@Override
public QueryResult<Variant> getAllVariantsByRegionAndStudy(
Region region, String sourceId, QueryOptions options) {
Long start, end, dbstart, dbend;
start = System.currentTimeMillis();
QueryResult<Variant> queryResult =
new QueryResult<>(
String.format("%s:%d-%d", region.getChromosome(), region.getStart(), region.getEnd()));
List<Variant> results = new LinkedList<>();
boolean includeSamples;
boolean includeStats;
boolean includeEffects;
if (!options.containsKey("samples")
&& !options.containsKey("stats")
&& !options.containsKey("effects")) {
includeSamples = true;
includeStats = true;
includeEffects = true;
} else {
includeSamples = options.containsKey("samples") && options.getBoolean("samples");
includeStats = options.containsKey("stats") && options.getBoolean("stats");
includeEffects = options.containsKey("effects") && options.getBoolean("effects");
}
try {
String startRow = buildRowkey(region.getChromosome(), Long.toString(region.getStart()));
String stopRow = buildRowkey(region.getChromosome(), Long.toString(region.getEnd()));
HTable table = new HTable(admin.getConfiguration(), tableName);
dbstart = System.currentTimeMillis();
Scan regionScan = new Scan(startRow.getBytes(), stopRow.getBytes());
ResultScanner scanres = table.getScanner(regionScan);
dbend = System.currentTimeMillis();
queryResult.setDbTime(dbend - dbstart);
// Iterate over results and, optionally, their samples and statistics
for (Result result : scanres) {
String[] rowkeyParts = new String(result.getRow(), CHARSET_UTF_8).split("_");
String chromosome = rowkeyParts[0].replaceFirst("^0+(?!$)", "");
int position = Integer.parseInt(rowkeyParts[1]);
// Get basic result fields from Protocol Buffers message
NavigableMap<byte[], byte[]> infoMap = result.getFamilyMap("i".getBytes());
byte[] byteInfo = infoMap.get((sourceId + "_data").getBytes());
VariantFieldsProtos.VariantInfo protoInfo =
VariantFieldsProtos.VariantInfo.parseFrom(byteInfo);
String reference = protoInfo.getReference();
String alternate = StringUtils.join(protoInfo.getAlternateList(), ",");
String format = StringUtils.join(protoInfo.getFormatList(), ":");
Variant variant = new Variant(chromosome, position, position, reference, alternate);
// Set samples if requested
if (includeSamples) {
NavigableMap<byte[], byte[]> sampleMap = result.getFamilyMap("d".getBytes());
Map<String, Map<String, String>> resultSampleMap = new HashMap<>();
// Set samples
for (byte[] s : sampleMap.keySet()) {
String sampleName = (new String(s, CHARSET_UTF_8)).replaceAll(sourceId + "_", "");
VariantFieldsProtos.VariantSample sample =
VariantFieldsProtos.VariantSample.parseFrom(sampleMap.get(s));
String sample1 = sample.getSample();
String[] values = sample1.split(":");
String[] fields = format.split(":");
Map<String, String> singleSampleMap = new HashMap<>();
for (int i = 0; i < fields.length; i++) {
singleSampleMap.put(fields[i], values[i]);
}
// TODO
// variant.addSampleData(sampleName, singleSampleMap);
}
}
// Set stats if requested
if (includeStats) {
byte[] byteStats = infoMap.get((sourceId + "_stats").getBytes());
VariantFieldsProtos.VariantStats protoStats =
VariantFieldsProtos.VariantStats.parseFrom(byteStats);
VariantStats variantStats =
new VariantStats(
chromosome,
position,
reference,
alternate,
protoStats.getMaf(),
protoStats.getMgf(),
protoStats.getMafAllele(),
protoStats.getMgfGenotype(),
protoStats.getMissingAlleles(),
protoStats.getMissingGenotypes(),
protoStats.getMendelianErrors(),
protoStats.getIsIndel(),
protoStats.getCasesPercentDominant(),
protoStats.getControlsPercentDominant(),
protoStats.getCasesPercentRecessive(),
protoStats.getControlsPercentRecessive());
variant.setStats(variantStats);
}
if (includeEffects) {
QueryResult<VariantEffect> queryEffects = getEffectsByVariant(variant, options);
variant.setEffect(queryEffects.getResult());
}
results.add(variant);
}
} catch (IOException e) {
System.err.println(e.getClass().getName() + ": " + e.getMessage());
}
queryResult.setResult(results);
queryResult.setNumResults(results.size());
end = System.currentTimeMillis();
queryResult.setTime(end - start);
return queryResult;
}
public QueryResult<File> index(
int fileId, int outDirId, String storageEngine, String sessionId, QueryOptions options)
throws IOException, CatalogException, AnalysisExecutionException {
if (options == null) {
options = new QueryOptions();
}
final boolean execute = options.getBoolean(AnalysisJobExecuter.EXECUTE);
final boolean simulate = options.getBoolean(AnalysisJobExecuter.SIMULATE);
