@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();
}
/**
* 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;
}
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);
}
/**
* 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);
}