From c1600b8517c88a920b650a4845e820c03b1ba48a Mon Sep 17 00:00:00 2001
From: kohleman <kohleman>
Date: Tue, 14 Dec 2010 10:52:13 +0000
Subject: [PATCH] adopted the Sample Library Strategy/Selection
SVN: 19121
---
.../cisd/dsu/dss/plugins/DataSetToSOFT.java | 112 +++++++++++-------
.../dsu/dss/plugins/DataSetToSOFTTest.java | 78 ++++++------
2 files changed, 110 insertions(+), 80 deletions(-)
diff --git a/deep_sequencing_unit/source/java/ch/ethz/bsse/cisd/dsu/dss/plugins/DataSetToSOFT.java b/deep_sequencing_unit/source/java/ch/ethz/bsse/cisd/dsu/dss/plugins/DataSetToSOFT.java
index 4071b8c4239..f080ae98f77 100644
--- a/deep_sequencing_unit/source/java/ch/ethz/bsse/cisd/dsu/dss/plugins/DataSetToSOFT.java
+++ b/deep_sequencing_unit/source/java/ch/ethz/bsse/cisd/dsu/dss/plugins/DataSetToSOFT.java
@@ -54,12 +54,13 @@ import ch.systemsx.cisd.openbis.generic.shared.dto.identifier.SampleIdentifierFa
/**
* Processing plugin for exporting meta data of a flow lane as a SOFT file.
- *
+ *
* @author Franz-Josef Elmer
*/
public class DataSetToSOFT implements IProcessingPluginTask
{
- @Private static final String EXTERNAL_SAMPLE_NAME_PROPERTY = "EXTERNAL_SAMPLE_NAME";
+ @Private
+ static final String EXTERNAL_SAMPLE_NAME_PROPERTY = "EXTERNAL_SAMPLE_NAME";
private static final String EMPTY = "<<<NEED_TO_BE_FILLED>>>";
@@ -79,34 +80,35 @@ public class DataSetToSOFT implements IProcessingPluginTask
private static final class SOFTBuilder
{
private final StringBuilder builder = new StringBuilder();
-
+
void addSample(Sample sample, String propertyTypeCode)
{
addLine('^', "SAMPLE", getProperty(sample, propertyTypeCode));
}
-
- void addSampleProperty(String key, Sample sample, String propertyTypeCode, Map<String, String> translation)
+
+ void addSampleProperty(String key, Sample sample, String propertyTypeCode,
+ Map<String, String> translation)
{
String property = getProperty(sample, propertyTypeCode);
String translatedProperty = translation.get(property);
addProperty(key, translatedProperty == null ? property : translatedProperty);
}
-
+
void addSampleProperty(String key, Sample sample, String propertyTypeCode)
{
addProperty(key, getProperty(sample, propertyTypeCode));
}
-
+
void addProperty(String key, String property)
{
addLine('!', "Sample_" + key, property);
}
-
+
private void addLine(char prefix, String key, String value)
{
builder.append(prefix).append(key).append(" = ").append(value).append('\n');
}
-
+
private String getProperty(Sample sample, String propertyTypeCode)
{
String property = tryToGetProperty(sample, propertyTypeCode);
@@ -119,7 +121,7 @@ public class DataSetToSOFT implements IProcessingPluginTask
return builder.toString();
}
}
-
+
private static final String tryToGetProperty(Sample sample, String propertyTypeCode)
{
List<IEntityProperty> properties = sample.getProperties();
@@ -132,12 +134,12 @@ public class DataSetToSOFT implements IProcessingPluginTask
}
return null;
}
-
- private static final Logger operationLog =
- LogFactory.getLogger(LogCategory.OPERATION, DataSetToSOFT.class);
+
+ private static final Logger operationLog = LogFactory.getLogger(LogCategory.OPERATION,
+ DataSetToSOFT.class);
private static final long serialVersionUID = 1L;
-
+
private final File storeRoot;
private final Map<String, String> translation;
@@ -148,7 +150,7 @@ public class DataSetToSOFT implements IProcessingPluginTask
{
this(properties, storeRoot, null);
}
-
+
DataSetToSOFT(Properties properties, File storeRoot, IEncapsulatedOpenBISService service)
{
this.storeRoot = storeRoot;
@@ -166,7 +168,7 @@ public class DataSetToSOFT implements IProcessingPluginTask
