/**
* Calculates a histogram using the estimateRoi method to estimate the effective yield doing x
* sequencing for x=1..10.
*/
public Histogram<Double> calculateRoiHistogram() {
if (ESTIMATED_LIBRARY_SIZE == null) {
try {
calculateDerivedMetrics();
if (ESTIMATED_LIBRARY_SIZE == null) return null;
} catch (IllegalStateException ise) {
return null;
}
}
long uniquePairs = READ_PAIRS_EXAMINED - READ_PAIR_DUPLICATES;
Histogram<Double> histo = new Histogram<Double>();
for (double x = 1; x <= 100; x += 1) {
histo.increment(x, estimateRoi(ESTIMATED_LIBRARY_SIZE, x, READ_PAIRS_EXAMINED, uniquePairs));
}
return histo;
}
Histogram<Integer> getMeanQualityHistogram() {
final String label = useOriginalQualities ? "MEAN_ORIGINAL_QUALITY" : "MEAN_QUALITY";
final Histogram<Integer> meanQualities = new Histogram<>("CYCLE", label);
int firstReadLength = 0;
for (int cycle = 0; cycle < firstReadTotalsByCycle.length; ++cycle) {
if (firstReadTotalsByCycle[cycle] > 0) {
meanQualities.increment(
cycle, firstReadTotalsByCycle[cycle] / firstReadCountsByCycle[cycle]);
firstReadLength = cycle;
}
}
for (int i = 0; i < secondReadTotalsByCycle.length; ++i) {
if (secondReadCountsByCycle[i] > 0) {
final int cycle = firstReadLength + i;
meanQualities.increment(cycle, secondReadTotalsByCycle[i] / secondReadCountsByCycle[i]);
}
}
return meanQualities;
}
@Override
protected int doWork() {
IOUtil.assertFileIsReadable(INPUT);
IOUtil.assertFileIsWritable(OUTPUT);
IOUtil.assertFileIsReadable(REFERENCE_SEQUENCE);
// Setup all the inputs
final ProgressLogger progress = new ProgressLogger(log, 10000000, "Processed", "loci");
final ReferenceSequenceFileWalker refWalker =
new ReferenceSequenceFileWalker(REFERENCE_SEQUENCE);
final SamReader in =
SamReaderFactory.makeDefault().referenceSequence(REFERENCE_SEQUENCE).open(INPUT);
final SamLocusIterator iterator = getLocusIterator(in);
final List<SamRecordFilter> filters = new ArrayList<SamRecordFilter>();
final CountingFilter dupeFilter = new CountingDuplicateFilter();
final CountingFilter mapqFilter = new CountingMapQFilter(MINIMUM_MAPPING_QUALITY);
final CountingPairedFilter pairFilter = new CountingPairedFilter();
filters.add(mapqFilter);
filters.add(dupeFilter);
if (!COUNT_UNPAIRED) {
filters.add(pairFilter);
}
filters.add(
new SecondaryAlignmentFilter()); // Not a counting filter because we never want to count
// reads twice
iterator.setSamFilters(filters);
iterator.setEmitUncoveredLoci(true);
iterator.setMappingQualityScoreCutoff(0); // Handled separately because we want to count bases
iterator.setQualityScoreCutoff(0); // Handled separately because we want to count bases
iterator.setIncludeNonPfReads(false);
final int max = COVERAGE_CAP;
final long[] HistogramArray = new long[max + 1];
final long[] baseQHistogramArray = new long[Byte.MAX_VALUE];
final boolean usingStopAfter = STOP_AFTER > 0;
final long stopAfter = STOP_AFTER - 1;
long counter = 0;
long basesExcludedByBaseq = 0;
long basesExcludedByOverlap = 0;
long basesExcludedByCapping = 0;
// Loop through all the loci
while (iterator.hasNext()) {
final SamLocusIterator.LocusInfo info = iterator.next();
// Check that the reference is not N
final ReferenceSequence ref = refWalker.get(info.getSequenceIndex());
final byte base = ref.getBases()[info.getPosition() - 1];
if (base == 'N') continue;
