Examples with MeanVarianceRelation ubic.gemma.model.expression.bioAssayData.MeanVarianceRelation used on opensource projects

Example 1 with MeanVarianceRelation

use of ubic.gemma.model.expression.bioAssayData.MeanVarianceRelation in project Gemma by PavlidisLab.

the class MeanVarianceServiceTest method testServiceCreateOneColor.

@Test
public final void testServiceCreateOneColor() {
    qt = this.createOrUpdateQt(ScaleType.LOG2);
    qt.setIsNormalized(false);
    quantitationTypeService.update(qt);
    // update ArrayDesign to ONECOLOR
    Collection<ArrayDesign> aas = eeService.getArrayDesignsUsed(ee);
    assertEquals(1, aas.size());
    ArrayDesign des = aas.iterator().next();
    des.setTechnologyType(TechnologyType.ONECOLOR);
    arrayDesignService.update(des);
    // check that ArrayDesign is the right TechnologyType
    aas = eeService.getArrayDesignsUsed(ee);
    assertEquals(1, aas.size());
    des = aas.iterator().next();
    assertEquals(TechnologyType.ONECOLOR, des.getTechnologyType());
    MeanVarianceRelation mvr = meanVarianceService.create(ee, true);
    // convert byte[] to array[]
    // warning: order may have changed
    double[] means = MeanVarianceServiceTest.bac.byteArrayToDoubles(mvr.getMeans());
    double[] variances = MeanVarianceServiceTest.bac.byteArrayToDoubles(mvr.getVariances());
    Arrays.sort(means);
    Arrays.sort(variances);
    // check sizes
    int expectedMeanVarianceLength = 75;
    // NAs removed
    int expectedLowessLength = 75;
    assertEquals(expectedMeanVarianceLength, means.length);
    assertEquals(expectedMeanVarianceLength, variances.length);
    // check results
    int idx = 0;
    assertEquals(-0.3484, means[idx], 0.0001);
    assertEquals(0.001569, variances[idx], 0.0001);
    idx = expectedLowessLength - 1;
    assertEquals(0.05115, means[idx], 0.0001);
    assertEquals(0.12014, variances[idx], 0.0001);
}

Example 2 with MeanVarianceRelation

use of ubic.gemma.model.expression.bioAssayData.MeanVarianceRelation in project Gemma by PavlidisLab.

the class MeanVarianceServiceTest method testServiceLinearNormalized.

@Test
public final void testServiceLinearNormalized() {
    assertEquals(97, ee.getProcessedExpressionDataVectors().size());
    MeanVarianceRelation mvr = meanVarianceService.create(ee, true);
    // convert byte[] to array[]
    // warning: order may have changed
    double[] means = MeanVarianceServiceTest.bac.byteArrayToDoubles(mvr.getMeans());
    double[] variances = MeanVarianceServiceTest.bac.byteArrayToDoubles(mvr.getVariances());
    Arrays.sort(means);
    Arrays.sort(variances);
    // after filtering
    int expectedLength = 97;
    assertEquals(expectedLength, means.length);
    assertEquals(expectedLength, variances.length);
    // duplicate rows removed
    expectedLength = 95;
    int idx = 0;
    assertEquals(-1.9858, means[idx], 0.0001);
    assertEquals(0, variances[idx], 0.0001);
    idx = expectedLength - 1;
    assertEquals(0.02509, means[idx], 0.0001);
    assertEquals(0.09943, variances[idx], 0.0001);
}

Example 3 with MeanVarianceRelation

use of ubic.gemma.model.expression.bioAssayData.MeanVarianceRelation in project Gemma by PavlidisLab.

the class MeanVarianceServiceTest method testServiceCreateTwoColor.

@Test
public final void testServiceCreateTwoColor() {
    qt = this.createOrUpdateQt(ScaleType.LOG2);
    qt.setIsNormalized(false);
    quantitationTypeService.update(qt);
    // update ArrayDesign to TWOCOLOR
    Collection<ArrayDesign> aas = eeService.getArrayDesignsUsed(ee);
    assertEquals(1, aas.size());
    ArrayDesign des = aas.iterator().next();
    des.setTechnologyType(TechnologyType.TWOCOLOR);
    arrayDesignService.update(des);
    aclTestUtils.checkHasAcl(des);
    // check that ArrayDesign is the right TechnologyType
    aas = eeService.getArrayDesignsUsed(ee);
    assertEquals(1, aas.size());
    des = aas.iterator().next();
    assertEquals(TechnologyType.TWOCOLOR, des.getTechnologyType());
    MeanVarianceRelation mvr = meanVarianceService.create(ee, true);
    aclTestUtils.checkEEAcls(ee);
    ee = eeService.thaw(ee);
    assertEquals(97, ee.getProcessedExpressionDataVectors().size());
    // convert byte[] to array[]
    // warning: order may have changed
    double[] means = MeanVarianceServiceTest.bac.byteArrayToDoubles(mvr.getMeans());
    double[] variances = MeanVarianceServiceTest.bac.byteArrayToDoubles(mvr.getVariances());
    Arrays.sort(means);
    Arrays.sort(variances);
    // after filtering
    int expectedLength = 75;
    assertEquals(expectedLength, means.length);
    assertEquals(expectedLength, variances.length);
    int idx = 0;
    assertEquals(-0.34836, means[idx], 0.0001);
    assertEquals(0.001569, variances[idx], 0.0001);
    idx = expectedLength - 1;
    assertEquals(0.05115, means[idx], 0.0001);
    assertEquals(0.12014, variances[idx], 0.0001);
}

Example 4 with MeanVarianceRelation

use of ubic.gemma.model.expression.bioAssayData.MeanVarianceRelation in project Gemma by PavlidisLab.

the class MeanVarianceServiceImpl method create.

