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());
}
// command line utility
public static void main(String[] args) throws IOException, DaoException {
if (args.length != 2) {
System.out.printf(
"command line usage: importGistic.pl <gistic-data-file.txt> <cancer-study-id>\n"
+ "\t <gistic-data-file.txt> Note that gistic-data-file.txt must be a massaged file, it does not come straight from the Broad\n"
+ "\t <cancer-study-id> e.g. 'tcga_gbm'");
return;
}
SpringUtil.initDataSource();
GisticReader gisticReader = new GisticReader();
File gistic_f = new File(args[0]);
int cancerStudyInternalId = gisticReader.getCancerStudyInternalId(args[1]);
ProgressMonitor.setConsoleMode(false);
System.out.println("Reading data from: " + gistic_f.getAbsolutePath());
System.out.println("CancerStudyId: " + cancerStudyInternalId);
int lines = FileUtil.getNumLines(gistic_f);
System.out.println(" --> total number of lines: " + lines);
ProgressMonitor.setMaxValue(lines);
ArrayList<Gistic> gistics = null;
gistics = gisticReader.parse(gistic_f, cancerStudyInternalId);
if (gistics == null) {
System.out.println("Error: didn't get any data");
return;
}
// add to CGDS database
for (Gistic g : gistics) {
try {
DaoGistic.addGistic(g);
} catch (validationException e) {
// only catching validationException, not DaoException
logger.debug(e);
} catch (DaoException e) {
System.err.println(e);
}
}
ConsoleUtil.showWarnings();
}
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 runImportCnaData2() throws DaoException, IOException {
DaoGeneticAlteration dao = DaoGeneticAlteration.getInstance();
ProgressMonitor.setConsoleMode(false);
// TBD: change this to use getResourceAsStream()
File file = new File("src/test/resources/cna_test2.txt");
ImportTabDelimData parser = new ImportTabDelimData(file, geneticProfileId, null);
int numLines = FileUtil.getNumLines(file);
parser.importData(numLines);
String value = dao.getGeneticAlteration(geneticProfileId, sample1, 207);
assertEquals(value, "0");
value = dao.getGeneticAlteration(geneticProfileId, sample4, 207);
assertEquals(value, "-1");
value = dao.getGeneticAlteration(geneticProfileId, sample2, 207);
assertEquals(value, "0");
value = dao.getGeneticAlteration(geneticProfileId, sample2, 10000);
assertEquals(value, "2");
value = dao.getGeneticAlteration(geneticProfileId, sample3, 10000);
assertEquals(value, "2");
int cnaStatus = Integer.parseInt(dao.getGeneticAlteration(geneticProfileId, sample3, 10000));
assertEquals(CopyNumberStatus.COPY_NUMBER_AMPLIFICATION, cnaStatus);
cnaStatus = Integer.parseInt(dao.getGeneticAlteration(geneticProfileId, sample2, 10000));
assertEquals(CopyNumberStatus.COPY_NUMBER_AMPLIFICATION, cnaStatus);
cnaStatus = Integer.parseInt(dao.getGeneticAlteration(geneticProfileId, sample4, 207));
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 sampleIds = DaoSampleProfile.getAllSampleIdsInProfile(geneticProfileId);
assertEquals(14, sampleIds.size());
}
/**
* 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);
}