/**
* Given an HttpServletRequest, determine all cancer_study_ids associated with it. cancer study
* identifiers can be inferred from profile_ids, case_list_ids, or case_ids. this returns the set
* of ALL POSSIBLE cancer study identifiers
*
* @param request
* @return the cancer_study_ids associated with the request, which will be empty if none can be
* determined; or empty set if a problem arises.
* @throws DaoException
* @throws ProtocolException
*/
public static HashSet<String> getCancerStudyIDs(HttpServletRequest request)
throws DaoException, ProtocolException {
HashSet<String> cancerStudies = new HashSet<String>();
// a CANCER_STUDY_ID is explicitly provided, as in getGeneticProfiles, getCaseLists, etc.
// make sure the cancer_study_id provided in the request points to a real study
String studyIDstring = getCancerStudyId(request);
if (studyIDstring != null) {
if (DaoCancerStudy.doesCancerStudyExistByStableId(studyIDstring)) {
cancerStudies.add(studyIDstring);
}
return cancerStudies;
}
// a genetic_profile_id is explicitly provided, as in getProfileData
if (null != request.getParameter(WebService.GENETIC_PROFILE_ID)) {
ArrayList<String> geneticProfileIds = getGeneticProfileId(request);
for (String geneticProfileId : geneticProfileIds) {
// that's the point of this code??
// if (geneticProfileId == null) {
// return cancerStudies;
// }
GeneticProfile aGeneticProfile =
DaoGeneticProfile.getGeneticProfileByStableId(geneticProfileId);
if (aGeneticProfile != null
&& DaoCancerStudy.doesCancerStudyExistByInternalId(
aGeneticProfile.getCancerStudyId())) {
cancerStudies.add(
DaoCancerStudy.getCancerStudyByInternalId(aGeneticProfile.getCancerStudyId())
.getCancerStudyStableId());
}
}
return cancerStudies;
}
// a patient_set_id is explicitly provided, as in getProfileData, getMutationData,
// getClinicalData, etc.
String sampleSetId = request.getParameter(WebService.CASE_SET_ID);
if (sampleSetId != null) {
DaoSampleList aDaoSampleList = new DaoSampleList();
SampleList aSampleList = aDaoSampleList.getSampleListByStableId(sampleSetId);
if (aSampleList != null
&& DaoCancerStudy.doesCancerStudyExistByInternalId(aSampleList.getCancerStudyId())) {
cancerStudies.add(
DaoCancerStudy.getCancerStudyByInternalId(aSampleList.getCancerStudyId())
.getCancerStudyStableId());
}
return cancerStudies;
}
return cancerStudies;
}
@Before
public void setUp() throws DaoException {
DaoCancerStudy.reCacheAll();
DaoGeneOptimized.getInstance().reCache();
ProgressMonitor.resetWarnings();
studyId = DaoCancerStudy.getCancerStudyByStableId("study_tcga_pub").getInternalId();
GeneticProfile newGeneticProfile = new GeneticProfile();
newGeneticProfile.setCancerStudyId(studyId);
newGeneticProfile.setGeneticAlterationType(GeneticAlterationType.COPY_NUMBER_ALTERATION);
newGeneticProfile.setStableId("study_tcga_pub_test");
newGeneticProfile.setProfileName("Barry CNA Results");
newGeneticProfile.setDatatype("test");
DaoGeneticProfile.addGeneticProfile(newGeneticProfile);
geneticProfileId =
DaoGeneticProfile.getGeneticProfileByStableId("study_tcga_pub_test").getGeneticProfileId();
sample1 =
DaoSample.getSampleByCancerStudyAndSampleId(studyId, "TCGA-A1-A0SB-01").getInternalId();
sample2 =
DaoSample.getSampleByCancerStudyAndSampleId(studyId, "TCGA-A1-A0SD-01").getInternalId();
sample3 =
DaoSample.getSampleByCancerStudyAndSampleId(studyId, "TCGA-A1-A0SE-01").getInternalId();
sample4 =
DaoSample.getSampleByCancerStudyAndSampleId(studyId, "TCGA-A1-A0SF-01").getInternalId();
sample5 =
DaoSample.getSampleByCancerStudyAndSampleId(studyId, "TCGA-A1-A0SG-01").getInternalId();
}
private void runImportRnaData1() throws DaoException, IOException {
DaoGeneOptimized daoGene = DaoGeneOptimized.getInstance();
DaoGeneticAlteration dao = DaoGeneticAlteration.getInstance();