translation.put("MRNA", "other");
translation.put("PROCESSED_DNA_LIBRARY", EMPTY);
}
-
+
public ProcessingStatus process(List<DatasetDescription> datasets,
DataSetProcessingContext context)
{
@@ -193,26 +195,37 @@ public class DataSetToSOFT implements IProcessingPluginTask
continue;
}
String checkSum = calculateCheckSum(srfFile);
-
+
SOFTBuilder softBuilder = new SOFTBuilder();
softBuilder.addSample(sequencingSample, EXTERNAL_SAMPLE_NAME_PROPERTY);
softBuilder.addProperty("type", "SRA");
- softBuilder.addSampleProperty("title", sequencingSample, EXTERNAL_SAMPLE_NAME_PROPERTY);
- softBuilder.addSampleProperty("source_name", sequencingSample, "SAMPLE_SOURCE_NAME");
- softBuilder.addSampleProperty("organism", sequencingSample, "NCBI_ORGANISM_TAXONOMY");
- softBuilder.addSampleProperty("characteristics", sequencingSample, "SAMPLE_CHARACTERISTICS");
- softBuilder.addSampleProperty("biomaterial_provider", sequencingSample, "CONTACT_PERSON_NAME");
- softBuilder.addSampleProperty("molecule", sequencingSample, "SEQUENCING_APPLICATION");
- softBuilder.addSampleProperty("extract_protocol", sequencingSample, "SAMPLE_EXTRACT_PROTOCOL");
- softBuilder.addSampleProperty("data_processing", sequencingSample, "SAMPLE_DATA_PROCESSING");
- softBuilder.addSampleProperty("library_strategy", sequencingSample, "SEQUENCING_APPLICATION");
- softBuilder.addSampleProperty("library_source", sequencingSample, "SAMPLE_KIND", translation);
- softBuilder.addSampleProperty("library_selection", sequencingSample, "SAMPLE_KIND");
- softBuilder.addSampleProperty("instrument_model", flowCellSample, "GENOME_ANALYZER");
+ softBuilder.addSampleProperty("title", sequencingSample,
+ EXTERNAL_SAMPLE_NAME_PROPERTY);
+ softBuilder
+ .addSampleProperty("source_name", sequencingSample, "SAMPLE_SOURCE_NAME");
+ softBuilder.addSampleProperty("organism", sequencingSample,
+ "NCBI_ORGANISM_TAXONOMY");
+ softBuilder.addSampleProperty("characteristics", sequencingSample,
+ "SAMPLE_CHARACTERISTICS");
+ softBuilder.addSampleProperty("biomaterial_provider", sequencingSample,
+ "CONTACT_PERSON_NAME");
+ softBuilder.addSampleProperty("molecule", sequencingSample, "SAMPLE_MOLECULE");
+ softBuilder.addSampleProperty("extract_protocol", sequencingSample,
+ "SAMPLE_EXTRACT_PROTOCOL");
+ softBuilder.addSampleProperty("data_processing", sequencingSample,
+ "SAMPLE_DATA_PROCESSING");
+ softBuilder.addSampleProperty("library_strategy", sequencingSample,
+ "SAMPLE_LIBRARY_STRATEGY");
+ softBuilder.addSampleProperty("library_source", sequencingSample, "SAMPLE_KIND",
+ translation);
+ softBuilder.addSampleProperty("library_selection", sequencingSample,
+ "SAMPLE_LIBRARY_SELECTION");
+ softBuilder
+ .addSampleProperty("instrument_model", flowCellSample, "GENOME_ANALYZER");
softBuilder.addProperty("raw_file_1", srfFile.getName());
softBuilder.addProperty("raw_file_type_1", "srf");
softBuilder.addProperty("file_checksum_1", checkSum);
-
+
String subject = createSubject(sequencingSample);
String content = createContent(sequencingSample, flowLaneSample, srfDataSet);
String fileName = createSoftFileName(sequencingSample, flowLaneSample);
@@ -222,13 +235,14 @@ public class DataSetToSOFT implements IProcessingPluginTask
status.addDatasetStatus(datasetDescription, Status.OK);
} catch (Exception ex)
{
- status.addDatasetStatus(datasetDescription, Status.createError("Exception occured: " + ex));
+ status.addDatasetStatus(datasetDescription,
+ Status.createError("Exception occured: " + ex));
operationLog.error("Exception occured while processing " + datasetDescription, ex);
}
}
return status;
}
-
+
private String createSubject(Sample sequencingSample)
