Performance: Eliminate per-sequence char[] allocations in seven QC modules

uk/ac/babraham/FastQC/Modules/BasicStats.java

@@ -123,9 +123,10 @@ public void processSequence(Sequence sequence) {
 			if (sequence.getSequence().length() > maxLength) maxLength = sequence.getSequence().length();
 		}
 
-		char [] chars = sequence.getSequence().toCharArray();
-		for (int c=0;c<chars.length;c++) {			
-			switch (chars[c]) {
+		String seq = sequence.getSequence();
+		int seqLen = seq.length();
+		for (int c=0;c<seqLen;c++) {
+			switch (seq.charAt(c)) {
 				case 'G': ++gCount;break;
 				case 'A': ++aCount;break;
 				case 'T': ++tCount;break;
@@ -134,10 +135,12 @@ public void processSequence(Sequence sequence) {
 			}
 		}
 
-		chars = sequence.getQualityString().toCharArray();
-		for (int c=0;c<chars.length;c++) {
-			if (chars[c] < lowestChar) {
-				lowestChar = chars[c];
+		String qual = sequence.getQualityString();
+		int qualLen = qual.length();
+		for (int c=0;c<qualLen;c++) {
+			char ch = qual.charAt(c);
+			if (ch < lowestChar) {
+				lowestChar = ch;
 			}
 		}
 	}

uk/ac/babraham/FastQC/Modules/NContent.java

@@ -87,12 +87,13 @@ private synchronized void getPercentages () {
 
 	public void processSequence(Sequence sequence) {
 		calculated = false;
-		char [] seq = sequence.getSequence().toCharArray();
-		if (nCounts.length < seq.length) {
+		String seq = sequence.getSequence();
+		int seqLen = seq.length();
+		if (nCounts.length < seqLen) {
 			// We need to expand the size of the data structures
 
-			long [] nCountsNew = new long [seq.length];
-			long [] notNCountsNew = new long [seq.length];
+			long [] nCountsNew = new long [seqLen];
+			long [] notNCountsNew = new long [seqLen];
 
 			for (int i=0;i<nCounts.length;i++) {
 				nCountsNew[i] = nCounts[i];
@@ -103,8 +104,8 @@ public void processSequence(Sequence sequence) {
 			notNCounts = notNCountsNew;
 		}
 
-		for (int i=0;i<seq.length;i++) {
-			if (seq[i] == 'N') {
+		for (int i=0;i<seqLen;i++) {
+			if (seq.charAt(i) == 'N') {
 				++nCounts[i];
 			}
 			else {

uk/ac/babraham/FastQC/Modules/PerBaseQualityScores.java

@@ -129,10 +129,11 @@ private synchronized void getPercentages () {
 	public void processSequence(Sequence sequence) {
 
 		calculated = false;
-		char [] qual = sequence.getQualityString().toCharArray();
-		if (qualityCounts.length < qual.length) {
-
-			QualityCount [] qualityCountsNew = new QualityCount[qual.length];
+		String qual = sequence.getQualityString();
+		int qualLen = qual.length();
+		if (qualityCounts.length < qualLen) {
+
+			QualityCount [] qualityCountsNew = new QualityCount[qualLen];
 
 			for (int i=0;i<qualityCounts.length;i++) {
 				qualityCountsNew[i] = qualityCounts[i];
@@ -144,8 +145,8 @@ public void processSequence(Sequence sequence) {
 
 		}
 
-		for (int i=0;i<qual.length;i++) {
-			qualityCounts[i].addValue(qual[i]);
+		for (int i=0;i<qualLen;i++) {
+			qualityCounts[i].addValue(qual.charAt(i));
 		}
 
 	}

uk/ac/babraham/FastQC/Modules/PerBaseSequenceContent.java

@@ -113,13 +113,14 @@ private synchronized void getPercentages () {
 
 	public void processSequence(Sequence sequence) {
 		calculated = false;
-		char [] seq = sequence.getSequence().toCharArray();
-		if (gCounts.length < seq.length) {
-
-			long [] gCountsNew = new long [seq.length];
-			long [] aCountsNew = new long [seq.length];
-			long [] cCountsNew = new long [seq.length];
-			long [] tCountsNew = new long [seq.length];
+		String seq = sequence.getSequence();
+		int seqLen = seq.length();
+		if (gCounts.length < seqLen) {
+
+			long [] gCountsNew = new long [seqLen];
+			long [] aCountsNew = new long [seqLen];
+			long [] cCountsNew = new long [seqLen];
+			long [] tCountsNew = new long [seqLen];
 
 			for (int i=0;i<gCounts.length;i++) {
 				gCountsNew[i] = gCounts[i];
@@ -134,17 +135,18 @@ public void processSequence(Sequence sequence) {
 			cCounts = cCountsNew;
 		}
 
-		for (int i=0;i<seq.length;i++) {
-			if (seq[i] == 'G') {
+		for (int i=0;i<seqLen;i++) {
+			char b = seq.charAt(i);
+			if (b == 'G') {
 				++gCounts[i];
 			}
-			else if (seq[i] == 'A') {
+			else if (b == 'A') {
 				++aCounts[i];
 			}
-			else if (seq[i] == 'T') {
+			else if (b == 'T') {
 				++tCounts[i];
 			}
-			else if (seq[i] == 'C') {
+			else if (b == 'C') {
 				++cCounts[i];
 			}
 		}

uk/ac/babraham/FastQC/Modules/PerSequenceGCContent.java

@@ -171,42 +171,44 @@ public void processSequence(Sequence sequence) {
 		// encounter we need to reduce the number of models.  We can do this by
 		// rounding off the sequence once we get above a certain size
 