final boolean recordOutput = options.getBoolean(AnalysisJobExecuter.RECORD_OUTPUT);
final long start = System.currentTimeMillis();
/** Query catalog for user data. * */
String userId = catalogManager.getUserIdBySessionId(sessionId);
File file = catalogManager.getFile(fileId, sessionId).first();
File outDir = catalogManager.getFile(outDirId, sessionId).first();
int studyIdByOutDirId = catalogManager.getStudyIdByFileId(outDirId);
Study study = catalogManager.getStudy(studyIdByOutDirId, sessionId).getResult().get(0);
if (file.getType() != File.Type.FILE) {
throw new CatalogException(
"Expected file type = " + File.Type.FILE + " instead of " + file.getType());
}
final String dbName;
if (options.containsKey(DB_NAME)) {
dbName = options.getString(DB_NAME);
} else {
if (study.getAttributes().containsKey(DB_NAME)
&& study.getAttributes().get(DB_NAME) != null) {
dbName = study.getAttributes().get(DB_NAME).toString();
} else {
int projectId = catalogManager.getProjectIdByStudyId(study.getId());
String alias =
catalogManager
.getProject(projectId, new QueryOptions("include", "alias"), sessionId)
.first()
.getAlias();
dbName =
Config.getAnalysisProperties()
.getProperty(OPENCGA_ANALYSIS_STORAGE_DATABASE_PREFIX, "opencga_")
+ userId
+ "_"
+ alias;
}
}
// TODO: Check if file can be indexed
// ObjectMap to fill with modifications over the indexed file (like new attributes or jobId)
ObjectMap indexFileModifyParams = new ObjectMap("attributes", new ObjectMap());
/** Create temporal Job Outdir * */
final URI temporalOutDirUri;
final String randomString = "I_" + StringUtils.randomString(10);
if (simulate) {
temporalOutDirUri = createSimulatedOutDirUri(randomString);
} else {
temporalOutDirUri =
catalogManager.createJobOutDir(studyIdByOutDirId, randomString, sessionId);
}
List<Sample> sampleList;
/** Create index file* */
final File index;
if (options.containsKey(INDEX_FILE_ID)) {
logger.debug("Using an existing indexedFile.");
int indexFileId = options.getInt(INDEX_FILE_ID);
index = catalogManager.getFile(indexFileId, sessionId).first();
if (index.getType() != File.Type.INDEX) {
throw new CatalogException("Expected {type: INDEX} in IndexedFile " + indexFileId);
}
if (index.getStatus() != File.Status.READY) {
throw new CatalogException("Expected {status: READY} in IndexedFile " + indexFileId);
}
if (simulate) {
index.setStatus(File.Status.INDEXING);
} else {
ObjectMap parameters = new ObjectMap("status", File.Status.INDEXING);
catalogManager.modifyFile(index.getId(), parameters, sessionId);
}
/** Get file samples * */
sampleList =
catalogManager
.getAllSamples(study.getId(), new QueryOptions("id", index.getSampleIds()), sessionId)
.getResult();
} else {
/** Get file samples * */
sampleList = getFileSamples(study, file, indexFileModifyParams, simulate, options, sessionId);
String indexedFileDescription = "Indexation of " + file.getName() + " (" + fileId + ")";
String indexedFileName = file.getName() + "." + storageEngine;
String indexedFilePath = Paths.get(outDir.getPath(), indexedFileName).toString();
if (simulate) {
index =
new File(
-10,
indexedFileName,
File.Type.INDEX,
file.getFormat(),
file.getBioformat(),
indexedFilePath,
userId,
TimeUtils.getTime(),
indexedFileDescription,
File.Status.INDEXING,
-1,
-1,
null,
-1,
null,
null,
new HashMap<String, Object>());
} else {
index =
catalogManager
.createFile(
studyIdByOutDirId,
File.Type.INDEX,
file.getFormat(),
file.getBioformat(),
indexedFilePath,
null,
null,
indexedFileDescription,
File.Status.INDEXING,
0,
-1,
null,
-1,
null,
null,
false,
null,
sessionId)
.first();
}
}
/** Create commandLine * */
String commandLine =
createCommandLine(
study,
file,
index,
sampleList,
storageEngine,
temporalOutDirUri,
indexFileModifyParams,
dbName,
options);
if (options.containsKey(PARAMETERS)) {
List<String> extraParams = options.getAsStringList(PARAMETERS);
for (String extraParam : extraParams) {
commandLine += " " + extraParam;
}
}
/** Create job * */
ObjectMap jobResourceManagerAttributes = new ObjectMap();
jobResourceManagerAttributes.put(Job.TYPE, Job.Type.INDEX);
jobResourceManagerAttributes.put(Job.INDEXED_FILE_ID, index.getId());
String jobName = "index";
String jobDescription = "Indexing file " + file.getName() + " (" + fileId + ")";
final Job job =
AnalysisJobExecuter.createJob(
catalogManager,
studyIdByOutDirId,
jobName,
OPENCGA_STORAGE_BIN_NAME,
jobDescription,
outDir,
Collections.<Integer>emptyList(),
sessionId,
randomString,
temporalOutDirUri,
commandLine,
execute,
simulate,
recordOutput,
jobResourceManagerAttributes)
.first();
