public static HashMap<Long, UUID> readSCTIDtoUUIDMapInfo(File snomedLocation)
throws ClassNotFoundException, IOException {
File snomedJbinFile = null;
if (snomedLocation.isDirectory()) {
for (File f : snomedLocation.listFiles()) {
if (f.isFile() && f.getName().toLowerCase().endsWith(".jbin")) {
snomedJbinFile = f;
break;
}
}
} else {
snomedJbinFile = snomedLocation;
}
if (snomedJbinFile == null) {
throw new IOException("Couldn't find sct data file in " + snomedLocation);
}
UUID sctIDType = IsaacMetadataAuxiliaryBinding.SNOMED_INTEGER_ID.getPrimodialUuid();
// Read in the SCT data
HashMap<Long, UUID> sctConcepts = new HashMap<>();
ConsoleUtil.println("Reading " + snomedJbinFile.getName());
DataInputStream in = new DataInputStream(new FileInputStream(snomedJbinFile));
while (in.available() > 0) {
if (sctConcepts.size() % 1000 == 0) {
ConsoleUtil.showProgress();
}
TtkConceptChronicle concept = new TtkConceptChronicle(in);
if (concept.getConceptAttributes() != null
&& concept.getConceptAttributes().getAdditionalIdComponents() != null) {
for (TtkIdentifier id : concept.getConceptAttributes().getAdditionalIdComponents()) {
if (sctIDType.equals(id.getAuthorityUuid())) {
// Store these by SCTID, because there is no reliable way to generate a UUID from a
// SCTID.
sctConcepts.put(
Long.parseLong(id.getDenotation().toString()), concept.getPrimordialUuid());
break;
}
}
}
}
in.close();
ConsoleUtil.println("Read UUIDs from SCT file - read " + sctConcepts.size() + " concepts");
return sctConcepts;
}
// 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();
}
@Override
public void execute() throws MojoExecutionException {
ConverterUUID.disableUUIDMap_ =
((createDebugUUIDMap == null || createDebugUUIDMap.length() == 0)
? false
: Boolean.parseBoolean(createDebugUUIDMap));
if (ConverterUUID.disableUUIDMap_) {
ConsoleUtil.println(
"The UUID Debug map is disabled - this also prevents duplicate ID detection");
}
// Set up the output
if (!outputDirectory.exists()) {
outputDirectory.mkdirs();
}
checkSkipListSupport();
}
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);
}