daoGene.addGene(new CanonicalGene(999999780, "A"));
daoGene.addGene(new CanonicalGene(999995982, "B"));
daoGene.addGene(new CanonicalGene(999993310, "C"));
daoGene.addGene(new CanonicalGene(999997849, "D"));
daoGene.addGene(new CanonicalGene(999992978, "E"));
daoGene.addGene(new CanonicalGene(999997067, "F"));
daoGene.addGene(new CanonicalGene(999911099, "G"));
daoGene.addGene(new CanonicalGene(999999675, "6352"));
GeneticProfile geneticProfile = new GeneticProfile();
geneticProfile.setCancerStudyId(studyId);
geneticProfile.setStableId("gbm_mrna");
geneticProfile.setGeneticAlterationType(GeneticAlterationType.MRNA_EXPRESSION);
geneticProfile.setDatatype("CONTINUOUS");
geneticProfile.setProfileName("MRNA Data");
geneticProfile.setProfileDescription("mRNA Data");
DaoGeneticProfile.addGeneticProfile(geneticProfile);
int newGeneticProfileId =
DaoGeneticProfile.getGeneticProfileByStableId("gbm_mrna").getGeneticProfileId();
ProgressMonitor.setConsoleMode(true);
// TBD: change this to use getResourceAsStream()
File file = new File("src/test/resources/mrna_test.txt");
ImportTabDelimData parser = new ImportTabDelimData(file, newGeneticProfileId, null);
int numLines = FileUtil.getNumLines(file);
parser.importData(numLines);
ConsoleUtil.showMessages();
int sampleId = DaoSample.getSampleByCancerStudyAndSampleId(studyId, "DD639").getInternalId();
String value = dao.getGeneticAlteration(newGeneticProfileId, sampleId, 999992978);
assertEquals("2.01", value);
sampleId = DaoSample.getSampleByCancerStudyAndSampleId(studyId, "DD638").getInternalId();
value = dao.getGeneticAlteration(newGeneticProfileId, sampleId, 999997849);
assertEquals("0.55", value);
}
private void processRequestStatistics(
HttpServletRequest request, HttpServletResponse response, String type)
throws ServletException, IOException {
String cancerStudyIds = request.getParameter(QueryBuilder.CANCER_STUDY_ID);
try {
boolean includeMut = "mut".equalsIgnoreCase(type);
boolean includeCna = "cna".equalsIgnoreCase(type);
Map<String, Map<String, Object>> data = new HashMap<String, Map<String, Object>>();
// get list of cancer studies
AccessControl accessControl = getaccessControl();
List<CancerStudy> cancerStudies;
if (cancerStudyIds == null || cancerStudyIds.isEmpty()) {
cancerStudies = accessControl.getCancerStudies();
} else {
cancerStudies = new ArrayList<CancerStudy>();
for (String studyId : cancerStudyIds.split("[ ,]+")) {
CancerStudy study = DaoCancerStudy.getCancerStudyByStableId(studyId);
if (study != null && !accessControl.isAccessibleCancerStudy(studyId).isEmpty()) {
cancerStudies.add(study);
}
}
}
for (CancerStudy cancerStudy : cancerStudies) {
if (cancerStudy.getCancerStudyStableId().equalsIgnoreCase("all")) {
continue;
}
Map<String, Object> row = new HashMap<String, Object>();
data.put(cancerStudy.getCancerStudyStableId(), row);
if (!includeMut && !includeMut) {
row.put("name", cancerStudy.getName());
String pmid = cancerStudy.getPmid();
if (pmid != null) {
row.put("pmid", pmid);
}
String citation = cancerStudy.getCitation();
if (citation != null) {
row.put("citation", citation);
}
row.put(
"cases", DaoPatient.getPatientsByCancerStudyId(cancerStudy.getInternalId()).size());
}
if (includeMut) {
GeneticProfile mutProfile = cancerStudy.getMutationProfile();
if (mutProfile == null) {
row.put("mut", 0);
} else {
int mutEvents = DaoMutation.countMutationEvents(mutProfile.getGeneticProfileId());
int samplesWithMut =
DaoSampleProfile.countSamplesInProfile(mutProfile.getGeneticProfileId());
row.put("mut", 1.0 * mutEvents / samplesWithMut);
}
}
if (includeCna) {
GeneticProfile cnaProfile = cancerStudy.getCopyNumberAlterationProfile(false);
if (cnaProfile == null) {
row.put("cna", 0);
} else {
List<Integer> samples =