if (simulate) {
index.getAttributes().put("job", job);
// index.getAttributes().putAll(indexFileModifyParams.getMap("attributes"));
index.setSampleIds(indexFileModifyParams.getAsIntegerList("sampleIds"));
// VariantSource variantSource = (VariantSource)
// index.getAttributes().get("variantSource");
// for (Map.Entry<String, Integer> entry :
// variantSource.getSamplesPosition().entrySet()) {
// System.out.println("entry.getKey() = " + entry.getKey());
// System.out.println("entry.getValue() = " + entry.getValue());
// }
// for (String s : variantSource.getSamples()) {
// System.out.println("sample = " + s);
// }
// variantSource.setSamplesPosition(new HashMap<String, Integer>());
return new QueryResult<>(
"indexFile",
(int) (System.currentTimeMillis() - start),
1,
1,
"",
"",
Collections.singletonList(index));
} else {
/** Update IndexFile to add extra information (jobId, sampleIds, attributes, ...) * */
indexFileModifyParams.put("jobId", job.getId());
Set<Integer> jobIds;
try {
jobIds = new HashSet<>(new ObjectMap(index.getAttributes()).getAsIntegerList("jobIds"));
} catch (Exception ignore) {
jobIds = new HashSet<>(1);
}
if (index.getJobId() > 0) {
jobIds.add(index.getJobId());
}
jobIds.add(job.getId());
indexFileModifyParams.getMap("attributes").put("jobIds", jobIds);
catalogManager.modifyFile(index.getId(), indexFileModifyParams, sessionId).getResult();
return new QueryResult<>(
"indexFile",
(int) (System.currentTimeMillis() - start),
1,
1,
"",
"",
catalogManager.getFile(index.getId(), sessionId).getResult());
}
}
public QueryResult getSimpleVariantsByRegion(
Region region, String sourceId, QueryOptions options) {
Long start, end, dbstart, dbend;
start = System.currentTimeMillis();
boolean includeStats;
boolean includeEffects;
if (!options.containsKey("stats") && !options.containsKey("effects")) {
includeStats = true;
includeEffects = true;
} else {
includeStats = options.containsKey("stats") && options.getBoolean("stats");
includeEffects = options.containsKey("effects") && options.getBoolean("effects");
}
QueryResult<Variant> queryResult =
new QueryResult<>(
String.format("%s:%d-%d", region.getChromosome(), region.getStart(), region.getEnd()));
List<Variant> results = new ArrayList<>();
String startRow = buildRowkey(region.getChromosome(), Long.toString(region.getStart()));
String stopRow = buildRowkey(region.getChromosome(), Long.toString(region.getEnd()));
BasicDBObject query =
new BasicDBObject("position", new BasicDBObject("$gte", startRow).append("$lte", stopRow))
.append("sources.sourceId", sourceId);
DBCollection collection = db.getCollection("variants");
dbstart = System.currentTimeMillis();
DBCursor variantInStudies = collection.find(query);
dbend = System.currentTimeMillis();
queryResult.setDbTime(dbend - dbstart);
for (DBObject result : variantInStudies) {
String[] rowkeyParts = result.get("position").toString().split("_");
String chromosome = rowkeyParts[0].replaceFirst("^0+(?!$)", "");
int position = Integer.parseInt(rowkeyParts[1]);
BasicDBList studies = (BasicDBList) result.get("sources");
BasicDBObject st = (BasicDBObject) studies.get(0);
String ref = (String) st.get("ref");
String alt = StringUtils.join((ArrayList<String>) st.get("alt"), ",");
// TODO Needs rework
Variant variant = new Variant(chromosome, position, position, ref, alt);
// Set stats informations
if (includeStats) {
VariantStats stats = new VariantStats();
BasicDBObject mongoStats = (BasicDBObject) st.get("stats");
stats.setMaf((float) (double) mongoStats.get("maf"));
stats.setMafAllele((String) mongoStats.get("alleleMaf"));
stats.setMissingGenotypes((int) mongoStats.get("missing"));
List<Genotype> genotypeCount = new ArrayList<>();
for (BasicDBObject s : (List<BasicDBObject>) mongoStats.get("genotypeCount")) {
for (Map.Entry<String, Object> entry : s.entrySet()) {
Genotype genotype = new Genotype(entry.getKey());
genotype.setCount((Integer) entry.getValue());
genotypeCount.add(genotype);
}
}
stats.setGenotypes(genotypeCount);
variant.setStats(stats);
}
// TODO Set consequence type names
if (includeEffects) {
BasicDBList mongoEffects = (BasicDBList) st.get("effects");
if (mongoEffects != null) {
for (Object e : mongoEffects) {
String effectObo = e.toString();
VariantEffect effect = new VariantEffect();
effect.setConsequenceTypeObo(effectObo);
variant.addEffect(effect);
}
}
}
results.add(variant);
}
queryResult.setResult(results);
queryResult.setNumResults(results.size());
end = System.currentTimeMillis();
queryResult.setTime(end - start);
return queryResult;
}