPosition, RegionCounter> motifRefPositions);
+	String getName();
+	boolean applyTo(final SAMRecord read, Map<ChrPosition, RegionCounter> motifRefPositions);
 }

qmotif/src/org/qcmg/motif/Algorithm.java

@@ -9,7 +9,7 @@
 import java.io.File;
 import java.util.HashSet;
 import java.util.List;
-import java.util.concurrent.atomic.AtomicLong;
+import java.util.concurrent.atomic.LongAdder;
 
 import org.ini4j.Ini;
 import org.qcmg.common.model.ChrPosition;
@@ -23,13 +23,13 @@
 public final class Configuration {
 	private final int numberThreads;
 	private final File outputFile;
-	private final HashSet<Pair<File, File>> filePairs = new HashSet<Pair<File, File>>();
+	private final HashSet<Pair<File, File>> filePairs = new HashSet<>();
 	private final QueryExecutor filter;
 	private final Algorithm algorithm;
 	private final LoggerInfo loggerInfo;
 	private final String validation;
-	private final AtomicLong countReadFromInput;
-	private final AtomicLong countReadToCoverage;
+	private final LongAdder countReadFromInput;
+	private final LongAdder countReadToCoverage;
 	private final List<ChrPosition> excludes;
 	private final List<ChrPosition> includes;
 	private final Integer windowSize;
@@ -67,7 +67,7 @@ public Configuration(final Options options) throws Exception {
 			stageTwoMotifs = new Motifs(Boolean.parseBoolean(IniUtils.getEntry(iniFile, MotifConstants.PARAMS, MotifConstants.STAGE_2_STRING_REV_COMP)), s2Motifs.split(Constants.COMMA_STRING));
 		}
 
-		mAndR = new MotifsAndRegexes(stageOneMotifs, stageOneRegex, stageTwoMotifs, stageTwoRegex, windowSize.intValue());
+		mAndR = new MotifsAndRegexes(stageOneMotifs, stageOneRegex, stageTwoMotifs, stageTwoRegex, windowSize);
 		algorithm = new MotifCoverageAlgorithm(mAndR);
 
 		// get excludes and includes
@@ -83,7 +83,7 @@ public Configuration(final Options options) throws Exception {
 		inferFilePairings(bamFileNames, baiFileNames);
 
 		if (options.hasNumberThreadsOption()) {
-			numberThreads = options.getNumberThreads().intValue();
+			numberThreads = options.getNumberThreads();
 		} else {
 			numberThreads = 1;
 		}
@@ -96,8 +96,8 @@ public Configuration(final Options options) throws Exception {
 
 		outputFile = new File(options.getOutputXmlFileName());
 		checkFileExistence();
-		countReadFromInput = new AtomicLong();
-		countReadToCoverage = new AtomicLong();
+		countReadFromInput = new LongAdder();
+		countReadToCoverage = new LongAdder();
 
 		String includesOnlyModeS = IniUtils.getEntry(iniFile, MotifConstants.PARAMS, MotifConstants.INCLUDES_ONLY_MODE);
 		if (null != includesOnlyModeS) {
@@ -106,22 +106,18 @@ public Configuration(final Options options) throws Exception {
 
 	}
 
-	public AtomicLong getInputReadsCount(){
+	public LongAdder getInputReadsCount(){
 		return countReadFromInput;
 	}
 
-	public AtomicLong getCoverageReadsCount(){
+	public LongAdder getCoverageReadsCount(){
 		return countReadToCoverage;
 	}
 
 	public int getNumberThreads() {
 		return numberThreads;
 	}
 
-	public File getOutputFile() {
-		return outputFile;
-	}
-
 	public HashSet<Pair<File, File>> getFilePairs() {
 		return filePairs;
 	}

qmotif/src/org/qcmg/motif/Configuration.java

@@ -12,8 +12,7 @@
 import java.util.Iterator;
 import java.util.List;
 import java.util.Map;
-import java.util.concurrent.atomic.AtomicLong;
-import java.util.regex.Pattern;
+import java.util.concurrent.atomic.LongAdder;
 
 import htsjdk.samtools.SamReader;
 import htsjdk.samtools.SAMRecord;
@@ -33,44 +32,39 @@ class CoverageJob implements Job {
 	private final ChrPosition cp;
 	private final QLogger logger;
 	private final QueryExecutor filter;
-	private final HashSet<SamReader> fileReaders = new HashSet<SamReader>();
+	private final HashSet<File> fileReaders = new HashSet<>();
 	private final Algorithm alg;
-	private final AtomicLong counterIn;
-	private final AtomicLong counterOut;
+	private final LongAdder counterIn;
+	private final LongAdder counterOut;
 	private final List<ChrPosition> includes;
 	private final List<ChrPosition> excludes;
 	private Map<ChrPosition, RegionCounter> chrSpecificRegions;
 	private int windowSize = 1000000;		// default to a mill
 	private final AbstractQueue<SAMRecord> outputQueue;
+	String validation;
 