DaoSampleProfile.getAllSampleIdsInProfile(cnaProfile.getGeneticProfileId());
Map<Integer, Double> fracs =
DaoCopyNumberSegment.getCopyNumberActeredFraction(
samples,
cnaProfile.getCancerStudyId(),
GlobalProperties.getPatientViewGenomicOverviewCnaCutoff()[0]);
double aveFrac = 0;
for (double frac : fracs.values()) {
aveFrac += frac;
}
aveFrac /= samples.size();
row.put("cna", aveFrac);
}
}
}
response.setContentType("application/json");
PrintWriter out = response.getWriter();
try {
JSONValue.writeJSONString(data, out);
} finally {
out.close();
}
} catch (DaoException ex) {
throw new ServletException(ex);
} catch (ProtocolException ex) {
throw new ServletException(ex);
}
}
/**
* Test importing of data_rppa file.
*
* @throws Exception All Errors.
*/
@Test
public void testImportRppaData() throws Exception {
MySQLbulkLoader.bulkLoadOn();
DaoGeneOptimized daoGene = DaoGeneOptimized.getInstance();
DaoGeneticAlteration dao = DaoGeneticAlteration.getInstance();
// Genes with alias:
daoGene.addGene(makeGeneWithAlias(999999931, "TESTACACA", "TESTACC1"));
daoGene.addGene(makeGeneWithAlias(999999207, "TESTAKT1", "TESTAKT"));
daoGene.addGene(makeGeneWithAlias(999999597, "TESTSANDER", "TESTACC1"));
daoGene.addGene(makeGeneWithAlias(999997158, "TESTTP53BP1", "TEST53BP1"));
// test for NA being a special case in RPPA, and not the usual alias
daoGene.addGene(makeGeneWithAlias(999997504, "XK", "NA"));
// Other genes:
daoGene.addGene(new CanonicalGene(999999932, "TESTACACB"));
daoGene.addGene(new CanonicalGene(999999208, "TESTAKT2"));
daoGene.addGene(new CanonicalGene(999999369, "TESTARAF"));
daoGene.addGene(new CanonicalGene(999991978, "TESTEIF4EBP1"));
daoGene.addGene(new CanonicalGene(999995562, "TESTPRKAA1"));
daoGene.addGene(new CanonicalGene(999997531, "TESTYWHAE"));
daoGene.addGene(new CanonicalGene(999910000, "TESTAKT3"));
daoGene.addGene(new CanonicalGene(999995578, "TESTPRKCA"));
GeneticProfile geneticProfile = new GeneticProfile();
geneticProfile.setCancerStudyId(studyId);
geneticProfile.setStableId("gbm_rppa");
geneticProfile.setGeneticAlterationType(GeneticAlterationType.PROTEIN_LEVEL);
geneticProfile.setDatatype("LOG2-VALUE");
geneticProfile.setProfileName("RPPA Data");
geneticProfile.setProfileDescription("RPPA Data");
DaoGeneticProfile.addGeneticProfile(geneticProfile);
int newGeneticProfileId =
DaoGeneticProfile.getGeneticProfileByStableId("gbm_rppa").getGeneticProfileId();
ProgressMonitor.setConsoleMode(true);
// TBD: change this to use getResourceAsStream()
File file = new File("src/test/resources/tabDelimitedData/data_rppa.txt");
ImportTabDelimData parser = new ImportTabDelimData(file, newGeneticProfileId, null);
int numLines = FileUtil.getNumLines(file);
parser.importData(numLines);
ConsoleUtil.showMessages();
int sampleId = DaoSample.getSampleByCancerStudyAndSampleId(studyId, "SAMPLE1").getInternalId();
String value = dao.getGeneticAlteration(newGeneticProfileId, sampleId, 999997531);
assertEquals("1.5", value);
sampleId = DaoSample.getSampleByCancerStudyAndSampleId(studyId, "SAMPLE4").getInternalId();
value = dao.getGeneticAlteration(newGeneticProfileId, sampleId, 999997531);
assertEquals("2", value);
sampleId = DaoSample.getSampleByCancerStudyAndSampleId(studyId, "SAMPLE4").getInternalId();
value = dao.getGeneticAlteration(newGeneticProfileId, sampleId, 999997504);
assertEquals(
"NaN",
value); // "NA" is not expected to be stored because of workaround for bug in firehose. See
// also https://github.com/cBioPortal/cbioportal/issues/839#issuecomment-203523078
sampleId = DaoSample.getSampleByCancerStudyAndSampleId(studyId, "SAMPLE1").getInternalId();
value = dao.getGeneticAlteration(newGeneticProfileId, sampleId, 999995578);
assertEquals("1.5", value);
}
/**
* Test importing of data_expression file.