// Figure out the coverage while not counting overlapping reads twice, and excluding various
// things
final HashSet<String> readNames = new HashSet<String>(info.getRecordAndPositions().size());
int pileupSize = 0;
for (final SamLocusIterator.RecordAndOffset recs : info.getRecordAndPositions()) {
if (recs.getBaseQuality() < MINIMUM_BASE_QUALITY) {
++basesExcludedByBaseq;
continue;
}
if (!readNames.add(recs.getRecord().getReadName())) {
++basesExcludedByOverlap;
continue;
}
pileupSize++;
if (pileupSize <= max) {
baseQHistogramArray[recs.getRecord().getBaseQualities()[recs.getOffset()]]++;
}
}
final int depth = Math.min(readNames.size(), max);
if (depth < readNames.size()) basesExcludedByCapping += readNames.size() - max;
HistogramArray[depth]++;
// Record progress and perhaps stop
progress.record(info.getSequenceName(), info.getPosition());
if (usingStopAfter && ++counter > stopAfter) break;
}
// Construct and write the outputs
final Histogram<Integer> histo = new Histogram<Integer>("coverage", "count");
for (int i = 0; i < HistogramArray.length; ++i) {
histo.increment(i, HistogramArray[i]);
}
// Construct and write the outputs
final Histogram<Integer> baseQHisto = new Histogram<Integer>("value", "baseq_count");
for (int i = 0; i < baseQHistogramArray.length; ++i) {
baseQHisto.increment(i, baseQHistogramArray[i]);
}
final WgsMetrics metrics = generateWgsMetrics();
metrics.GENOME_TERRITORY = (long) histo.getSumOfValues();
metrics.MEAN_COVERAGE = histo.getMean();
metrics.SD_COVERAGE = histo.getStandardDeviation();
metrics.MEDIAN_COVERAGE = histo.getMedian();
metrics.MAD_COVERAGE = histo.getMedianAbsoluteDeviation();
final long basesExcludedByDupes = getBasesExcludedBy(dupeFilter);
final long basesExcludedByMapq = getBasesExcludedBy(mapqFilter);
final long basesExcludedByPairing = getBasesExcludedBy(pairFilter);
final double total = histo.getSum();
final double totalWithExcludes =
total
+ basesExcludedByDupes
+ basesExcludedByMapq
+ basesExcludedByPairing
+ basesExcludedByBaseq
+ basesExcludedByOverlap
+ basesExcludedByCapping;
metrics.PCT_EXC_DUPE = basesExcludedByDupes / totalWithExcludes;
metrics.PCT_EXC_MAPQ = basesExcludedByMapq / totalWithExcludes;
metrics.PCT_EXC_UNPAIRED = basesExcludedByPairing / totalWithExcludes;
metrics.PCT_EXC_BASEQ = basesExcludedByBaseq / totalWithExcludes;
metrics.PCT_EXC_OVERLAP = basesExcludedByOverlap / totalWithExcludes;
metrics.PCT_EXC_CAPPED = basesExcludedByCapping / totalWithExcludes;
metrics.PCT_EXC_TOTAL = (totalWithExcludes - total) / totalWithExcludes;
metrics.PCT_1X =
MathUtil.sum(HistogramArray, 1, HistogramArray.length) / (double) metrics.GENOME_TERRITORY;
metrics.PCT_5X =
MathUtil.sum(HistogramArray, 5, HistogramArray.length) / (double) metrics.GENOME_TERRITORY;
metrics.PCT_10X =
MathUtil.sum(HistogramArray, 10, HistogramArray.length) / (double) metrics.GENOME_TERRITORY;
metrics.PCT_15X =
MathUtil.sum(HistogramArray, 15, HistogramArray.length) / (double) metrics.GENOME_TERRITORY;
metrics.PCT_20X =
MathUtil.sum(HistogramArray, 20, HistogramArray.length) / (double) metrics.GENOME_TERRITORY;
metrics.PCT_25X =
MathUtil.sum(HistogramArray, 25, HistogramArray.length) / (double) metrics.GENOME_TERRITORY;
metrics.PCT_30X =
MathUtil.sum(HistogramArray, 30, HistogramArray.length) / (double) metrics.GENOME_TERRITORY;
