/**
* Process a single {@link SequenceFile}
*
* @param sequenceFile file to process
* @throws FileProcessorException if an error occurs while processing
*/
private void processSingleFile(SequenceFile sequenceFile) throws FileProcessorException {
Path fileToProcess = sequenceFile.getFile();
AnalysisFastQC.AnalysisFastQCBuilder analysis =
AnalysisFastQC.builder()
.executionManagerAnalysisId(EXECUTION_MANAGER_ANALYSIS_ID)
.description(
messageSource.getMessage(
"fastqc.file.processor.analysis.description",
null,
LocaleContextHolder.getLocale()));
try {
uk.ac.babraham.FastQC.Sequence.SequenceFile fastQCSequenceFile =
SequenceFactory.getSequenceFile(fileToProcess.toFile());
BasicStats basicStats = new BasicStats();
PerBaseQualityScores pbqs = new PerBaseQualityScores();
PerSequenceQualityScores psqs = new PerSequenceQualityScores();
OverRepresentedSeqs overRep = new OverRepresentedSeqs();
QCModule[] moduleList = new QCModule[] {basicStats, pbqs, psqs, overRep};
logger.debug("Launching FastQC analysis modules on all sequences.");
while (fastQCSequenceFile.hasNext()) {
Sequence sequence = fastQCSequenceFile.next();
for (QCModule module : moduleList) {
module.processSequence(sequence);
}
}
logger.debug("Finished FastQC analysis modules.");
handleBasicStats(basicStats, analysis);
handlePerBaseQualityScores(pbqs, analysis);
handlePerSequenceQualityScores(psqs, analysis);
handleDuplicationLevel(overRep.duplicationLevelModule(), analysis);
Set<OverrepresentedSequence> overrepresentedSequences =
handleOverRepresentedSequences(overRep);
logger.trace("Saving FastQC analysis.");
analysis.overrepresentedSequences(overrepresentedSequences);
sequenceFile.setFastQCAnalysis(analysis.build());
sequenceFileRepository.save(sequenceFile);
} catch (Exception e) {
logger.error("FastQC failed to process the sequence file. Stack trace follows.", e);
throw new FileProcessorException("FastQC failed to parse the sequence file.", e);
}
}
/**
* Handle getting over represented sequences from fastqc.
*
* @param seqs overrepresented sequences.
* @return a collection of {@link OverrepresentedSequence} corresponding to the FastQC {@link
* OverRepresentedSeqs}.
*/
private Set<OverrepresentedSequence> handleOverRepresentedSequences(OverRepresentedSeqs seqs) {
OverrepresentedSeq[] sequences = seqs.getOverrepresentedSequences();
if (sequences == null) {
return Collections.emptySet();
}
Set<OverrepresentedSequence> overrepresentedSequences = new HashSet<>(sequences.length);
for (OverrepresentedSeq s : sequences) {
String sequenceString = s.seq();
int count = s.count();
BigDecimal percent = BigDecimal.valueOf(s.percentage());
String possibleSource = s.contaminantHit();
overrepresentedSequences.add(
new OverrepresentedSequence(sequenceString, count, percent, possibleSource));
}
return overrepresentedSequences;
}
