Merge pull request #2379 from opencb/TASK-4158

TASK-4158

opencga-analysis/src/main/java/org/opencb/opencga/analysis/clinical/ClinicalAnalysisLoadTask.java

@@ -0,0 +1,63 @@
+package org.opencb.opencga.analysis.clinical;
+
+import org.apache.commons.lang3.StringUtils;
+import org.opencb.opencga.analysis.tools.OpenCgaToolScopeStudy;
+import org.opencb.opencga.catalog.managers.FileManager;
+import org.opencb.opencga.catalog.models.ClinicalAnalysisLoadResult;
+import org.opencb.opencga.core.exceptions.ToolException;
+import org.opencb.opencga.core.models.clinical.ClinicalAnalysisLoadParams;
+import org.opencb.opencga.core.models.common.Enums;
+import org.opencb.opencga.core.models.file.File;
+import org.opencb.opencga.core.tools.annotations.Tool;
+import org.opencb.opencga.core.tools.annotations.ToolParams;
+
+import java.nio.file.Path;
+import java.nio.file.Paths;
+import java.util.Map;
+
+@Tool(id = ClinicalAnalysisLoadTask.ID, resource = Enums.Resource.CLINICAL_ANALYSIS, description = ClinicalAnalysisLoadTask.DESCRIPTION)
+public class ClinicalAnalysisLoadTask extends OpenCgaToolScopeStudy {
+    public final static String ID = "load";
+    public static final String DESCRIPTION = "Load clinical analyses from a file";
+
+    private Path filePath;
+
+    @ToolParams
+    protected ClinicalAnalysisLoadParams params = new ClinicalAnalysisLoadParams();
+
+    @Override
+    protected void check() throws Exception {
+        super.check();
+
+        String fileStr = params.getFile();
+        if (StringUtils.isEmpty(fileStr)) {
+            throw new ToolException("Missing input file when loading clinical analyses.");
+        }
+
+        File file = catalogManager.getFileManager().get(getStudy(), fileStr, FileManager.INCLUDE_FILE_URI_PATH, token).first();
+        filePath = Paths.get(file.getUri());
+        if (!filePath.toFile().exists()) {
+            throw new ToolException("Input file '" + fileStr + "' does not exist: " + filePath);
+        }
+    }
+
+    @Override
+    protected void run() throws Exception {
+        step(() -> {
+            ClinicalAnalysisLoadResult loadResult = catalogManager.getClinicalAnalysisManager().load(getStudy(), filePath, token);
+
+            // Add results as attributes
+            addAttribute("Num. clinical analyses loaded", loadResult.getNumLoaded());
+            addAttribute("Num. clinical analyses not loaded", loadResult.getFailures().size());
+            addAttribute("Loading time (in sec.)", loadResult.getTime());
+            addAttribute("Clinical analyses file name", loadResult.getFilename());
+
+            // Add warnings with the not loaded clinical analysis
+            if (loadResult.getFailures().size() > 0) {
+                for (Map.Entry<String, String> entry : loadResult.getFailures().entrySet()) {
+                    addWarning("Clinical analysis " + entry.getKey() + " could not be loaded due to error: " + entry.getValue());
+                }
+            }
+        });
+    }
+}

opencga-analysis/src/main/java/org/opencb/opencga/analysis/tools/ToolFactory.java

@@ -37,14 +37,16 @@ public class ToolFactory {
     private static Map<String, Set<Class<? extends OpenCgaTool>>> duplicatedTools;
     private static List<Class<? extends OpenCgaTool>> toolsList;
 
-    private static synchronized Map<String, Class<? extends OpenCgaTool>> loadTools() {
+    public static final String DEFAULT_PACKAGE = "org.opencb.opencga";
+
+    private static synchronized Map<String, Class<? extends OpenCgaTool>> loadTools(List<String> packages) {
         if (toolsCache == null) {
             Reflections reflections = new Reflections(new ConfigurationBuilder()
                     .setScanners(
                             new SubTypesScanner(),
                             new TypeAnnotationsScanner().filterResultsBy(s -> StringUtils.equals(s, Tool.class.getName()))
                     )
-                    .addUrls(getUrls())
+                    .addUrls(getUrlsFromPackages(packages))
                     .filterInputsBy(input -> input != null && input.endsWith(".class"))
             );
 