*
* @throws Exception All Errors.
*/
@Test
public void testImportmRnaData2() throws Exception {
MySQLbulkLoader.bulkLoadOn();
DaoGeneOptimized daoGene = DaoGeneOptimized.getInstance();
DaoGeneticAlteration dao = DaoGeneticAlteration.getInstance();
// Gene with alias:
daoGene.addGene(makeGeneWithAlias(999997504, "TESTXK", "NA"));
// Other genes:
daoGene.addGene(new CanonicalGene(999999999, "TESTNAT1"));
daoGene.addGene(new CanonicalGene(999997124, "TESTTNF"));
daoGene.addGene(new CanonicalGene(999991111, "TESTCHEK1"));
daoGene.addGene(new CanonicalGene(999999919, "TESTABCA1"));
// will get generated negative id:
daoGene.addGene(new CanonicalGene(-1, "TESTphosphoprotein"));
GeneticProfile geneticProfile = new GeneticProfile();
geneticProfile.setCancerStudyId(studyId);
geneticProfile.setStableId("gbm_mrna");
geneticProfile.setGeneticAlterationType(GeneticAlterationType.MRNA_EXPRESSION);
geneticProfile.setDatatype("CONTINUOUS");
geneticProfile.setProfileName("MRNA Data");
geneticProfile.setProfileDescription("mRNA Data");
DaoGeneticProfile.addGeneticProfile(geneticProfile);
int newGeneticProfileId =
DaoGeneticProfile.getGeneticProfileByStableId("gbm_mrna").getGeneticProfileId();
ProgressMonitor.setConsoleMode(true);
// TBD: change this to use getResourceAsStream()
File file = new File("src/test/resources/tabDelimitedData/data_expression2.txt");
ImportTabDelimData parser = new ImportTabDelimData(file, newGeneticProfileId, null);
int numLines = FileUtil.getNumLines(file);
parser.importData(numLines);
// check if expected warnings are given:
ArrayList<String> warnings = ProgressMonitor.getWarnings();
int countDuplicatedRowWarnings = 0;
int countInvalidEntrez = 0;
int countSkippedWarnings = 0;
for (String warning : warnings) {
if (warning.contains("Duplicated row")) {
countDuplicatedRowWarnings++;
}
if (warning.contains("invalid Entrez_Id")) {
// invalid Entrez
countInvalidEntrez++;
}
if (warning.contains("Record will be skipped")) {
// Entrez is a valid number, but not found
countSkippedWarnings++;
}
}
// check that we have 11 warning messages:
assertEquals(2, countDuplicatedRowWarnings);
assertEquals(3, countInvalidEntrez);
assertEquals(6, countSkippedWarnings);
Set<Long> entrezGeneIds = DaoGeneticAlteration.getGenesIdInProfile(newGeneticProfileId);
// data will be loaded for 5 of the genes
assertEquals(5, entrezGeneIds.size());
HashMap<Long, HashMap<Integer, String>> dataMap =
dao.getGeneticAlterationMap(newGeneticProfileId, entrezGeneIds);
assertEquals(5, dataMap.entrySet().size());
int sampleId = DaoSample.getSampleByCancerStudyAndSampleId(studyId, "SAMPLE1").getInternalId();
String value = dao.getGeneticAlteration(newGeneticProfileId, sampleId, 999997124);
assertEquals("770", value);
sampleId = DaoSample.getSampleByCancerStudyAndSampleId(studyId, "SAMPLE3").getInternalId();
value = dao.getGeneticAlteration(newGeneticProfileId, sampleId, 999997124);
assertEquals("220", value);
// gene should also be loaded via its alias "NA" as defined above:
sampleId = DaoSample.getSampleByCancerStudyAndSampleId(studyId, "SAMPLE3").getInternalId();
value = dao.getGeneticAlteration(newGeneticProfileId, sampleId, 999997504);
assertEquals("9940", value);
}
/**
* Processes requests for both HTTP <code>GET</code> and <code>POST</code> methods.
