Handle more cases for splitting datasets

Make splitting the experiment transactional by isolating the relevant
operations in a helper service.

Use generic BulkDataSlicerUtils and SingleCellSlicerUtils for slicing
samples in splits which handle all the possible fixed-size encoding
cases and single-cell data slicing.

Add options to the splitExperiment CLI to skip post-processing and
delete the original experiment automatically.

Author: arteymix

gemma-cli/src/main/java/ubic/gemma/apps/SplitExperimentCli.java

@@ -20,19 +20,14 @@
 package ubic.gemma.apps;
 
 import org.apache.commons.cli.CommandLine;
-import org.apache.commons.cli.Option;
 import org.apache.commons.cli.Options;
 import org.apache.commons.cli.ParseException;
 import org.springframework.beans.factory.annotation.Autowired;
+import ubic.gemma.cli.util.EntityLocator;
 import ubic.gemma.core.analysis.preprocess.SplitExperimentService;
 import ubic.gemma.model.analysis.expression.ExpressionExperimentSet;
-import ubic.gemma.model.expression.experiment.ExperimentFactorUtils;
 import ubic.gemma.model.expression.experiment.ExperimentalFactor;
-import ubic.gemma.model.expression.experiment.ExperimentalFactorValueObject;
 import ubic.gemma.model.expression.experiment.ExpressionExperiment;
-import ubic.gemma.persistence.service.expression.experiment.ExperimentalFactorService;
-
-import java.util.Collection;
 
 /**
  * Split an experiment into parts based on an experimental factor
@@ -41,18 +36,18 @@
  */
 public class SplitExperimentCli extends ExpressionExperimentManipulatingCLI {
 
-    /**
-     *
-     */
-    private static final String FACTOR_OPTION = "factor";
+    private static final String FACTOR_OPTION = "factor",
+            SKIP_POST_PROCESSING_OPTION = "nopost",
+            DELETE_ORIGINAL_EXPERIMENT_OPTION = "deleteOriginalExperiment";
 
     @Autowired
     private SplitExperimentService serv;
     @Autowired
-    private ExperimentalFactorService efs;
+    private EntityLocator entityLocator;
 
-    private Long factorId;
-    private String factorName;
+    private String factorIdentifier;
+    private boolean skipPostProcessing;
+    private boolean deleteOriginalExperiment;
 
     public SplitExperimentCli() {
         super();
@@ -71,80 +66,23 @@ public String getShortDesc() {
 
     @Override
     protected void buildExperimentOptions( Options options ) {
-        options.addOption( Option.builder( FACTOR_OPTION ).hasArg()
-                .desc( "ID numbers, categories or names of the factor to use, with spaces replaced by underscores (must not be 'batch')" )
-                .build() );
+        options.addRequiredOption( FACTOR_OPTION, "factor", true, "ID numbers, categories or names of the factor to use, with spaces replaced by underscores (must not be 'batch')" );
+        options.addOption( SKIP_POST_PROCESSING_OPTION, "no-post-processing", false, "Skip post-processing of resulting splits if applicable." );
+        options.addOption( DELETE_ORIGINAL_EXPERIMENT_OPTION, "delete-original-experiment", false, "Delete the original experiment once the split succeeds." );
     }
 
     @Override
     protected void processExperimentOptions( CommandLine commandLine ) throws ParseException {
-        if ( !commandLine.hasOption( FACTOR_OPTION ) ) {
-            throw new IllegalArgumentException( "Please specify the factor" );
-        }
-        String rawFactor = commandLine.getOptionValue( FACTOR_OPTION );
-        try {
-            this.factorId = Long.parseLong( rawFactor );
-        } catch ( NumberFormatException e ) {
-            this.factorName = rawFactor;
-        }
+        factorIdentifier = commandLine.getOptionValue( FACTOR_OPTION );
+        skipPostProcessing = commandLine.hasOption( SKIP_POST_PROCESSING_OPTION );
+        deleteOriginalExperiment = commandLine.hasOption( DELETE_ORIGINAL_EXPERIMENT_OPTION );
     }
 
     @Override
     protected void processExpressionExperiment( ExpressionExperiment ee ) {
         ee = this.eeService.thawLite( ee );
-        ExperimentalFactor splitOn = this.guessFactor( ee );
-        ExpressionExperimentSet eeSet = serv.split( ee, splitOn, true );
+        ExperimentalFactor splitOn = entityLocator.locateExperimentalFactor( ee, factorIdentifier );
+        ExpressionExperimentSet eeSet = serv.split( ee, splitOn, !skipPostProcessing, deleteOriginalExperiment );
         addSuccessObject( ee, "Experiment was split on " + splitOn + " into " + eeSet.getExperiments().size() + " parts." );
     }
-
-    /**
-     * Adapted from code in DifferentialExpressionAnalysisCli
-     */
-    private ExperimentalFactor guessFactor( ExpressionExperiment ee ) {
-        if ( ee.getExperimentalDesign() == null ) {
-            throw new IllegalStateException( ee + " does not have an experimental design, it cannot be split on a factor." );
-        }
-
-        if ( this.factorName != null ) {
-
-            Collection<ExperimentalFactor> experimentalFactors = ee.getExperimentalDesign().getExperimentalFactors();
-            for ( ExperimentalFactor experimentalFactor : experimentalFactors ) {
-
-                // has already implemented way of figuring out human-friendly name of factor value.
-                ExperimentalFactorValueObject fvo = new ExperimentalFactorValueObject( experimentalFactor );
-
-                // do not attempt to switch on 'batch'
-                if ( ExperimentFactorUtils.isBatchFactor( experimentalFactor ) ) {
-                    continue;
-                }
-
-                if ( factorName.contains( experimentalFactor.getName().replaceAll( " ", "_" ) ) ) {
-                    return experimentalFactor;
-                } else if ( fvo.getCategory() != null && factorName
-                        .contains( fvo.getCategory().replaceAll( " ", "_" ) ) ) {
-                    return experimentalFactor;
-                }
-            }
-
-            throw new IllegalArgumentException( "Didn't find factor the provided factor name " );
-
-        }
-
-        ExperimentalFactor factor = efs.loadOrFail( factorId );
-        factor = efs.thaw( factor );
-        if ( factor == null ) {
-            throw new IllegalArgumentException( "No factor for id=" + factorId );
-        }
-        if ( !factor.getExperimentalDesign().equals( ee.getExperimentalDesign() ) ) {
-            throw new IllegalArgumentException( "Factor with id=" + factorId + " does not belong to " + ee );
-        }
-
-        if ( ExperimentFactorUtils.isBatchFactor( factor ) ) {
-            throw new IllegalArgumentException( "Selected factor looks like batch, split not allowed, choose another factor instead" );
-        }
-
-        return factor;
-
-    }
-
 }