{
Template template = E_MAIL_SUBJECT_TEMPLATE.createFreshCopy();
@@ -236,7 +250,8 @@ public class DataSetToSOFT implements IProcessingPluginTask
return template.createText();
}
- private String createContent(Sample sequencingSample, Sample flowLaneSample, ExternalData dataSet)
+ private String createContent(Sample sequencingSample, Sample flowLaneSample,
+ ExternalData dataSet)
{
Template template = E_MAIL_CONTENT_TEMPLATE.createFreshCopy();
bindExternalSampleName(template, sequencingSample);
@@ -244,22 +259,24 @@ public class DataSetToSOFT implements IProcessingPluginTask
template.bind("data-set", dataSet.getCode());
return template.createText();
}
-
+
private String createSoftFileName(Sample sequencingSample, Sample flowLaneSample)
{
Template template = SOFT_FILE_NAME_TEMPLATE.createFreshCopy();
- String externalSampleName = tryToGetProperty(sequencingSample, EXTERNAL_SAMPLE_NAME_PROPERTY);
+ String externalSampleName =
+ tryToGetProperty(sequencingSample, EXTERNAL_SAMPLE_NAME_PROPERTY);
template.bind("external-sample-name", externalSampleName.replace(' ', '_'));
template.bind("flow-lane", flowLaneSample.getCode().replace(':', '-'));
return template.createText();
}
-
+
private void bindExternalSampleName(Template template, Sample sequencingSample)
{
- String externalSampleName = tryToGetProperty(sequencingSample, EXTERNAL_SAMPLE_NAME_PROPERTY);
+ String externalSampleName =
+ tryToGetProperty(sequencingSample, EXTERNAL_SAMPLE_NAME_PROPERTY);
template.bind("external-sample-name", externalSampleName);
}
-
+
private String calculateCheckSum(File srfFile)
{
String checkSum;
@@ -272,13 +289,13 @@ public class DataSetToSOFT implements IProcessingPluginTask
}
return checkSum;
}
-
+
private File tryToFindSrfFile(DatasetDescription datasetDescription)
{
File root = new File(storeRoot, datasetDescription.getDataSetLocation());
return tryToFindSrfFile(root);
}
-
+
private File tryToFindSrfFile(File file)
{
if (file.isFile() && file.getName().endsWith(".srf"))
@@ -299,22 +316,25 @@ public class DataSetToSOFT implements IProcessingPluginTask
}
return null;
}
-
+
private Sample getFlowLaneSample(ExternalData dataSet)
{
SampleIdentifier identifier = SampleIdentifierFactory.parse(dataSet.getSampleIdentifier());
return getService().tryGetSampleWithExperiment(identifier);
}
-
+
private Sample getFlowCellSample(Sample flowLaneSample)
{
- SampleIdentifier identifier = SampleIdentifierFactory.parse(flowLaneSample.getContainer().getIdentifier());
+ SampleIdentifier identifier =
+ SampleIdentifierFactory.parse(flowLaneSample.getContainer().getIdentifier());
return getService().tryGetSampleWithExperiment(identifier);
}
-
+
private Sample getSequencingSample(Sample flowLaneSample)
{
- List<Sample> parents = getService().listSamples(ListSampleCriteria.createForChild(new TechId(flowLaneSample.getId())));
+ List<Sample> parents =
+ getService().listSamples(
+ ListSampleCriteria.createForChild(new TechId(flowLaneSample.getId())));
return parents.get(0);
}
@@ -328,7 +348,7 @@ public class DataSetToSOFT implements IProcessingPluginTask
throw CheckedExceptionTunnel.wrapIfNecessary(ex);
}
}
-
+
private IEncapsulatedOpenBISService getService()
{
if (service == null)
diff --git a/deep_sequencing_unit/sourceTest/java/ch/ethz/bsse/cisd/dsu/dss/plugins/DataSetToSOFTTest.java b/deep_sequencing_unit/sourceTest/java/ch/ethz/bsse/cisd/dsu/dss/plugins/DataSetToSOFTTest.java
index 0233f85afb1..7092a28b1aa 100644
--- a/deep_sequencing_unit/sourceTest/java/ch/ethz/bsse/cisd/dsu/dss/plugins/DataSetToSOFTTest.java
+++ b/deep_sequencing_unit/sourceTest/java/ch/ethz/bsse/cisd/dsu/dss/plugins/DataSetToSOFTTest.java