metrics.PCT_40X =
MathUtil.sum(HistogramArray, 40, HistogramArray.length) / (double) metrics.GENOME_TERRITORY;
metrics.PCT_50X =
MathUtil.sum(HistogramArray, 50, HistogramArray.length) / (double) metrics.GENOME_TERRITORY;
metrics.PCT_60X =
MathUtil.sum(HistogramArray, 60, HistogramArray.length) / (double) metrics.GENOME_TERRITORY;
metrics.PCT_70X =
MathUtil.sum(HistogramArray, 70, HistogramArray.length) / (double) metrics.GENOME_TERRITORY;
metrics.PCT_80X =
MathUtil.sum(HistogramArray, 80, HistogramArray.length) / (double) metrics.GENOME_TERRITORY;
metrics.PCT_90X =
MathUtil.sum(HistogramArray, 90, HistogramArray.length) / (double) metrics.GENOME_TERRITORY;
metrics.PCT_100X =
MathUtil.sum(HistogramArray, 100, HistogramArray.length)
/ (double) metrics.GENOME_TERRITORY;
final MetricsFile<WgsMetrics, Integer> out = getMetricsFile();
out.addMetric(metrics);
out.addHistogram(histo);
if (INCLUDE_BQ_HISTOGRAM) {
out.addHistogram(baseQHisto);
}
out.write(OUTPUT);
return 0;
}
/**
* Main method for the program. Checks that all input files are present and readable and that the
* output file can be written to. Then iterates through all the records accumulating metrics.
* Finally writes metrics file
*/
protected int doWork() {
IOUtil.assertFileIsReadable(INPUT);
IOUtil.assertFileIsWritable(OUTPUT);
final SamReader reader =
SamReaderFactory.makeDefault().referenceSequence(REFERENCE_SEQUENCE).open(INPUT);
final Histogram<Integer> mismatchesHist = new Histogram<Integer>("Predicted", "Mismatches");
final Histogram<Integer> totalHist = new Histogram<Integer>("Predicted", "Total_Bases");
final Map<String, Histogram> mismatchesByTypeHist = new HashMap<String, Histogram>();
final Map<String, Histogram> totalByTypeHist = new HashMap<String, Histogram>();
// Set up the histograms
byte[] bases = {'A', 'C', 'G', 'T'};
for (final byte base : bases) {
final Histogram<Integer> h = new Histogram<Integer>("Predicted", (char) base + ">");
mismatchesByTypeHist.put((char) base + ">", h);
final Histogram<Integer> h2 = new Histogram<Integer>("Predicted", ">" + (char) base);
mismatchesByTypeHist.put(">" + (char) base, h2);
}
for (final byte base : bases) {
final Histogram<Integer> h = new Histogram<Integer>("Predicted", (char) base + ">");
totalByTypeHist.put((char) base + ">", h);
final Histogram<Integer> h2 = new Histogram<Integer>("Predicted", ">" + (char) base);
totalByTypeHist.put(">" + (char) base, h2);
}
for (final SAMRecord record : reader) {
// Ignore these as we don't know the truth
if (record.getReadUnmappedFlag() || record.isSecondaryOrSupplementary()) {
continue;
}
final byte[] readBases = record.getReadBases();
final byte[] readQualities = record.getBaseQualities();
final byte[] refBases = SequenceUtil.makeReferenceFromAlignment(record, false);
// We've seen stranger things
if (readQualities.length != readBases.length) {
throw new PicardException(
"Missing Qualities ("
+ readQualities.length
+ ","
+ readBases.length
+ ") : "
+ record.getSAMString());
}
if (refBases.length != readBases.length) {
throw new PicardException(
"The read length did not match the inferred reference length, please check your MD and CIGAR.");
}
int cycleIndex; // zero-based
if (record.getReadNegativeStrandFlag()) {
cycleIndex = readBases.length - 1 + CYCLE_OFFSET;
} else {
