public ObjectNode processMutHm(long entrezGeneId, ArrayList<Integer> sampleList, HashMap mutHm) {
ObjectNode _datum = mapper.createObjectNode();
// create map to pair sample and value
HashMap<Integer, String> mapSampleValue = new HashMap<>();
for (Integer sampleId :
sampleList) { // Assign every sample (included non mutated ones) values -- mutated ->
// Mutation Type, non-mutated -> "Non"
String mutationStatus = "Non";
String tmpStr =
new StringBuilder()
.append(Integer.toString(sampleId))
.append(Long.toString(entrezGeneId))
.toString();
if (mutHm.containsKey(tmpStr)) mutationStatus = "Mutated";
mapSampleValue.put(sampleId, mutationStatus);
}
// remove empty entry
Iterator it = mapSampleValue.entrySet().iterator();
while (it.hasNext()) {
Map.Entry pair = (Map.Entry) it.next();
if (pair.getValue().equals("NA")
|| pair.getValue().equals("NaN")
|| pair.getValue().equals("null")) {
it.remove();
}
}
// get Gene Name and Cytoband
DaoGeneOptimized daoGeneOptimized = DaoGeneOptimized.getInstance();
String geneName = daoGeneOptimized.getGene(entrezGeneId).getHugoGeneSymbolAllCaps();
String cytoband = daoGeneOptimized.getGene(entrezGeneId).getCytoband();
// statistics analysis
if (!(Arrays.asList(queriedGenes)).contains(geneName)) { // remove queried genes from result
_datum.put(COL_NAME_GENE, geneName);
_datum.put(COL_NAME_CYTOBAND, cytoband);
_datum.put(COL_NAME_PCT_ALTERED, calcPct(mapSampleValue, profileType, "altered"));
_datum.put(COL_NAME_PCT_UNALTERED, calcPct(mapSampleValue, profileType, "unaltered"));
_datum.put(
COL_NAME_RATIO,
calcRatio(
calcPct(mapSampleValue, profileType, "altered"),
calcPct(mapSampleValue, profileType, "unaltered")));
_datum.put(COL_NAME_DIRECTION, "place holder"); // calculation is done by the front-end
_datum.put(COL_NAME_P_VALUE, calcPval(mapSampleValue, profileType, geneticProfileStableId));
if (!(calcPct(mapSampleValue, profileType, "altered") == 0.0
&& calcPct(mapSampleValue, profileType, "unaltered") == 0.0)
&& !Double.isNaN(calcPval(mapSampleValue, profileType, geneticProfileStableId))) {
return _datum;
}
}
return null;
}
public void testImporter() throws Exception {
ResetDatabase.resetDatabase();
DaoGeneOptimized daoGeneOptimized = DaoGeneOptimized.getInstance();
String[] genes = {
"F2", "EGFR", "FCGR3B", "C1R", "C1QA", "C1QB", "C1QC", "FCGR3A", "C1S", "FCGR1A", "FCGR2A",
"FCGR2B", "FCGR2C", "IL2RA", "IL2RB", "IL2RG"
};
for (String gene : genes) {
daoGeneOptimized.addGene(new CanonicalGene(gene));
}
Calendar cal = Calendar.getInstance();
SimpleDateFormat format = new SimpleDateFormat("yyyy-MM");
String today = format.format(cal.getTime());
DrugDataResource pihelper =
new DrugDataResource(
"PiHelper",
"https://bitbucket.org/armish/pihelper/downloads/pihelper_data_20121107.zip",
today);
PiHelperImporter importer = new PiHelperImporter(pihelper);
ClassLoader classLoader = this.getClass().getClassLoader();
importer.setDrugInfoFile(classLoader.getResourceAsStream("test_pihelper_drugs.tsv"));
importer.setDrugTargetsFile(classLoader.getResourceAsStream("test_pihelper_drugtargets.tsv"));
importer.importData();
DaoDrug daoDrug = DaoDrug.getInstance();
DaoDrugInteraction daoDrugInteraction = DaoDrugInteraction.getInstance();
ArrayList<Drug> allDrugs = daoDrug.getAllDrugs();
int count = allDrugs.size();
assertEquals(6, count);
assertEquals(16, daoDrugInteraction.getCount());
int[] numOfTargets = {1, 12, 0, 3, 0, 0};
for (int i = 0; i < count; i++) {
assertEquals(numOfTargets[i], daoDrugInteraction.getTargets(allDrugs.get(i)).size());
