public List<Variant> getRecordSimpleStats(
String study, int missing_gt, float maf, String maf_allele) {
BasicDBObject compare =
new BasicDBObject("studies.stats.allele_maf", maf_allele)
.append("studies.stats.MAF", maf)
.append("studies.stats.missing", missing_gt);
List<Get> hbaseQuery = new ArrayList<>();
DBCollection collection = db.getCollection("variants");
Iterator<DBObject> result = collection.find(compare);
String chromosome = new String();
while (result.hasNext()) {
DBObject variant = result.next();
String position = variant.get("_id").toString();
// hbase query construction
Get get = new Get(position.getBytes());
hbaseQuery.add(get);
}
// Complete results, from HBase
tableName = study;
effectTableName = tableName + "effect";
Map<String, Variant> resultsMap = new HashMap<>();
try {
HTable table = new HTable(admin.getConfiguration(), tableName);
HTable effectTable = new HTable(admin.getConfiguration(), effectTableName);
Result[] hbaseResultEffect = effectTable.get(hbaseQuery);
Result[] hbaseResultStats = table.get(hbaseQuery);
// List<Variant> results = new LinkedList<>();
for (Result r : hbaseResultStats) {
String position = new String(r.getRow(), CHARSET_UTF_8);
String[] aux = position.split("_");
String inner_position = aux[1];
String chr = aux[0];
// position parsing
if (chr.startsWith("0")) {
chr = chr.substring(1);
}
while (inner_position.startsWith("0")) {
inner_position = inner_position.substring(1);
}
List<VariantFieldsProtos.VariantSample> samples = new LinkedList<>();
NavigableMap<byte[], byte[]> infoMap = r.getFamilyMap("i".getBytes());
byte[] byteStats = infoMap.get((study + "_stats").getBytes());
VariantFieldsProtos.VariantStats stats =
VariantFieldsProtos.VariantStats.parseFrom(byteStats);
byte[] byteInfo = infoMap.get((study + "_data").getBytes());
VariantFieldsProtos.VariantInfo info = VariantFieldsProtos.VariantInfo.parseFrom(byteInfo);
String alternate = StringUtils.join(info.getAlternateList(), ", ");
String reference = info.getReference();
Variant partialResult =
new Variant(
chr,
Integer.parseInt(inner_position),
Integer.parseInt(inner_position),
reference,
alternate);
String format = StringUtils.join(info.getFormatList(), ":");
NavigableMap<byte[], byte[]> sampleMap = r.getFamilyMap("d".getBytes());
Map<String, Map<String, String>> resultSampleMap = new HashMap<>();
// StringBuilder sampleRaw = new StringBuilder();
for (byte[] s : sampleMap.keySet()) {
String qual = (new String(s, CHARSET_UTF_8)).replaceAll(study + "_", "");
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]);
}
resultSampleMap.put(qual, singleSampleMap);
}
VariantStats variantStats =
new VariantStats(
chromosome,
Integer.parseInt(inner_position),
reference,
alternate,
stats.getMaf(),
stats.getMgf(),
stats.getMafAllele(),
stats.getMgfGenotype(),
stats.getMissingAlleles(),
stats.getMissingGenotypes(),
stats.getMendelianErrors(),
stats.getIsIndel(),
stats.getCasesPercentDominant(),
stats.getControlsPercentDominant(),
stats.getCasesPercentRecessive(),
stats.getControlsPercentRecessive());
partialResult.setStats(variantStats);
resultsMap.put(new String(r.getRow(), CHARSET_UTF_8), partialResult);
}
for (Result r : hbaseResultEffect) {
if (!r.isEmpty()) {
NavigableMap<byte[], byte[]> effectMap = r.getFamilyMap("e".getBytes());
Variant partialResult = resultsMap.get(new String(r.getRow(), CHARSET_UTF_8));
System.out.println("Recuperado " + partialResult.toString());
String s = partialResult.getReference() + "_" + partialResult.getAlternate();
VariantEffectProtos.EffectInfo effectInfo =
VariantEffectProtos.EffectInfo.parseFrom(effectMap.get(s.getBytes()));
VariantEffect variantEffect =
new VariantEffect(
partialResult.getChromosome(),
(int) partialResult.getStart(),
partialResult.getReference(),
partialResult.getAlternate(),
effectInfo.getFeatureId(),
effectInfo.getFeatureName(),
effectInfo.getFeatureType(),
effectInfo.getFeatureBiotype(),
effectInfo.getFeatureChromosome(),
effectInfo.getFeatureStart(),
effectInfo.getFeatureEnd(),
effectInfo.getFeatureStrand(),
effectInfo.getSnpId(),
effectInfo.getAncestral(),
effectInfo.getAlternative(),
effectInfo.getGeneId(),
effectInfo.getTranscriptId(),
effectInfo.getGeneName(),
effectInfo.getConsequenceType(),
effectInfo.getConsequenceTypeObo(),
effectInfo.getConsequenceTypeDesc(),
effectInfo.getConsequenceTypeType(),
effectInfo.getAaPosition(),
effectInfo.getAminoacidChange(),
effectInfo.getCodonChange());
resultsMap.put(new String(r.getRow(), CHARSET_UTF_8), partialResult);
}
}
} catch (InvalidProtocolBufferException e) {
System.err.println(e.getClass().getName() + ": " + e.getMessage());
} catch (IOException e) {
System.err.println(e.getClass().getName() + ": " + e.getMessage());
}
List<Variant> results = new ArrayList<>(resultsMap.values());
return results;
}
@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 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;
}