/**
* Uncompress a file to a temporary file.
*
* @param context Step context
* @param file file to uncompress
* @return the path to the uncompressed file
* @throws IOException if an error occurs while creating the uncompressed file
*/
private File uncompressFile(final TaskContext context, final DataFile file) throws IOException {
checkNotNull(file, "file argument cannot be null");
final DataFile realFile;
final DataProtocol protocol = file.getProtocol();
// Get the underlying file if the file protocol is a storage protocol
if (protocol instanceof StorageDataProtocol) {
realFile = ((StorageDataProtocol) protocol).getUnderLyingData(file);
} else {
realFile = file;
}
final File outputFile =
Files.createTempFile(
context.getLocalTempDirectory().toPath(),
MODULE_NAME + "-",
realFile.getExtension())
.toFile();
context.getLogger().fine("Uncompress/copy " + realFile + " to " + outputFile);
DataFiles.copy(realFile, new DataFile(outputFile));
return outputFile;
}
/**
* Set the "gtf.feature.exon" and "gtf.tag.exon.parent.transcript" parameter from the expression
* step parameters.
*
* @param context the context of the task
* @throws EoulsanException if more than one expression step exists
*/
private void searchExpressionStepParameters(final TaskContext context) throws EoulsanException {
int count = 0;
for (Step step : context.getWorkflow().getSteps()) {
if (AbstractExpressionModule.MODULE_NAME.equals(step.getModuleName())) {
for (Parameter p : step.getParameters()) {
switch (p.getName()) {
case AbstractExpressionModule.OLD_GENOMIC_TYPE_PARAMETER_NAME:
case AbstractExpressionModule.GENOMIC_TYPE_PARAMETER_NAME:
gtfFeatureExon = p.getStringValue();
break;
case AbstractExpressionModule.OLD_ATTRIBUTE_ID_PARAMETER_NAME:
case AbstractExpressionModule.ATTRIBUTE_ID_PARAMETER_NAME:
gtfTagExonParentTranscript = p.getStringValue();
break;
default:
break;
}
}
count++;
}
}
if (count == 0) {
throw new EoulsanException("No expression step found in the workflow");
}
if (count > 1) {
throw new EoulsanException("Found more than one expression step in the workflow");
}
}
@Override
public TaskResult execute(final TaskContext context, final TaskStatus status) {
try {
final StringBuilder additionalArguments = new StringBuilder();
final Map<String, String> additionalDescription = new HashMap<>();
final List<File> temporaryFiles = new ArrayList<>();
// Search expression parameter is needed
if (this.useExpressionStepParameters) {
searchExpressionStepParameters(context);
}
if (this.gtfFile) {
// Get the annotation data
final Data annotationData =
context.getInputData(this.gtfFormat ? ANNOTATION_GTF : ANNOTATION_GFF);
// Get the annotation DataFile
final DataFile gffFile = annotationData.getDataFile();
final File gffFilePath = uncompressFileIfNecessary(context, temporaryFiles, gffFile);
additionalArguments.append("--sjdbGTFfile");
additionalArguments.append(' ');
additionalArguments.append(gffFilePath.getAbsolutePath());
additionalArguments.append(' ');
additionalDescription.put("sjdbGTFfile", computeMD5SumFile(gffFilePath));
}
if (this.overhang != null) {
additionalArguments.append("--sjdbOverhang");
additionalArguments.append(' ');
additionalArguments.append(this.overhang.toString());
additionalArguments.append(' ');
additionalDescription.put("sjdbOverhang", this.overhang.toString());
}
if (this.gtfTagExonParentTranscript != null) {
additionalArguments.append("--sjdbGTFtagExonParentTranscript");
additionalArguments.append(' ');
additionalArguments.append(this.gtfTagExonParentTranscript);
additionalArguments.append(' ');
additionalDescription.put(
"sjdbGTFtagExonParentTranscript", this.gtfTagExonParentTranscript);
}
if (this.gtfFeatureExon != null) {
additionalArguments.append("--sjdbGTFfeatureExon");
additionalArguments.append(' ');
additionalArguments.append(this.gtfFeatureExon);
additionalArguments.append(' ');
additionalDescription.put("sjdbGTFfeatureExon", this.gtfFeatureExon);
}
if (this.chrStartEndFilename != null) {
DataFile chrStartEndFile = new DataFile(this.chrStartEndFilename);
if (!chrStartEndFile.exists()) {
throw new IOException("Unable to read chromosome startend file: " + chrStartEndFile);
}
final File chrStartEndFilePath =
uncompressFileIfNecessary(context, temporaryFiles, chrStartEndFile);
additionalArguments.append("--sjdbFileChrStartEnd");
additionalArguments.append(' ');
additionalArguments.append(chrStartEndFilePath.getAbsolutePath());
additionalArguments.append(' ');
additionalDescription.put("sjdbFileChrStartEnd", computeMD5SumFile(chrStartEndFilePath));
}
if (this.genomeSAindexNbases != null) {
additionalArguments.append("--genomeSAindexNbases");
additionalArguments.append(' ');
additionalArguments.append(this.genomeSAindexNbases.toString());
additionalArguments.append(' ');
additionalDescription.put("genomeSAindexNbases", this.genomeSAindexNbases.toString());
}
if (this.genomeChrBinNbits != null) {
additionalArguments.append("--genomeChrBinNbits");
additionalArguments.append(' ');
additionalArguments.append(this.genomeChrBinNbits.toString());
additionalArguments.append(' ');
additionalDescription.put("genomeChrBinNbits", this.genomeChrBinNbits.toString());
}
status.setProgressMessage(this.mapper.getMapperName() + " index creation");
// Create the index
GenomeMapperIndexGeneratorModule.execute(
this.mapper,
context,
additionalArguments.toString(),
additionalDescription,
Common.getThreadsNumber(this.localThreads, this.maxLocalThreads));
// Remove temporary files
for (File temporaryFile : temporaryFiles) {
if (!temporaryFile.delete()) {
context.getLogger().warning("Cannot remove temporary file: " + temporaryFile);
}
}
} catch (IOException | EoulsanException e) {
return status.createTaskResult(e);
}
return status.createTaskResult();
}
