private BasicDBList processGeneList(String genes) {
BasicDBList list = new BasicDBList();
Client wsRestClient = Client.create();
WebResource webResource =
wsRestClient.resource("http://ws.bioinfo.cipf.es/cellbase/rest/latest/hsa/feature/gene/");
ObjectMapper mapper = new ObjectMapper();
String response =
webResource.path(genes).path("info").queryParam("of", "json").get(String.class);
try {
JsonNode actualObj = mapper.readTree(response);
Iterator<JsonNode> it = actualObj.iterator();
Iterator<JsonNode> aux;
while (it.hasNext()) {
JsonNode node = it.next();
if (node.isArray()) {
aux = node.iterator();
while (aux.hasNext()) {
JsonNode auxNode = aux.next();
DBObject regionClause = new BasicDBObject("chr", auxNode.get("chromosome").asText());
regionClause.put(
"pos",
new BasicDBObject("$gte", auxNode.get("start").asInt())
.append("$lte", auxNode.get("end").asInt()));
list.add(regionClause);
}
}
}
} catch (IOException e) {
e.printStackTrace();
}
return list;
}
public QueryResult<VariantInfo> getRecordsMongo(
int page, int start, int limit, MutableInt count, Map<String, String> options) {
long startTime = System.currentTimeMillis();
QueryResult<VariantInfo> queryResult = new QueryResult<>();
List<VariantInfo> res = new ArrayList<>();
String sourceId = options.get("studyId");
DBCollection coll = db.getCollection("variants");
BasicDBObject elemMatch = new BasicDBObject("sourceId", sourceId);
DBObject query = new BasicDBObject();
BasicDBList orList = new BasicDBList();
Map<String, List<String>> sampleGenotypes = processSamplesGT(options);
System.out.println("map = " + options);
if (options.containsKey("region") && !options.get("region").equals("")) {
String[] regions = options.get("region").split(",");
Pattern pattern = Pattern.compile("(\\w+):(\\d+)-(\\d+)");
Matcher matcher, matcherChr;
for (int i = 0; i < regions.length; i++) {
String region = regions[i];
matcher = pattern.matcher(region);
if (matcher.find()) {
String chr = matcher.group(1);
int s = Integer.valueOf(matcher.group(2));
int e = Integer.valueOf(matcher.group(3));
DBObject regionClause = new BasicDBObject("chr", chr);
regionClause.put("pos", new BasicDBObject("$gte", s).append("$lte", e));
orList.add(regionClause);
} else {
Pattern patternChr = Pattern.compile("(\\w+)");
matcherChr = patternChr.matcher(region);
if (matcherChr.find()) {
String chr = matcherChr.group();
DBObject regionClause = new BasicDBObject("chr", chr);
orList.add(regionClause);
}
}
}
query.put("$or", orList);
} else if (options.containsKey("genes") && !options.get("genes").equals("")) {
orList = processGeneList(options.get("genes"));
if (orList.size() > 0) {
query.put("$or", orList);
} else {
queryResult.setWarningMsg("Wrong gene name");
queryResult.setResult(res);
queryResult.setNumResults(res.size());
return queryResult;
}
}
if (options.containsKey("conseq_type") && !options.get("conseq_type").equals("")) {
String[] cts = options.get("conseq_type").split(",");
BasicDBList ctList = new BasicDBList();
for (String ct : cts) {
ctList.add(ct);
}
elemMatch.put("effects", new BasicDBObject("$in", ctList));
}
if (sampleGenotypes.size() > 0) {
for (Map.Entry<String, List<String>> entry : sampleGenotypes.entrySet()) {
BasicDBList gtList = new BasicDBList();
for (String gt : entry.getValue()) {
gtList.add(gt);
}
elemMatch.put("samples." + entry.getKey() + ".GT", new BasicDBObject("$in", gtList));
}
}
if (options.containsKey("miss_gt") && !options.get("miss_gt").equalsIgnoreCase("")) {
Integer val = Integer.valueOf(options.get("miss_gt"));
Object missGt = getMongoOption(options.get("option_miss_gt"), val);
elemMatch.put("stats.missGenotypes", missGt);
}
BasicDBList andControls = new BasicDBList();
if (options.containsKey("maf_1000g_controls")
&& !options.get("maf_1000g_controls").equalsIgnoreCase("")) {
BasicDBList or = new BasicDBList();
