/**
* 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);
}