@Override
public MeanVarianceRelation create(ExpressionExperiment ee, boolean forceRecompute) {
    if (ee == null) {
        log.warn("Experiment is null");
        return null;
    }
    log.info("Starting mean-variance computation");
    MeanVarianceRelation mvr = ee.getMeanVarianceRelation();
    if (forceRecompute || mvr == null) {
        log.info("Recomputing mean-variance");
        ee = expressionExperimentService.thawLiter(ee);
        mvr = ee.getMeanVarianceRelation();
        ExpressionDataDoubleMatrix intensities = meanVarianceServiceHelper.getIntensities(ee);
        if (intensities == null) {
            throw new IllegalStateException("Could not locate intensity matrix for " + ee.getShortName());
        }
        Collection<QuantitationType> qtList = expressionExperimentService.getPreferredQuantitationType(ee);
        QuantitationType qt;
        if (qtList.size() == 0) {
            log.error("Did not find any preferred quantitation type. Mean-variance relation was not computed.");
        } else {
            qt = qtList.iterator().next();
            if (qtList.size() > 1) {
                // Really this should be an error condition.
                log.warn("Found more than one preferred quantitation type. Only the first preferred quantitation type (" + qt + ") will be used.");
            }
            try {
                intensities = ExpressionDataDoubleMatrixUtil.filterAndLog2Transform(qt, intensities);
            } catch (UnknownLogScaleException e) {
                log.warn("Problem log transforming data. Check that the appropriate log scale is used. Mean-variance will be computed as is.");
            }
            mvr = calculateMeanVariance(intensities, mvr);
            meanVarianceServiceHelper.createMeanVariance(ee, mvr);
        }
    }
    log.info("Mean-variance computation is complete");
    return mvr;
}

Example 5 with MeanVarianceRelation

use of ubic.gemma.model.expression.bioAssayData.MeanVarianceRelation in project Gemma by PavlidisLab.

the class MeanVarianceServiceTest method testServiceCreateCountData.

@Test
public final void testServiceCreateCountData() throws Exception {
    // so it doesn't look for soft files
    geoService.setGeoDomainObjectGenerator(new GeoDomainObjectGenerator());
    ee = eeService.findByShortName("GSE29006");
    if (ee != null) {
        eeService.remove(ee);
    }
    assertNull(eeService.findByShortName("GSE29006"));
    try {
        Collection<?> results = geoService.fetchAndLoad("GSE29006", false, false, false);
        ee = (ExpressionExperiment) results.iterator().next();
    } catch (AlreadyExistsInSystemException e) {
        throw new IllegalStateException("Need to remove this data set before test is run");
    }
    ee = eeService.thaw(ee);
    qt = this.createOrUpdateQt(ScaleType.COUNT);
    // Load the data from a text file.
    DoubleMatrixReader reader = new DoubleMatrixReader();
    try (InputStream countData = this.getClass().getResourceAsStream("/data/loader/expression/flatfileload/GSE29006_expression_count.test.txt");
        InputStream rpkmData = this.getClass().getResourceAsStream("/data/loader/expression/flatfileload/GSE29006_expression_RPKM.test.txt")) {
        DoubleMatrix<String, String> countMatrix = reader.read(countData);
        DoubleMatrix<String, String> rpkmMatrix = reader.read(rpkmData);
        List<String> probeNames = countMatrix.getRowNames();
        // we have to find the right generic platform to use.
        ArrayDesign targetArrayDesign = this.getTestPersistentArrayDesign(probeNames, taxonService.findByCommonName("human"));
        targetArrayDesign = arrayDesignService.thaw(targetArrayDesign);
        try {
            dataUpdater.addCountData(ee, targetArrayDesign, countMatrix, rpkmMatrix, 36, true, false);
            fail("Should have gotten an exception");
        } catch (IllegalArgumentException e) {
        // Expected
        }
        dataUpdater.addCountData(ee, targetArrayDesign, countMatrix, rpkmMatrix, 36, true, true);
    }
    ee = eeService.thaw(this.ee);
    assertNotNull(ee.getId());
    MeanVarianceRelation mvr = meanVarianceService.create(ee, true);
    // convert byte[] to array[]
    // warning: order may have changed
    double[] means = MeanVarianceServiceTest.bac.byteArrayToDoubles(mvr.getMeans());
    double[] variances = MeanVarianceServiceTest.bac.byteArrayToDoubles(mvr.getVariances());
    if (means != null) {
        Arrays.sort(means);
    }
    if (variances != null) {
        Arrays.sort(variances);
    }
    // check sizes
    int expectedMeanVarianceLength = 199;
    // NAs removed
    int expectedLowessLength = 197;
    assert means != null;
    assertEquals(expectedMeanVarianceLength, means.length);
    assert variances != null;
    assertEquals(expectedMeanVarianceLength, variances.length);
    int idx = 0;
    assertEquals(1.037011, means[idx], 0.0001);
    assertEquals(0.00023724336, variances[idx], 0.000001);
    idx = expectedLowessLength - 1;
    assertEquals(15.23313, means[idx], 0.0001);
    assertEquals(4.84529, variances[idx], 0.0001);
}