or.add(new BasicDBObject("attributes.1000G_maf", new BasicDBObject("$exists", false)));
or.add(
new BasicDBObject(
"attributes.1000G_maf",
new BasicDBObject("$lte", options.get("maf_1000g_controls"))));
andControls.add(new BasicDBObject("$or", or));
}
if (options.containsKey("maf_1000g_afr_controls")
&& !options.get("maf_1000g_afr_controls").equalsIgnoreCase("")) {
BasicDBList or = new BasicDBList();
or.add(new BasicDBObject("attributes.1000G_AFR_maf", new BasicDBObject("$exists", false)));
or.add(
new BasicDBObject(
"attributes.1000G_AFR_maf",
new BasicDBObject("$lte", options.get("maf_1000g_afr_controls"))));
andControls.add(new BasicDBObject("$or", or));
}
if (options.containsKey("maf_1000g_asi_controls")
&& !options.get("maf_1000g_asi_controls").equalsIgnoreCase("")) {
BasicDBList or = new BasicDBList();
or.add(new BasicDBObject("attributes.1000G_ASI_maf", new BasicDBObject("$exists", false)));
or.add(
new BasicDBObject(
"attributes.1000G_ASI_maf",
new BasicDBObject("$lte", options.get("maf_1000g_asi_controls"))));
andControls.add(new BasicDBObject("$or", or));
}
if (options.containsKey("maf_1000g_eur_controls")
&& !options.get("maf_1000g_eur_controls").equalsIgnoreCase("")) {
System.out.print("EUR");
BasicDBList or = new BasicDBList();
or.add(new BasicDBObject("attributes.1000G_EUR_maf", new BasicDBObject("$exists", false)));
or.add(
new BasicDBObject(
"attributes.1000G_EUR_maf",
new BasicDBObject("$lte", options.get("maf_1000g_eur_controls"))));
andControls.add(new BasicDBObject("$or", or));
}
if (options.containsKey("maf_1000g_ame_controls")
&& !options.get("maf_1000g_ame_controls").equalsIgnoreCase("")) {
System.out.print("AME");
BasicDBList or = new BasicDBList();
or.add(new BasicDBObject("attributes.1000G_AME_maf", new BasicDBObject("$exists", false)));
or.add(
new BasicDBObject(
"attributes.1000G_AME_maf",
new BasicDBObject("$lte", options.get("maf_1000g_ame_controls"))));
andControls.add(new BasicDBObject("$or", or));
}
if (options.containsKey("maf_evs_controls")
&& !options.get("maf_evs_controls").equalsIgnoreCase("")) {
BasicDBList or = new BasicDBList();
or.add(new BasicDBObject("attributes.EVS_maf", new BasicDBObject("$exists", false)));
or.add(
new BasicDBObject(
"attributes.EVS_maf", new BasicDBObject("$lte", options.get("maf_evs_controls"))));
andControls.add(new BasicDBObject("$or", or));
}
if (options.containsKey("maf_bier_controls")
&& !options.get("maf_bier_controls").equalsIgnoreCase("")) {
BasicDBList or = new BasicDBList();
or.add(new BasicDBObject("attributes.BIER_maf", new BasicDBObject("$exists", false)));
or.add(
new BasicDBObject(
"attributes.BIER_maf", new BasicDBObject("$lte", options.get("maf_bier_controls"))));
andControls.add(new BasicDBObject("$or", or));
}
if (andControls.size() > 0) {
elemMatch.append("$and", andControls);
}
query.put("sources", new BasicDBObject("$elemMatch", elemMatch));
System.out.println("#############################");
System.out.println(query);
System.out.println("#############################");
long dbStart = System.currentTimeMillis();
DBObject sort = null;
DBCursor cursor;
if (options.containsKey("sort")) {
sort = getQuerySort(options.get("sort"));
cursor = coll.find(query).sort(sort).skip(start).limit(limit);
} else {
cursor = coll.find(query).skip(start).limit(limit);
}
count.setValue(cursor.count());
queryResult.setDbTime(dbStart - System.currentTimeMillis());
for (DBObject obj : cursor) {
BasicDBObject elem = (BasicDBObject) obj;
VariantInfo vi = new VariantInfo();
VariantStats vs = new VariantStats();
String chr = elem.getString("chr");
int pos = elem.getInt("pos");
vi.setChromosome(chr);
vi.setPosition(pos);
BasicDBList studies = (BasicDBList) elem.get("sources");
Iterator<Object> it = studies.iterator();
while (it.hasNext()) {
BasicDBObject study = (BasicDBObject) it.next();
if (study.getString("sourceId").equalsIgnoreCase(sourceId)) {
