private List<Sample> getFileSamples(
Study study,
File file,
ObjectMap indexFileModifyParams,
boolean simulate,
QueryOptions options,
String sessionId)
throws CatalogException {
List<Sample> sampleList;
QueryOptions queryOptions =
new QueryOptions(
"include",
Arrays.asList("projects.studies.samples.id", "projects.studies.samples.name"));
if (file.getSampleIds() == null || file.getSampleIds().isEmpty()) {
// Read samples from file
List<String> sampleNames = null;
switch (file.getBioformat()) {
case VARIANT:
{
if (file.getAttributes().containsKey("variantSource")) {
Object variantSource = file.getAttributes().get("variantSource");
if (variantSource instanceof VariantSource) {
sampleNames = ((VariantSource) variantSource).getSamples();
} else if (variantSource instanceof Map) {
sampleNames = new ObjectMap((Map) variantSource).getAsStringList("samples");
} else {
logger.warn(
"Unexpected object type of variantSource ({}) in file attributes. Expected {} or {}",
variantSource.getClass(),
VariantSource.class,
Map.class);
}
}
if (sampleNames == null) {
VariantSource variantSource = readVariantSource(catalogManager, study, file);
indexFileModifyParams
.get("attributes", ObjectMap.class)
.put("variantSource", variantSource);
sampleNames = variantSource.getSamples();
}
}
break;
default:
return new LinkedList<>();
// throw new CatalogException("Unknown to get samples names from
// bioformat " + file.getBioformat());
}
// Find matching samples in catalog with the sampleName from the VariantSource.
queryOptions.add("name", sampleNames);
sampleList = catalogManager.getAllSamples(study.getId(), queryOptions, sessionId).getResult();
// check if all file samples exists on Catalog
if (sampleList.size()
!= sampleNames.size()) { // Size does not match. Find the missing samples.
Set<String> set = new HashSet<>(sampleNames);
for (Sample sample : sampleList) {
set.remove(sample.getName());
}
logger.warn("Missing samples: m{}", set);
if (options.getBoolean(CREATE_MISSING_SAMPLES, true)) {
for (String sampleName : set) {
if (simulate) {
sampleList.add(new Sample(-1, sampleName, file.getName(), null, null));
} else {
sampleList.add(
catalogManager
.createSample(
study.getId(), sampleName, file.getName(), null, null, null, sessionId)
.first());
}
}
} else {
throw new CatalogException(
"Can not find samples " + set + " in catalog"); // FIXME: Create missing samples??
}
}
} else {
// Get samples from file.sampleIds
queryOptions.add("id", file.getSampleIds());
sampleList = catalogManager.getAllSamples(study.getId(), queryOptions, sessionId).getResult();
}
List<Integer> sampleIdsList = new ArrayList<>(sampleList.size());
for (Sample sample : sampleList) {
sampleIdsList.add(sample.getId());
//
// sampleIdsString.append(sample.getName()).append(":").append(sample.getId()).append(",");
}
indexFileModifyParams.put("sampleIds", sampleIdsList);
return sampleList;
}
/**
* @param study Study where file is located
* @param file File to be indexed
* @param indexFile Generated index file
* @param sampleList
* @param storageEngine StorageEngine to be used
* @param outDirUri Index outdir
* @param indexFileModifyParams This map will be used to modify the indexFile
* @param dbName
* @return CommandLine
* @throws org.opencb.opencga.catalog.db.CatalogDBException
* @throws CatalogIOManagerException
*/
private String createCommandLine(
Study study,
File file,
File indexFile,
List<Sample> sampleList,
String storageEngine,
URI outDirUri,
final ObjectMap indexFileModifyParams,
final String dbName,
QueryOptions options)
throws CatalogException {
// Create command line
String userId = file.getOwnerId();
String name = file.getName();
String commandLine;
ObjectMap indexAttributes = indexFileModifyParams.get("attributes", ObjectMap.class);
String opencgaStorageBin =
Paths.get(Config.getOpenCGAHome(), "bin", OPENCGA_STORAGE_BIN_NAME).toString();
if (file.getBioformat() == File.Bioformat.ALIGNMENT
|| name.endsWith(".bam")
|| name.endsWith(".sam")) {
int chunkSize = 200; // TODO: Read from properties.
commandLine =
new StringBuilder(opencgaStorageBin)
.append(" --storage-engine ")
.append(storageEngine)
.append(" index-alignments ")
.append(" --file-id ")
.append(indexFile.getId())
.append(" --database ")
.append(dbName)
.append(" --input ")
.append(catalogManager.getFileUri(file))
.append(" --calculate-coverage ")
.append(chunkSize)
.append(" --mean-coverage ")
.append(chunkSize)
.append(" --outdir ")
.append(outDirUri)
// .append(" --credentials ")
.toString();
indexAttributes.put("chunkSize", chunkSize);
} else if (name.endsWith(".fasta") || name.endsWith(".fasta.gz")) {
throw new UnsupportedOperationException();
} else if (file.getBioformat() == File.Bioformat.VARIANT
|| name.contains(".vcf")
|| name.contains(".vcf.gz")) {
StringBuilder sampleIdsString = new StringBuilder();
for (Sample sample : sampleList) {
sampleIdsString.append(sample.getName()).append(":").append(sample.getId()).append(",");
}
StringBuilder sb =
new StringBuilder(opencgaStorageBin)
.append(" --storage-engine ")
.append(storageEngine)
.append(" index-variants ")
.append(" --file-id ")
.append(indexFile.getId())
.append(" --study-name \'")
.append(study.getName())
.append("\'")
.append(" --study-id ")
.append(study.getId())
// .append(" --study-type ").append(study.getType())
.append(" --database ")
.append(dbName)
.append(" --input ")
.append(catalogManager.getFileUri(file))
.append(" --outdir ")
.append(outDirUri)
.append(" --include-genotypes ")
.append(" --compress-genotypes ")
.append(" --include-stats ")
// .append(" --sample-ids ").append(sampleIdsString)
// .append(" --credentials ")
;
if (options.getBoolean(VariantStorageManager.ANNOTATE, true)) {
sb.append(" --annotate ");
}
if (options.getBoolean(VariantStorageManager.INCLUDE_SRC, false)) {
sb.append(" --include-src ");
}
commandLine = sb.toString();
} else {
return null;
}
indexAttributes.put(INDEXED_FILE, file.getId());
indexAttributes.put(DB_NAME, dbName);
indexAttributes.put(STORAGE_ENGINE, storageEngine);
return commandLine;
}