@@ -85,9 +87,20 @@ private static synchronized Map<String, Class<? extends OpenCgaTool>> loadTools(
         }
         return toolsCache;
     }
+    static Collection<URL> getUrlsFromPackages(List<String> packages) {
+        Collection<URL> urls = new LinkedList<>();
+        for (String pack :packages){
+            for (URL url : ClasspathHelper.forPackage(pack)) {
+                String name = url.getPath().substring(url.getPath().lastIndexOf('/') + 1);
+                if (name.isEmpty() || (name.contains("opencga") && !name.contains("opencga-storage-hadoop-deps"))) {
+                    urls.add(url);
+                }
+            }
+        }
+        return urls;
+    }
 
     static Collection<URL> getUrls() {
-        // TODO: What if there are third party libraries that implement Tools?
         //  Currently they must contain "opencga" in the jar name.
         //  e.g.  acme-rockets-opencga-5.4.0.jar
         Collection<URL> urls = new LinkedList<>();
@@ -101,6 +114,10 @@ static Collection<URL> getUrls() {
     }
 
     public final Class<? extends OpenCgaTool> getToolClass(String toolId) throws ToolException {
+        return getToolClass(toolId, Collections.singletonList(DEFAULT_PACKAGE));
+    }
+
+    public final Class<? extends OpenCgaTool> getToolClass(String toolId, List<String> packages) throws ToolException {
         Objects.requireNonNull(toolId);
 
         Class<? extends OpenCgaTool> aClass = null;
@@ -112,7 +129,7 @@ public final Class<? extends OpenCgaTool> getToolClass(String toolId) throws Too
         } catch (ClassNotFoundException ignore) {
         }
         if (aClass == null) {
-            aClass = loadTools().get(toolId);
+            aClass = loadTools(packages).get(toolId);
         }
         if (aClass == null) {
             throw new ToolException("Tool '" + toolId + "' not found");
@@ -121,11 +138,19 @@ public final Class<? extends OpenCgaTool> getToolClass(String toolId) throws Too
     }
 
     public Tool getTool(String toolId) throws ToolException {
-        return getToolClass(toolId).getAnnotation(Tool.class);
+        return getTool(toolId, Collections.singletonList(DEFAULT_PACKAGE));
+    }
+
+    public Tool getTool(String toolId, List<String> packages) throws ToolException {
+        return getToolClass(toolId, packages).getAnnotation(Tool.class);
     }
 
     public final OpenCgaTool createTool(String toolId) throws ToolException {
-        return createTool(getToolClass(toolId));
+        return createTool(toolId, Collections.singletonList(DEFAULT_PACKAGE));
+    }
+
+    public final OpenCgaTool createTool(String toolId, List<String> packages) throws ToolException {
+        return createTool(getToolClass(toolId, packages));
     }
 
     public final OpenCgaTool createTool(Class<? extends OpenCgaTool> aClass) throws ToolException {
@@ -141,12 +166,22 @@ public final OpenCgaTool createTool(Class<? extends OpenCgaTool> aClass) throws
     }
 
     public Collection<Class<? extends OpenCgaTool>> getTools() {
-        loadTools();
+        loadTools(Collections.singletonList(DEFAULT_PACKAGE));
+        return toolsList;
+    }
+
+    public Collection<Class<? extends OpenCgaTool>> getTools(List<String> packages) {
+        loadTools(packages);
         return toolsList;
     }
 
     public Map<String, Set<Class<? extends OpenCgaTool>>> getDuplicatedTools() {
-        loadTools();
+        loadTools(Collections.singletonList(DEFAULT_PACKAGE));
+        return duplicatedTools;
+    }
+
+    public Map<String, Set<Class<? extends OpenCgaTool>>> getDuplicatedTools(List<String> packages) {
+        loadTools(packages);
         return duplicatedTools;
     }
 }

opencga-analysis/src/main/java/org/opencb/opencga/analysis/tools/ToolRunner.java

@@ -16,12 +16,14 @@
 
 package org.opencb.opencga.analysis.tools;
 