}
Drug cetuximab = daoDrug.getDrug("33612");
assertEquals(204, cetuximab.getNumberOfClinicalTrials().intValue());
assertTrue(cetuximab.isCancerDrug());
assertFalse(cetuximab.isNutraceuitical());
assertTrue(cetuximab.isApprovedFDA());
Drug etanercept = daoDrug.getDrug("33615");
assertEquals(-1, etanercept.getNumberOfClinicalTrials().intValue());
assertFalse(etanercept.isCancerDrug());
assertFalse(etanercept.isNutraceuitical());
assertTrue(etanercept.isApprovedFDA());
}
@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 runImportCnaData() throws DaoException, IOException {
DaoGeneticAlteration dao = DaoGeneticAlteration.getInstance();
DaoGeneOptimized daoGene = DaoGeneOptimized.getInstance();
// the largest current true Entrez gene ID counts 8 digits
daoGene.addGene(new CanonicalGene(999999207, "TESTAKT1"));
daoGene.addGene(new CanonicalGene(999999208, "TESTAKT2"));
daoGene.addGene(new CanonicalGene(999910000, "TESTAKT3"));
daoGene.addGene(new CanonicalGene(999999369, "TESTARAF"));
daoGene.addGene(new CanonicalGene(999999472, "TESTATM"));
daoGene.addGene(new CanonicalGene(999999673, "TESTBRAF"));
daoGene.addGene(new CanonicalGene(999999672, "TESTBRCA1"));
daoGene.addGene(new CanonicalGene(999999675, "TESTBRCA2"));
ProgressMonitor.setConsoleMode(false);
// TBD: change this to use getResourceAsStream()
File file = new File("src/test/resources/cna_test.txt");
ImportTabDelimData parser = new ImportTabDelimData(file, "Barry", geneticProfileId, null);
int numLines = FileUtil.getNumLines(file);
parser.importData(numLines);
String value = dao.getGeneticAlteration(geneticProfileId, sample1, 999999207);
assertEquals("0", value);
value = dao.getGeneticAlteration(geneticProfileId, sample4, 999999207);
assertEquals("-1", value);
value = dao.getGeneticAlteration(geneticProfileId, sample2, 999999207);
assertEquals("0", value);
value = dao.getGeneticAlteration(geneticProfileId, sample2, 999910000);
assertEquals("2", value);
value = dao.getGeneticAlteration(geneticProfileId, sample3, 999910000);
assertEquals("2", value);
int cnaStatus =
Integer.parseInt(dao.getGeneticAlteration(geneticProfileId, sample3, 999910000));
assertEquals(CopyNumberStatus.COPY_NUMBER_AMPLIFICATION, cnaStatus);
cnaStatus = Integer.parseInt(dao.getGeneticAlteration(geneticProfileId, sample2, 999910000));
assertEquals(CopyNumberStatus.COPY_NUMBER_AMPLIFICATION, cnaStatus);
cnaStatus = Integer.parseInt(dao.getGeneticAlteration(geneticProfileId, sample4, 999999207));
assertEquals(CopyNumberStatus.HEMIZYGOUS_DELETION, cnaStatus);
Patient patient = DaoPatient.getPatientByCancerStudyAndPatientId(studyId, "TCGA-A1-A0SB");
Sample sample =
DaoSample.getSampleByPatientAndSampleId(patient.getInternalId(), "TCGA-A1-A0SB-01");
assertTrue(
DaoSampleProfile.sampleExistsInGeneticProfile(sample.getInternalId(), geneticProfileId));
patient = DaoPatient.getPatientByCancerStudyAndPatientId(studyId, "TCGA-A1-A0SJ");
sample = DaoSample.getSampleByPatientAndSampleId(patient.getInternalId(), "TCGA-A1-A0SJ-01");
assertTrue(
DaoSampleProfile.sampleExistsInGeneticProfile(sample.getInternalId(), geneticProfileId));
ArrayList caseIds = DaoSampleProfile.getAllSampleIdsInProfile(geneticProfileId);
assertEquals(14, caseIds.size());
}
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);
}
@Override
public ObjectNode process(long entrezGeneId, String[] values, ArrayList<Integer> sampleList) {
ObjectNode _datum = mapper.createObjectNode();
// create map to pair sample and value