cycleIndex = CYCLE_OFFSET;
}
for (int i = 0; i < readBases.length; i++) {
if (-1 == CYCLE || cycleIndex == CYCLE) {
if ('-' != refBases[i] && '0' != refBases[i]) { // not insertion and not soft-clipped
if (!SequenceUtil.basesEqual(readBases[i], refBases[i])) { // mismatch
mismatchesHist.increment((int) readQualities[i]);
if (SequenceUtil.isValidBase(refBases[i])) {
mismatchesByTypeHist
.get((char) refBases[i] + ">")
.increment((int) readQualities[i]);
}
if (SequenceUtil.isValidBase(readBases[i])) {
mismatchesByTypeHist
.get(">" + (char) readBases[i])
.increment((int) readQualities[i]);
}
} else {
mismatchesHist.increment(
(int) readQualities[i], 0); // to make sure the bin will exist
}
totalHist.increment((int) readQualities[i]);
if (SequenceUtil.isValidBase(readBases[i])) {
totalByTypeHist.get(">" + (char) readBases[i]).increment((int) readQualities[i]);
}
if (SequenceUtil.isValidBase(refBases[i])) {
totalByTypeHist.get((char) refBases[i] + ">").increment((int) readQualities[i]);
}
}
}
cycleIndex += record.getReadNegativeStrandFlag() ? -1 : 1;
}
}
CloserUtil.close(reader);
final Histogram<Integer> hist = new Histogram<Integer>("Predicted", "Observed");
double sumOfSquaresError = 0.0;
// compute the aggregate phred values
for (final Integer key : mismatchesHist.keySet()) {
final double numMismatches = mismatchesHist.get(key).getValue();
final double numBases = totalHist.get(key).getValue();
final double phredErr = Math.log10(numMismatches / numBases) * -10.0;
sumOfSquaresError += (0 == numMismatches) ? 0.0 : (key - phredErr) * (key - phredErr);
hist.increment(key, phredErr);
// make sure the bin will exist
for (final byte base : bases) {
mismatchesByTypeHist.get(">" + (char) base).increment(key, 0.0);
mismatchesByTypeHist.get((char) base + ">").increment(key, 0.0);
totalByTypeHist.get(">" + (char) base).increment(key, 0.0);
totalByTypeHist.get((char) base + ">").increment(key, 0.0);
}
}
final QualityScoreAccuracyMetrics metrics = new QualityScoreAccuracyMetrics();
metrics.SUM_OF_SQUARE_ERROR = sumOfSquaresError;
final MetricsFile<QualityScoreAccuracyMetrics, Integer> out = getMetricsFile();
out.addMetric(metrics);
out.addHistogram(hist);
for (final byte base : bases) {
// >base : histograms for mismatches *to* the given base
Histogram<Integer> m = mismatchesByTypeHist.get(">" + (char) base);
Histogram<Integer> t = totalByTypeHist.get(">" + (char) base);
Histogram<Integer> h = new Histogram<Integer>(m.getBinLabel(), m.getValueLabel());
for (final Integer key : m.keySet()) {
final double numMismatches = m.get(key).getValue();
final double numBases = t.get(key).getValue();
final double phredErr = Math.log10(numMismatches / numBases) * -10.0;
h.increment(key, phredErr);
}
out.addHistogram(h);
// base> : histograms for mismatches *from* the given base
m = mismatchesByTypeHist.get((char) base + ">");
t = totalByTypeHist.get(">" + (char) base);
h = new Histogram<Integer>(m.getBinLabel(), m.getValueLabel());
for (final Integer key : m.keySet()) {
final double numMismatches = m.get(key).getValue();
final double numBases = t.get(key).getValue();
final double phredErr = Math.log10(numMismatches / numBases) * -10.0;
h.increment(key, phredErr);
}
out.addHistogram(h);
}
out.addHistogram(mismatchesHist);
out.addHistogram(totalHist);
out.write(OUTPUT);
return 0;
}