BasicDBObject stats = (BasicDBObject) study.get("stats");
String ref = study.getString("ref");
BasicDBList alt = (BasicDBList) study.get("alt");
vi.setRef(ref);
vi.setAlt(Joiner.on(",").join(alt.toArray()));
vs.setMaf((float) stats.getDouble("maf"));
vs.setMgf((float) stats.getDouble("mgf"));
vs.setMafAllele(stats.getString("alleleMaf"));
vs.setMgfAllele(stats.getString("genotypeMaf"));
vs.setMissingAlleles(stats.getInt("missAllele"));
vs.setMissingGenotypes(stats.getInt("missGenotypes"));
vs.setMendelinanErrors(stats.getInt("mendelErr"));
vs.setCasesPercentDominant((float) stats.getDouble("casesPercentDominant"));
vs.setControlsPercentDominant((float) stats.getDouble("controlsPercentDominant"));
vs.setCasesPercentRecessive((float) stats.getDouble("casesPercentRecessive"));
vs.setControlsPercentRecessive((float) stats.getDouble("controlsPercentRecessive"));
BasicDBObject samples = (BasicDBObject) study.get("samples");
for (String sampleName : samples.keySet()) {
DBObject sample = (DBObject) samples.get(sampleName);
if (sample.containsField("GT")) {
String sampleGT = (String) sample.get("GT");
vi.addSammpleGenotype(sampleName, sampleGT);
}
}
vi.setSnpid((String) study.get("snpId"));
if (study.containsField("effects")) {
BasicDBList conseqTypes = (BasicDBList) study.get("effects");
conseqTypes.remove("");
String cts = Joiner.on(",").join(conseqTypes.iterator());
vi.addConsequenceTypes(cts);
}
if (study.containsField("genes")) {
BasicDBList genesList = (BasicDBList) study.get("genes");
String genes = Joiner.on(",").join(genesList.iterator());
vi.addGenes(genes);
}
if (study.containsField("attributes")) {
BasicDBObject attr = (BasicDBObject) study.get("attributes");
if (attr.containsField("1000G_maf")) {
vi.addControl("1000G_maf", (String) attr.get("1000G_maf"));
vi.addControl("1000G_amaf", (String) attr.get("1000G_amaf"));
vi.addControl("1000G_gt", (String) attr.get("1000G_gt"));
}
if (attr.containsField("1000G_ASI_maf")) {
vi.addControl("1000G-ASI_maf", (String) attr.get("1000G_ASI_maf"));
vi.addControl("1000G-ASI_amaf", (String) attr.get("1000G_ASI_amaf"));
vi.addControl("1000G-ASI_gt", (String) attr.get("1000G_ASI_gt"));
}
if (attr.containsField("1000G_AFR_maf")) {
vi.addControl("1000G-AFR_maf", (String) attr.get("1000G_AFR_maf"));
vi.addControl("1000G-AFR_amaf", (String) attr.get("1000G_AFR_amaf"));
vi.addControl("1000G-AFR_gt", (String) attr.get("1000G_AFR_gt"));
}
if (attr.containsField("1000G_AME_maf")) {
vi.addControl("1000G-AME_maf", (String) attr.get("1000G_AME_maf"));
vi.addControl("1000G-AME_amaf", (String) attr.get("1000G_AME_amaf"));
vi.addControl("1000G-AME_gt", (String) attr.get("1000G_AME_gt"));
}
if (attr.containsField("1000G_EUR_maf")) {
vi.addControl("1000G-EUR_maf", (String) attr.get("1000G_EUR_maf"));
vi.addControl("1000G-EUR_amaf", (String) attr.get("1000G_EUR_amaf"));
vi.addControl("1000G-EUR_gt", (String) attr.get("1000G_EUR_gt"));
}
if (attr.containsField("EVS_maf")) {
vi.addControl("EVS_maf", (String) attr.get("EVS_maf"));
vi.addControl("EVS_amaf", (String) attr.get("EVS_amaf"));
vi.addControl("EVS_gt", (String) attr.get("EVS_gt"));
}
if (attr.containsField("BIER_maf")) {
vi.addControl("BIER_maf", (String) attr.get("BIER_maf"));
vi.addControl("BIER_amaf", (String) attr.get("BIER_amaf"));
vi.addControl("BIER_gt", (String) attr.get("BIER_gt"));
}
if (attr.containsField("PolyphenScore")) {
vi.setPolyphen_score(Double.parseDouble(attr.getString("PolyphenScore")));
vi.setPolyphen_effect(Integer.parseInt(attr.getString("PolyphenEffect")));
}
if (attr.containsField("SIFTScore")) {
vi.setSift_score(Double.parseDouble(attr.getString("SIFTScore")));
vi.setSift_effect(Integer.parseInt(attr.getString("SIFTEffect")));
}
}
continue;
}
}
vi.addStats(vs);
res.add(vi);
}
queryResult.setResult(res);
queryResult.setTime(startTime - System.currentTimeMillis());
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;
}