+import org.apache.commons.collections4.CollectionUtils;
 import org.opencb.commons.datastore.core.ObjectMap;
 import org.opencb.commons.datastore.core.QueryOptions;
 import org.opencb.opencga.analysis.variant.manager.VariantStorageManager;
 import org.opencb.opencga.catalog.exceptions.CatalogException;
 import org.opencb.opencga.catalog.managers.CatalogManager;
 import org.opencb.opencga.core.api.ParamConstants;
+import org.opencb.opencga.core.config.Configuration;
 import org.opencb.opencga.core.exceptions.ToolException;
 import org.opencb.opencga.core.models.job.Job;
 import org.opencb.opencga.core.tools.ToolParams;
@@ -32,6 +34,7 @@
 
 import java.nio.file.Path;
 import java.nio.file.Paths;
+import java.util.Collections;
 
 public class ToolRunner {
 
@@ -42,16 +45,30 @@ public class ToolRunner {
     private final String opencgaHome;
     private final ToolFactory toolFactory;
 
+    private final Configuration configuration;
+
     public ToolRunner(String opencgaHome, CatalogManager catalogManager, StorageEngineFactory storageEngineFactory) {
         this(opencgaHome, catalogManager, new VariantStorageManager(catalogManager, storageEngineFactory));
     }
 
+    public ToolRunner(String opencgaHome, CatalogManager catalogManager, StorageEngineFactory storageEngineFactory,
+                      Configuration configuration) {
+        this(opencgaHome, catalogManager, new VariantStorageManager(catalogManager, storageEngineFactory), configuration);
+    }
+
     public ToolRunner(String opencgaHome, CatalogManager catalogManager, VariantStorageManager variantStorageManager) {
+        this(opencgaHome, catalogManager, variantStorageManager, null);
+    }
+
+    public ToolRunner(String opencgaHome, CatalogManager catalogManager, VariantStorageManager variantStorageManager,
+                      Configuration configuration) {
         this.opencgaHome = opencgaHome;
         this.catalogManager = catalogManager;
 
         this.variantStorageManager = variantStorageManager;
         this.toolFactory = new ToolFactory();
+
+        this.configuration = configuration;
     }
 
     /**
@@ -102,9 +119,14 @@ public ExecutionResult execute(Job job, Path outDir, String token) throws Catalo
      * @throws ToolException if the execution fails
      */
     public ExecutionResult execute(String toolId, ObjectMap params, Path outDir, String jobId, String token) throws ToolException {
-        return toolFactory
-                .createTool(toolId)
-                .setUp(opencgaHome, catalogManager, variantStorageManager, params, outDir, jobId, token)
+        OpenCgaTool tool;
+        if (configuration != null && configuration.getAnalysis() != null
+                && CollectionUtils.isNotEmpty(configuration.getAnalysis().getPackages())) {
+            tool = toolFactory.createTool(toolId, configuration.getAnalysis().getPackages());
+        } else {
+            tool = toolFactory.createTool(toolId);
+        }
+        return tool.setUp(opencgaHome, catalogManager, variantStorageManager, params, outDir, jobId, token)
                 .start();
     }
 
@@ -177,6 +199,7 @@ public ExecutionResult execute(Class<? extends OpenCgaTool> tool, ToolParams too
      * @throws ToolException if the execution fails
      */
     public ExecutionResult execute(Class<? extends OpenCgaTool> tool, ObjectMap params, Path outDir, String jobId, String token) throws ToolException {
+
         return toolFactory
                 .createTool(tool)
                 .setUp(opencgaHome, catalogManager, variantStorageManager, params, outDir, jobId, token)

opencga-analysis/src/test/java/org/opencb/opencga/analysis/variant/VariantAnalysisTest.java