@@ -58,24 +58,31 @@ import ch.systemsx.cisd.openbis.generic.shared.dto.DatasetDescription;
import ch.systemsx.cisd.openbis.generic.shared.dto.identifier.SampleIdentifierFactory;
/**
- *
- *
* @author Franz-Josef Elmer
*/
-@Friend(toClasses=DataSetToSOFT.class)
+@Friend(toClasses = DataSetToSOFT.class)
public class DataSetToSOFTTest extends AbstractFileSystemTestCase
{
private static final String USER_EMAIL = "user@ho.me";
+
private static final String SRF_FILE_NAME = "sample.srf";
+
private static final String SRF_FILE_CONTENT = "hello world!";
+
private static final String DATASET_CODE = "ds-1";
+
private static final String FLOW_LANE_SAMPLE_CODE = "flow-lane:1";
-
+
private BufferedAppender logRecorder;
+
private Mockery context;
+
private IEncapsulatedOpenBISService service;
+
private IMailClient mailClient;
+
private DataSetProcessingContext dataSetProcessingContext;
+
private IProcessingPluginTask processingPlugin;
@BeforeMethod
@@ -120,65 +127,68 @@ public class DataSetToSOFTTest extends AbstractFileSystemTestCase
addProperty(sequencingSample, "SAMPLE_SOURCE_NAME", "source");
addProperty(sequencingSample, "NCBI_ORGANISM_TAXONOMY", "organism");
addProperty(sequencingSample, "CONTACT_PERSON_NAME", "person");
+ addProperty(sequencingSample, "SAMPLE_MOLECULE", "total_rna");
addProperty(sequencingSample, "SEQUENCING_APPLICATION", "application");
addProperty(sequencingSample, "SAMPLE_EXTRACT_PROTOCOL", "protocol");
addProperty(sequencingSample, "SAMPLE_DATA_PROCESSING", "processing");
+ addProperty(sequencingSample, "SAMPLE_LIBRARY_SELECTION", "cDNA");
+ addProperty(sequencingSample, "SAMPLE_LIBRARY_STRATEGY", "RNA-Seq");
addProperty(sequencingSample, "SAMPLE_KIND", "CHIP");
flowLaneSample.setContainer(flowCellSample);
prepareGetSample(flowLaneSample);
prepareGetSample(flowCellSample);
- final RecordingMatcher<ListSampleCriteria> listSampleCriteriaMatcher = new RecordingMatcher<ListSampleCriteria>();
+ final RecordingMatcher<ListSampleCriteria> listSampleCriteriaMatcher =
+ new RecordingMatcher<ListSampleCriteria>();
final RecordingMatcher<String> subjectMatcher = new RecordingMatcher<String>();
final RecordingMatcher<String> contentMatcher = new RecordingMatcher<String>();
final RecordingMatcher<String> fileNameMatcher = new RecordingMatcher<String>();
final RecordingMatcher<DataHandler> attachmentMatcher = new RecordingMatcher<DataHandler>();
- final RecordingMatcher<EMailAddress[]> addressesMatcher = new RecordingMatcher<EMailAddress[]>();
+ final RecordingMatcher<EMailAddress[]> addressesMatcher =
+ new RecordingMatcher<EMailAddress[]>();
context.checking(new Expectations()
{
{
one(service).listSamples(with(listSampleCriteriaMatcher));
will(returnValue(Arrays.asList(sequencingSample)));
-
+
one(mailClient).sendEmailMessageWithAttachment(with(subjectMatcher),
with(contentMatcher), with(fileNameMatcher), with(attachmentMatcher),
with(new IsNull<EMailAddress>()), with(new IsNull<EMailAddress>()),
with(addressesMatcher));
}
});
-
+
ProcessingStatus status = processingPlugin.process(dataSets, dataSetProcessingContext);
assertEquals(0, status.getErrorStatuses().size());
- assertEquals(flowLaneSample.getId(), listSampleCriteriaMatcher.getRecordedObjects().get(0).getChildSampleId().getId());
+ assertEquals(flowLaneSample.getId(), listSampleCriteriaMatcher.getRecordedObjects().get(0)
+ .getChildSampleId().getId());
Template template = initTemplate(DataSetToSOFT.E_MAIL_SUBJECT_TEMPLATE);
- assertEquals("[" + template.createText() + "]", subjectMatcher.getRecordedObjects().toString());
+ assertEquals("[" + template.createText() + "]", subjectMatcher.getRecordedObjects()