*
* @param request servlet request
* @param response servlet response
* @throws javax.servlet.ServletException if a servlet-specific error occurs
* @throws java.io.IOException if an I/O error occurs
*/
protected void doGet(HttpServletRequest request, HttpServletResponse response)
throws IOException, ServletException {
XDebug xdebug = new XDebug();
xdebug.startTimer();
response.setContentType("application/json");
PrintWriter writer = response.getWriter();
try {
List resultsList = new LinkedList();
// Get the gene list
String geneList = request.getParameter(QueryBuilder.GENE_LIST);
if (request instanceof XssRequestWrapper) {
geneList = ((XssRequestWrapper) request).getRawParameter(QueryBuilder.GENE_LIST);
}
String cancerStudyIdListString = request.getParameter(QueryBuilder.CANCER_STUDY_LIST);
String[] cancerStudyIdList = cancerStudyIdListString.split(",");
// Get the priority
Integer dataTypePriority;
try {
dataTypePriority =
Integer.parseInt(request.getParameter(QueryBuilder.DATA_PRIORITY).trim());
} catch (NumberFormatException e) {
dataTypePriority = 0;
}
// Cancer All Cancer Studies
List<CancerStudy> cancerStudiesList = accessControl.getCancerStudies();
HashMap<String, Boolean> studyMap = new HashMap<>();
for (String studyId : cancerStudyIdList) {
studyMap.put(studyId, Boolean.TRUE);
}
for (CancerStudy cancerStudy : cancerStudiesList) {
String cancerStudyId = cancerStudy.getCancerStudyStableId();
if (!studyMap.containsKey(cancerStudyId)) {
continue;
}
if (cancerStudyId.equalsIgnoreCase("all")) continue;
Map cancerMap = new LinkedHashMap();
cancerMap.put("studyId", cancerStudyId);
cancerMap.put("typeOfCancer", cancerStudy.getTypeOfCancerId());
// Get all Genetic Profiles Associated with this Cancer Study ID.
ArrayList<GeneticProfile> geneticProfileList =
GetGeneticProfiles.getGeneticProfiles(cancerStudyId);
// Get all Patient Lists Associated with this Cancer Study ID.