@@ -39,6 +39,7 @@
 import org.opencb.commons.datastore.core.Query;
 import org.opencb.commons.datastore.core.QueryOptions;
 import org.opencb.opencga.TestParamConstants;
+import org.opencb.opencga.analysis.clinical.ClinicalAnalysisLoadTask;
 import org.opencb.opencga.analysis.tools.ToolRunner;
 import org.opencb.opencga.analysis.variant.gwas.GwasAnalysis;
 import org.opencb.opencga.analysis.variant.hrdetect.HRDetectAnalysis;
@@ -62,12 +63,15 @@
 import org.opencb.opencga.core.config.storage.CellBaseConfiguration;
 import org.opencb.opencga.core.config.storage.StorageConfiguration;
 import org.opencb.opencga.core.exceptions.ToolException;
+import org.opencb.opencga.core.models.clinical.ClinicalAnalysis;
+import org.opencb.opencga.core.models.clinical.ClinicalAnalysisLoadParams;
 import org.opencb.opencga.core.models.cohort.Cohort;
 import org.opencb.opencga.core.models.cohort.CohortCreateParams;
 import org.opencb.opencga.core.models.cohort.CohortUpdateParams;
 import org.opencb.opencga.core.models.common.AnnotationSet;
 import org.opencb.opencga.core.models.family.Family;
 import org.opencb.opencga.core.models.file.File;
+import org.opencb.opencga.core.models.file.FileLinkParams;
 import org.opencb.opencga.core.models.individual.Individual;
 import org.opencb.opencga.core.models.individual.IndividualInternal;
 import org.opencb.opencga.core.models.individual.Location;
@@ -1063,6 +1067,46 @@ public void testPedigreeGraph() throws CatalogException {
         assertEquals(base64, family.getPedigreeGraph().getBase64());
     }
 
+    @Test
+    public void testClinicalAnalysisLoading() throws IOException, ToolException, CatalogException {
+        String fileStr = "clinical_analyses.json.gz";
+
+        String gzFile = getClass().getResource("/biofiles/" + fileStr).getFile();
+        File file = catalogManager.getFileManager().link(CANCER_STUDY, new FileLinkParams(gzFile, "ca", "", "", null, null, null, null,
+                null), true, token).first();
+        System.out.println("file ID = " + file.getId());
+        System.out.println("file name = " + file.getName());
+
+        // Run clinical analysis load task
+        Path loadingOutDir = Paths.get(opencga.createTmpOutdir("_clinical_analysis_outdir"));
+        System.out.println("Clinical analysis load task out dir = " + loadingOutDir);
+
+        ClinicalAnalysisLoadParams params = new ClinicalAnalysisLoadParams();
+        params.setFile(file.getId());
+
+        toolRunner.execute(ClinicalAnalysisLoadTask.class, params, new ObjectMap(ParamConstants.STUDY_PARAM,
+                CANCER_STUDY), loadingOutDir, null, token);
+
+        String ca1Id = "SAP-45016-1";
+        String ca2Id = "OPA-6607-1";
+
+        Query query = new Query();
+        OpenCGAResult<ClinicalAnalysis> result = catalogManager.getClinicalAnalysisManager().search(CANCER_STUDY, query, QueryOptions.empty(),
+                token);
+        Assert.assertTrue(result.getResults().stream().map(ca -> ca.getId()).collect(Collectors.toList()).contains(ca1Id));
+        Assert.assertTrue(result.getResults().stream().map(ca -> ca.getId()).collect(Collectors.toList()).contains(ca2Id));
+
+        query.put("id", ca1Id);
+        ClinicalAnalysis clinicalAnalysis = catalogManager.getClinicalAnalysisManager().search(CANCER_STUDY, query, QueryOptions.empty(),
+                token).first();
+        Assert.assertEquals(ca1Id, clinicalAnalysis.getId());
+
+        query.put("id", ca2Id);
+        clinicalAnalysis = catalogManager.getClinicalAnalysisManager().search(CANCER_STUDY, query, QueryOptions.empty(),
+                token).first();
+        Assert.assertEquals(ca2Id, clinicalAnalysis.getId());
+    }
+
     @Test
     public void testCellbaseConfigure() throws Exception {
         String project = "Project_test_cellbase_configure";