+ .toString());
Template content = initTemplate(DataSetToSOFT.E_MAIL_CONTENT_TEMPLATE);
content.bind("flow-lane", flowLaneSample.getCode());
content.bind("data-set", DATASET_CODE);
- assertEquals("[" + content.createText() + "]", contentMatcher.getRecordedObjects().toString());
+ assertEquals("[" + content.createText() + "]", contentMatcher.getRecordedObjects()
+ .toString());
Template fileNameTemplate = DataSetToSOFT.SOFT_FILE_NAME_TEMPLATE.createFreshCopy();
fileNameTemplate.bind("external-sample-name", "my_sample");
fileNameTemplate.bind("flow-lane", flowLaneSample.getCode().replace(':', '-'));
- assertEquals("[" + fileNameTemplate.createText() + "]", fileNameMatcher.getRecordedObjects().toString());
+ assertEquals("[" + fileNameTemplate.createText() + "]", fileNameMatcher
+ .getRecordedObjects().toString());
List<EMailAddress[]> emailAddresses = addressesMatcher.getRecordedObjects();
assertEquals(USER_EMAIL, emailAddresses.get(0)[0].tryGetEmailAddress());
DataSource dataSource = attachmentMatcher.getRecordedObjects().get(0).getDataSource();
String attachmentContent = getContent(dataSource);
- assertEquals("^SAMPLE = my sample\n"
- + "!Sample_type = SRA\n"
- + "!Sample_title = my sample\n"
- + "!Sample_source_name = source\n"
+ assertEquals("^SAMPLE = my sample\n" + "!Sample_type = SRA\n"
+ + "!Sample_title = my sample\n" + "!Sample_source_name = source\n"
+ "!Sample_organism = organism\n"
+ "!Sample_characteristics = <<<NEED_TO_BE_FILLED>>>\n"
- + "!Sample_biomaterial_provider = person\n"
- + "!Sample_molecule = application\n"
+ + "!Sample_biomaterial_provider = person\n" + "!Sample_molecule = total_rna\n"
+ "!Sample_extract_protocol = protocol\n"
- + "!Sample_data_processing = processing\n"
- + "!Sample_library_strategy = application\n"
- + "!Sample_library_source = genomic\n"
- + "!Sample_library_selection = CHIP\n"
- + "!Sample_instrument_model = my-analyzer\n"
- + "!Sample_raw_file_1 = sample.srf\n"
+ + "!Sample_data_processing = processing\n" + "!Sample_library_strategy = RNA-Seq\n"
+ + "!Sample_library_source = genomic\n" + "!Sample_library_selection = cDNA\n"
+ + "!Sample_instrument_model = my-analyzer\n" + "!Sample_raw_file_1 = sample.srf\n"
+ "!Sample_raw_file_type_1 = srf\n"
+ "!Sample_file_checksum_1 = fc3ff98e8c6a0d3087d515c0473f8677\n", attachmentContent);
context.assertIsSatisfied();
@@ -202,7 +212,7 @@ public class DataSetToSOFTTest extends AbstractFileSystemTestCase
template.bind("external-sample-name", "my sample");
return template;
}
-
+
private void prepareGetDataSet(final String dataSetCode, final Sample flowLaneSample)
{
context.checking(new Expectations()
@@ -216,18 +226,19 @@ public class DataSetToSOFTTest extends AbstractFileSystemTestCase
}
});
}
-
+
private void prepareGetSample(final Sample sample)
{
context.checking(new Expectations()
- {
{
- one(service).tryGetSampleWithExperiment(SampleIdentifierFactory.parse(sample.getIdentifier()));
- will(returnValue(sample));
- }
- });
+ {
+ one(service).tryGetSampleWithExperiment(
+ SampleIdentifierFactory.parse(sample.getIdentifier()));
+ will(returnValue(sample));
+ }
+ });
}
-
+
private void addProperty(Sample sample, String propertyTypeCode, String value)
{
List<IEntityProperty> properties = sample.getProperties();
@@ -243,7 +254,7 @@ public class DataSetToSOFTTest extends AbstractFileSystemTestCase
property.setValue(value);
properties.add(property);
}
-
+
private Sample sample(long id, String identifier)
{
Sample sample = new Sample();
@@ -252,4 +263,3 @@ public class DataSetToSOFTTest extends AbstractFileSystemTestCase
return sample;
}
}
-
--
GitLab