ArrayList<SampleList> sampleSetList = GetSampleLists.getSampleLists(cancerStudyId);
// Get the default patient set
AnnotatedSampleSets annotatedSampleSets =
new AnnotatedSampleSets(sampleSetList, dataTypePriority);
SampleList defaultSampleSet = annotatedSampleSets.getDefaultSampleList();
if (defaultSampleSet == null) {
continue;
}
List<String> sampleIds = defaultSampleSet.getSampleList();
// Get the default genomic profiles
CategorizedGeneticProfileSet categorizedGeneticProfileSet =
new CategorizedGeneticProfileSet(geneticProfileList);
HashMap<String, GeneticProfile> defaultGeneticProfileSet = null;
switch (dataTypePriority) {
case 2:
defaultGeneticProfileSet = categorizedGeneticProfileSet.getDefaultCopyNumberMap();
break;
case 1:
defaultGeneticProfileSet = categorizedGeneticProfileSet.getDefaultMutationMap();
break;
case 0:
default:
defaultGeneticProfileSet =
categorizedGeneticProfileSet.getDefaultMutationAndCopyNumberMap();
}
String mutationProfile = "", cnaProfile = "";
for (GeneticProfile geneticProfile : defaultGeneticProfileSet.values()) {
GeneticAlterationType geneticAlterationType = geneticProfile.getGeneticAlterationType();
if (geneticAlterationType.equals(GeneticAlterationType.COPY_NUMBER_ALTERATION)) {
cnaProfile = geneticProfile.getStableId();
} else if (geneticAlterationType.equals(GeneticAlterationType.MUTATION_EXTENDED)) {
mutationProfile = geneticProfile.getStableId();
}
}
cancerMap.put("mutationProfile", mutationProfile);
cancerMap.put("cnaProfile", cnaProfile);
cancerMap.put("caseSetId", defaultSampleSet.getStableId());
cancerMap.put("caseSetLength", sampleIds.size());
ProfileDataSummary genomicData =
getGenomicData(
cancerStudyId,
defaultGeneticProfileSet,
defaultSampleSet,
geneList,
sampleSetList,
request,
response);
ArrayList<GeneWithScore> geneFrequencyList = genomicData.getGeneFrequencyList();
ArrayList<String> genes = new ArrayList<String>();
for (GeneWithScore geneWithScore : geneFrequencyList) genes.add(geneWithScore.getGene());
int noOfMutated = 0,
noOfCnaUp = 0,
noOfCnaDown = 0,
noOfCnaLoss = 0,
noOfCnaGain = 0,
noOfOther = 0,
noOfAll = 0;
boolean skipStudy = defaultGeneticProfileSet.isEmpty();
if (!skipStudy) {
for (String sampleId : sampleIds) {
if (sampleId == null) {
continue;
}
if (!genomicData.isCaseAltered(sampleId)) continue;
boolean isAnyMutated = false,
isAnyCnaUp = false,
isAnyCnaDown = false,
isAnyCnaLoss = false,
isAnyCnaGain = false;
for (String gene : genes) {
isAnyMutated |= genomicData.isGeneMutated(gene, sampleId);
GeneticTypeLevel cnaLevel = genomicData.getCNALevel(gene, sampleId);
boolean isCnaUp = cnaLevel != null && cnaLevel.equals(GeneticTypeLevel.Amplified);
isAnyCnaUp |= isCnaUp;
boolean isCnaDown =
cnaLevel != null && cnaLevel.equals(GeneticTypeLevel.HomozygouslyDeleted);
isAnyCnaDown |= isCnaDown;
boolean isCnaLoss =
cnaLevel != null && cnaLevel.equals(GeneticTypeLevel.HemizygouslyDeleted);
isAnyCnaLoss |= isCnaLoss;
boolean isCnaGain = cnaLevel != null && cnaLevel.equals(GeneticTypeLevel.Gained);
isAnyCnaGain |= isCnaGain;
}
boolean isAnyCnaChanged = isAnyCnaUp || isAnyCnaDown;
if (isAnyMutated && !isAnyCnaChanged) noOfMutated++;
else if (isAnyMutated && isAnyCnaChanged) noOfOther++;
else if (isAnyCnaUp) noOfCnaUp++;
else if (isAnyCnaDown) noOfCnaDown++;
else if (isAnyCnaGain) noOfCnaGain++;
else if (isAnyCnaLoss) noOfCnaLoss++;
noOfAll++;
}
}
Map alterations = new LinkedHashMap();
cancerMap.put("alterations", alterations);
alterations.put("all", noOfAll);
alterations.put("mutation", noOfMutated);
alterations.put("cnaUp", noOfCnaUp);
alterations.put("cnaDown", noOfCnaDown);
alterations.put("cnaLoss", noOfCnaLoss);
alterations.put("cnaGain", noOfCnaGain);
alterations.put("other", noOfOther);
cancerMap.put("genes", genes);
cancerMap.put("skipped", skipStudy);
resultsList.add(cancerMap);
}
JSONValue.writeJSONString(resultsList, writer);
} catch (DaoException e) {
throw new ServletException(e);
} catch (ProtocolException e) {
throw new ServletException(e);
} finally {
writer.close();
}
}