opencga-app/app/cloud/docker/compose/conf/configuration.yml

@@ -107,6 +107,8 @@ healthCheck:
 
 
 analysis:
+  packages: # List of packages where to find analysis tools
+    - "org.opencb.opencga"
   scratchDir: ""    # Scratch folder for the analysis.
   execution:
     # Accepted values are "local", "SGE", "azure-batch", "k8s"

opencga-app/app/misc/clients/r_client_generator.py

@@ -26,7 +26,7 @@ def __init__(self, server_url, output_dir):
             'Analysis - Clinical': 'Clinical',
             'Operations - Variant Storage': 'Operation',
             'Meta': 'Meta',
-            'Cva': 'Cva',
+            'Cvdb': 'Cvdb',
             'GA4GH': 'GA4GH',
             'Admin': 'Admin'
         }

opencga-app/app/misc/clients/rest_client_generator.py

@@ -50,7 +50,7 @@ def __init__(self, rest_api_file, output_dir):
             'Analysis - Clinical': 'ClinicalAnalysis',
             'Operations - Variant Storage': 'VariantOperation',
             'Meta': 'Meta',
-            'Cva': 'Cva',
+            'Cvdb': 'Cvdb',
             'GA4GH': 'GA4GH',
             'Admin': 'Admin'
         }

opencga-app/src/main/java/org/opencb/opencga/app/cli/internal/InternalCliOptionsParser.java

@@ -35,6 +35,7 @@
 import static org.opencb.opencga.app.cli.internal.options.AlignmentCommandOptions.SamtoolsCommandOptions.SAMTOOLS_RUN_COMMAND;
 import static org.opencb.opencga.app.cli.internal.options.ClinicalCommandOptions.CancerTieringCommandOptions.CANCER_TIERING_INTERPRETATION_RUN_COMMAND;
 import static org.opencb.opencga.app.cli.internal.options.ClinicalCommandOptions.ExomiserInterpretationCommandOptions.EXOMISER_INTERPRETATION_RUN_COMMAND;
+import static org.opencb.opencga.app.cli.internal.options.ClinicalCommandOptions.ImportClinicalAnalysesCommandOptions.IMPORT_COMMAND;
 import static org.opencb.opencga.app.cli.internal.options.ClinicalCommandOptions.RgaAuxiliarSecondaryIndexCommandOptions.RGA_AUX_INDEX_RUN_COMMAND;
 import static org.opencb.opencga.app.cli.internal.options.ClinicalCommandOptions.RgaSecondaryIndexCommandOptions.RGA_INDEX_RUN_COMMAND;
 import static org.opencb.opencga.app.cli.internal.options.ClinicalCommandOptions.TeamCommandOptions.TEAM_INTERPRETATION_RUN_COMMAND;
@@ -226,6 +227,7 @@ public InternalCliOptionsParser() {
         clinicalSubCommands.addCommand(RGA_INDEX_RUN_COMMAND, clinicalCommandOptions.rgaSecondaryIndexCommandOptions);
         clinicalSubCommands.addCommand(RGA_AUX_INDEX_RUN_COMMAND, clinicalCommandOptions.rgaAuxiliarSecondaryIndexCommandOptions);
         clinicalSubCommands.addCommand(EXOMISER_INTERPRETATION_RUN_COMMAND, clinicalCommandOptions.exomiserInterpretationCommandOptions);
+        clinicalSubCommands.addCommand(IMPORT_COMMAND, clinicalCommandOptions.importClinicalAnalysesCommandOptions);
         clinicalSubCommands.addCommand("tsv-load", clinicalCommandOptions.tsvLoad);
 
         fileCommandOptions = new FileCommandOptions(commonCommandOptions, jCommander);