HashMap<Integer, String> mapSampleValue = new HashMap<>();
for (int i = 0; i < values.length; i++) {
String value = values[i];
Integer sampleId = sampleList.get(i);
mapSampleValue.put(sampleId, value);
}
// remove empty entry
Iterator it = mapSampleValue.entrySet().iterator();
while (it.hasNext()) {
Map.Entry pair = (Map.Entry) it.next();
if (pair.getValue().equals("NA")
|| pair.getValue().equals("NaN")
|| pair.getValue().equals("null")) {
it.remove();
}
}
// get Gene Name and Cytoband
DaoGeneOptimized daoGeneOptimized = DaoGeneOptimized.getInstance();
String geneName = daoGeneOptimized.getGene(entrezGeneId).getHugoGeneSymbolAllCaps();
String cytoband = daoGeneOptimized.getGene(entrezGeneId).getCytoband();
// statistics analysis
if (profileType.equals(GeneticAlterationType.COPY_NUMBER_ALTERATION.toString())) {
if (!(Arrays.asList(queriedGenes)).contains(geneName)) { // remove queried genes from result
_datum.put(COL_NAME_GENE, geneName);
_datum.put(COL_NAME_CYTOBAND, cytoband);
_datum.put(
COL_NAME_PCT_ALTERED,
Integer.toString(countAltered(mapSampleValue, profileType, "altered"))
+ "////"
+ Double.toString(calcPct(mapSampleValue, profileType, "altered")));
_datum.put(
COL_NAME_PCT_UNALTERED,
Integer.toString(countAltered(mapSampleValue, profileType, "unaltered"))
+ "////"
+ Double.toString(calcPct(mapSampleValue, profileType, "unaltered")));
_datum.put(
COL_NAME_RATIO,
calcRatio(
calcPct(mapSampleValue, profileType, "altered"),
calcPct(mapSampleValue, profileType, "unaltered")));
_datum.put(COL_NAME_DIRECTION, "place holder"); // calculation is done by the front-end
_datum.put(COL_NAME_P_VALUE, calcPval(mapSampleValue, profileType, geneticProfileStableId));
if (!(calcPct(mapSampleValue, profileType, "altered") == 0.0
&& calcPct(mapSampleValue, profileType, "unaltered") == 0.0)
&& !Double.isNaN(calcPval(mapSampleValue, profileType, geneticProfileStableId))) {
return _datum;
}
}
} else if (profileType.equals(GeneticAlterationType.MRNA_EXPRESSION.toString())) {
_datum.put(COL_NAME_GENE, geneName);
_datum.put(COL_NAME_CYTOBAND, cytoband);
_datum.put(
COL_NAME_MEAN_ALTERED, calcMean(mapSampleValue, "altered", geneticProfileStableId));
_datum.put(
COL_NAME_MEAN_UNALTERED, calcMean(mapSampleValue, "unaltered", geneticProfileStableId));
_datum.put(
COL_NAME_STDEV_ALTERED, calcSTDev(mapSampleValue, "altered", geneticProfileStableId));
_datum.put(
COL_NAME_STDEV_UNALTERED, calcSTDev(mapSampleValue, "unaltered", geneticProfileStableId));
_datum.put(COL_NAME_P_VALUE, calcPval(mapSampleValue, profileType, geneticProfileStableId));
if (!Double.isNaN(calcPval(mapSampleValue, profileType, geneticProfileStableId))) {
return _datum;
}
} else if (profileType.equals(GeneticAlterationType.PROTEIN_LEVEL.toString())) {
_datum.put(COL_NAME_GENE, geneName);
_datum.put(COL_NAME_CYTOBAND, cytoband);
_datum.put(
COL_NAME_MEAN_ALTERED, calcMean(mapSampleValue, "altered", geneticProfileStableId));
_datum.put(
COL_NAME_MEAN_UNALTERED, calcMean(mapSampleValue, "unaltered", geneticProfileStableId));
_datum.put(
COL_NAME_STDEV_ALTERED, calcSTDev(mapSampleValue, "altered", geneticProfileStableId));
_datum.put(
COL_NAME_STDEV_UNALTERED, calcSTDev(mapSampleValue, "unaltered", geneticProfileStableId));
_datum.put(COL_NAME_P_VALUE, calcPval(mapSampleValue, profileType, geneticProfileStableId));
if (!Double.isNaN(calcPval(mapSampleValue, profileType, geneticProfileStableId))) {
return _datum;
}
}
return null;
}
/**
* 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);
}