-		char [] seq = truncateSequence(sequence);
-
-		if (seq.length == 0) return; // Ignore empty sequences
+		String seq = truncateSequence(sequence);
+		int seqLen = seq.length();
+
+		if (seqLen == 0) return; // Ignore empty sequences
 
 
 		int thisSeqGCCount = 0;
-		for (int i=0;i<seq.length;i++) {
-			if (seq[i] == 'G' || seq[i] == 'C') {
+		for (int i=0;i<seqLen;i++) {
+			char b = seq.charAt(i);
+			if (b == 'G' || b == 'C') {
 				++thisSeqGCCount;
 			}
 		}
 
-		if (seq.length >= cachedModels.length) {
+		if (seqLen >= cachedModels.length) {
 			// We need to extend the length of cached models
 
-			GCModel [] longerModels = new GCModel[seq.length+1];
+			GCModel [] longerModels = new GCModel[seqLen+1];
 			for (int i=0;i<cachedModels.length;i++) {
 				longerModels[i] = cachedModels[i];
 			}
 
 			cachedModels = longerModels;
 		}
 
-		if (cachedModels[seq.length] == null) {
-			cachedModels[seq.length] = new GCModel(seq.length);
+		if (cachedModels[seqLen] == null) {
+			cachedModels[seqLen] = new GCModel(seqLen);
 		}
 
-		GCModelValue [] values = cachedModels[seq.length].getModelValues(thisSeqGCCount);
+		GCModelValue [] values = cachedModels[seqLen].getModelValues(thisSeqGCCount);
 
 		for (int i=0;i<values.length;i++) {
 			gcDistribution[values[i].percentage()] += values[i].increment();
 		}
 
 	}
 
-	private char [] truncateSequence (Sequence sequence) {
+	private String truncateSequence (Sequence sequence) {
 
 		String seq = sequence.getSequence();
 
@@ -215,14 +217,14 @@ public void processSequence(Sequence sequence) {
 
 		if (seq.length() > 1000) {
 			int length = (seq.length()/1000)*1000;
-			return seq.substring(0, length).toCharArray();
+			return seq.substring(0, length);
 		}
 		if (seq.length() > 100) {
 			int length = (seq.length()/100)*100;
-			return seq.substring(0, length).toCharArray();
+			return seq.substring(0, length);
 		}
 
-		return seq.toCharArray();		
+		return seq;
 
 	}
 

uk/ac/babraham/FastQC/Modules/PerSequenceQualityScores.java

@@ -88,18 +88,20 @@ private synchronized void calculateDistribution () {
 
 	public void processSequence(Sequence sequence) {
 
-		char [] seq = sequence.getQualityString().toCharArray();
+		String qual = sequence.getQualityString();
+		int qualLen = qual.length();
 		int averageQuality = 0;
 
-		for (int i=0;i<seq.length;i++) {
-			if (seq[i] < lowestChar) {
-				lowestChar = seq[i];
+		for (int i=0;i<qualLen;i++) {
+			char ch = qual.charAt(i);
+			if (ch < lowestChar) {
+				lowestChar = ch;
 			}
-			averageQuality += seq[i];
+			averageQuality += ch;
 		}
 
-		if (seq.length > 0) {
-			averageQuality /= seq.length;
+		if (qualLen > 0) {
+			averageQuality /= qualLen;
 
 			if (averageScoreCounts.containsKey(averageQuality)) {
 				long currentCount = averageScoreCounts.get(averageQuality);

uk/ac/babraham/FastQC/Modules/PerTileQualityScores.java

@@ -256,15 +256,16 @@ else if (splitID.length >=5) {
 			return;
 		}
 
-		char [] qual = sequence.getQualityString().toCharArray();
-		if (currentLength < qual.length) {
+		String qual = sequence.getQualityString();
+		int qualLen = qual.length();
+		if (currentLength < qualLen) {
 
 			Iterator<Integer> tiles = perTileQualityCounts.keySet().iterator();
 			while (tiles.hasNext()) {
 				int thisTile = tiles.next();
 
 				QualityCount [] qualityCounts = perTileQualityCounts.get(thisTile);
-				QualityCount [] qualityCountsNew = new QualityCount[qual.length];
+				QualityCount [] qualityCountsNew = new QualityCount[qualLen];
 
 				for (int i=0;i<qualityCounts.length;i++) {
 					qualityCountsNew[i] = qualityCounts[i];
@@ -275,7 +276,7 @@ else if (splitID.length >=5) {
 				perTileQualityCounts.put(thisTile, qualityCountsNew);
 			}
 
-			currentLength = qual.length;
+			currentLength = qualLen;
 
 		}
 
@@ -300,8 +301,8 @@ else if (splitID.length >=5) {
 
 		QualityCount [] qualityCounts = perTileQualityCounts.get(tile);
 
-		for (int i=0;i<qual.length;i++) {
-			qualityCounts[i].addValue(qual[i]);
+		for (int i=0;i<qualLen;i++) {
+			qualityCounts[i].addValue(qual.charAt(i));
 		}
 
 	}