 	CoverageJob(final ChrPosition cp,
 			final HashSet<Pair<File, File>> filePairs, final QueryExecutor filter,
-			final Algorithm algorithm, final AtomicLong counterIn,final AtomicLong counterOut, 
-			Integer windowSize, Pattern regex, AbstractQueue<SAMRecord> outputQueue) throws Exception {
-		this(cp, filePairs, filter, algorithm, counterIn, counterOut, null, windowSize, outputQueue, null, null);
-	}
-	CoverageJob(final ChrPosition cp,
-			final HashSet<Pair<File, File>> filePairs, final QueryExecutor filter,
-			final Algorithm algorithm, final AtomicLong counterIn, final AtomicLong counterOut, final String validation,
+			final Algorithm algorithm, final LongAdder counterIn, final LongAdder counterOut, final String validation,
 			 Integer windowSize, AbstractQueue<SAMRecord> outputQueue,
-			List<ChrPosition> includes, List<ChrPosition> excludes) throws Exception {
+			List<ChrPosition> includes, List<ChrPosition> excludes) {
 		assert (cp.getLength() > -1);
 		this.cp = cp;
 		this.alg = algorithm;
+		this.validation = validation;
 		this.counterIn = counterIn;
 		this.counterOut = counterOut;
 		this.filter = filter;
 		this.logger = QLoggerFactory.getLogger(CoverageJob.class);
 		this.includes = includes;
 		this.excludes = excludes;
 		if (null != windowSize) {
-			this.windowSize = windowSize.intValue();
+			this.windowSize = windowSize;
 		}
 		this.outputQueue = outputQueue;
 		for (final Pair<File, File> pair : filePairs) {
 			File bamFile = pair.left();
-			SamReader reader = SAMFileReaderFactory.createSAMFileReader(bamFile, validation);
-			fileReaders.add(reader);
+			fileReaders.add(bamFile);
 		}
 		logger.debug("Length of sequence to be processed by job '" + cp.toIGVString() + " : " + cp.getLength());
 	}
@@ -110,15 +104,15 @@ void constructCoverageMap() {
 	}
 
 	private void performCoverage() throws Exception {
- 		for (final SamReader fileReader : fileReaders) {
- 			
+ 		for (final File file : fileReaders) {
+			SamReader fileReader = SAMFileReaderFactory.createSAMFileReader(file, validation);
 			Iterator<SAMRecord> iter = "unmapped".equals(cp.getChromosome()) ? fileReader.queryUnmapped() : fileReader.query(cp.getChromosome(), cp.getStartPosition(), cp.getEndPosition(), true);
 			long recordCounterIn = 0;
 			long recordCounterOut = 0; 
 			while (iter.hasNext()) {
 				SAMRecord read = iter.next();
 
-				counterIn.incrementAndGet();   //count input read number
+				counterIn.increment();   //count input read number
 
 				if (++recordCounterIn % 10000000 == 0) {
 					logger.debug("Hit " + recordCounterIn + " record for " + cp.toIGVString());
@@ -127,26 +121,25 @@ private void performCoverage() throws Exception {
 				if (MotifConstants.UNMAPPED.equals(cp.getChromosome()) || read.getReferenceName().equals(cp.getChromosome())) {
 					if (null == filter) {
 						recordCounterOut ++;
-						counterOut.incrementAndGet();    //count output read number
+						counterOut.increment();    //count output read number
 						if (alg.applyTo(read, chrSpecificRegions)) outputQueue.add(read);
 					} else if (filter.Execute(read)) {
 						recordCounterOut ++;
-						counterOut.incrementAndGet(); //count output read number
+						counterOut.increment(); //count output read number
 						if (alg.applyTo(read, chrSpecificRegions)) outputQueue.add(read);
 					}
 				} else {
 					logger.info("ref names did not match!");
 				}
 			}
 			fileReader.close();
+
+            String sb = cp.toIGVString() + ":\n" +
+                    "read " + recordCounterIn + " records, of which " + recordCounterOut + " satisfied the query\n" +
+                    "number in counterIn instance is: " + counterIn.longValue() + "\n" +
+                    "number in counterOut instance is: " + counterOut.longValue() + "\n";
 
-			StringBuilder sb = new StringBuilder();
-			sb.append(cp.toIGVString()).append(":\n");
-			sb.append("read ").append(recordCounterIn).append(" records, of which ").append(recordCounterOut).append(" satisfied the query\n");
-			sb.append("number in counterIn instance is: ").append(counterIn.get()).append("\n");
-			sb.append("number in counterOut instance is: ").append(counterOut.get()).append("\n");
-			
-			logger.info(sb.toString());
+			logger.info(sb);
 		}
 	}
 }

qmotif/src/org/qcmg/motif/CoverageJob.java

@@ -9,12 +9,11 @@
 import java.util.Map;
 
 import org.qcmg.common.model.ChrPosition;
-import org.qcmg.common.model.ChrRangePosition;
 import org.qcmg.motif.util.RegionCounter;
 
 interface Job {
-	public Map<ChrPosition, RegionCounter> getResults();
-	public void run() throws Exception;
+	Map<ChrPosition, RegionCounter> getResults();
+	void run() throws Exception;
 	@Override
-	public String toString();
+    String toString();
 }