Refactor patient and sample removal code

cBioPortal · Jun 27, 2024 · bd54ed2 · bd54ed2
1 parent 5753d1d
commit bd54ed2
Show file tree

Hide file tree

Showing 5 changed files with 94 additions and 81 deletions.
diff --git a/src/main/java/org/mskcc/cbio/portal/dao/DaoGeneticAlteration.java b/src/main/java/org/mskcc/cbio/portal/dao/DaoGeneticAlteration.java
@@ -493,59 +493,4 @@ public void deleteAllRecords() throws DaoException {
             JdbcUtil.closeAll(DaoGeneticAlteration.class, con, pstmt, rs);
         }
     }
-
-    /**
-     * Removes sample in genetic alterations' data for a study
-     * @param internalStudyId - internal id of study to remove samples in genetic alterations data
-     * @param internalSampleIdsToRemove - internal ids of samples to remove
-     * @throws DaoException
-     */
-    public void removeSamplesInGeneticAlterationsForStudy(int internalStudyId, Set<Integer> internalSampleIdsToRemove) throws DaoException {
-        List<GeneticProfile> geneticProfiles = DaoGeneticProfile.getAllGeneticProfiles(internalStudyId);
-        for (GeneticProfile geneticProfile : geneticProfiles) {
-            Set<Integer> removedInternalSampleIds = removeSamplesInGeneticAlterationsForGeneticProfile(geneticProfile, internalSampleIdsToRemove);
-            log.debug("Genetic alterations data for {} sample ids ouf of {} requested have been removed for genetic profile with stable id={}",
-                    removedInternalSampleIds, internalSampleIdsToRemove, geneticProfile.getStableId());
-        }
-    }
-
-    /**
-     * Removes sample in genetic alterations' data for a genetic profile
-     * @param geneticProfile - genetic profile to remove samples in genetic alteration data
-     * @param internalSampleIdsToRemove - internal ids of samples to remove
-     * @return set of sample internal ids that were actually removed
-     * @throws DaoException
-     */
-    public Set<Integer> removeSamplesInGeneticAlterationsForGeneticProfile(GeneticProfile geneticProfile, Set<Integer> internalSampleIdsToRemove) throws DaoException {
-        int geneticProfileId = geneticProfile.getGeneticProfileId();
-        List<Integer> orderedSampleList = DaoGeneticProfileSamples.getOrderedSampleList(geneticProfileId);
-        Set<Integer> actualInternalSampleIdsToRemove = orderedSampleList.stream()
-                .filter(internalSampleIdsToRemove::contains).collect(Collectors.toUnmodifiableSet());
-        if (!actualInternalSampleIdsToRemove.isEmpty()) {
-            if (GeneticAlterationType.GENESET_SCORE.equals(geneticProfile.getGeneticAlterationType())) {
-                List<String> sampleStableIds = actualInternalSampleIdsToRemove.stream()
-                        .map(internalSampleID ->
-                                DaoSample.getSampleById(internalSampleID).getStableId())
-                        .toList();
-                throw new RuntimeException("Sample(s) with stable id "
-                        + String.join(", ", sampleStableIds)
-                        + " can't be removed as it contains GSVA data." +
-                        " Consider dropping and re-uploading the whole study.");
-            }
-            orderedSampleList.removeAll(actualInternalSampleIdsToRemove);
-            HashMap<Integer, HashMap<Integer, String>> geneticAlterationMapForEntityIds = DaoGeneticAlteration.getInstance().getGeneticAlterationMapForEntityIds(geneticProfileId, null);
-            DaoGeneticAlteration.getInstance().deleteAllRecordsInGeneticProfile(geneticProfileId);
-            if (!orderedSampleList.isEmpty()) {
-                for (Map.Entry<Integer, HashMap<Integer, String>> entry : geneticAlterationMapForEntityIds.entrySet()) {
-                    String[] values = orderedSampleList.stream().map(isid -> entry.getValue().get(isid)).toArray(String[]::new);
-                    DaoGeneticAlteration.getInstance().addGeneticAlterationsForGeneticEntity(geneticProfileId, entry.getKey(), values);
-                }
-            }
-            DaoGeneticProfileSamples.deleteAllSamplesInGeneticProfile(geneticProfileId);
-            if (!orderedSampleList.isEmpty()) {
-                DaoGeneticProfileSamples.addGeneticProfileSamples(geneticProfileId, orderedSampleList);
-            }
-        }
-        return actualInternalSampleIdsToRemove;
-    }
 }
diff --git a/src/main/java/org/mskcc/cbio/portal/dao/DaoPatient.java b/src/main/java/org/mskcc/cbio/portal/dao/DaoPatient.java
@@ -221,23 +221,26 @@ private static Patient extractPatient(ResultSet rs) throws SQLException
 		}
     }
 
-    public static void removePatientsEverywhereInStudy(int internalStudyId, Set<String> patientStableIds) throws DaoException {
-        log.info("Removing {} patients from study with internal id={} ...", patientStableIds, internalStudyId);
-        Set<Integer> internalPatientIds = findInternalPatientIdsInStudy(internalStudyId, patientStableIds);
-        for (Integer internalPatientId : internalPatientIds) {
-            Set<Integer> patientInternalSampleIds = DaoSample.getSamplesByPatientId(internalPatientId).stream()
-                    .map(Sample::getInternalId).collect(Collectors.toSet());
-            DaoGeneticAlteration.getInstance().removeSamplesInGeneticAlterationsForStudy(internalStudyId, patientInternalSampleIds);
-        }
-        DaoPatient.deletePatients(internalPatientIds);
-        log.info("Removing {} patients from study with internal id={} done.", patientStableIds, internalStudyId);
-    }
-
-    public static void deletePatients(Collection<Integer> internalPatientIds) throws DaoException
+    /**
+     * Removes patients information from the study
+     * @param internalStudyId - id of the study that contains the patients
+     * @param patientStableIds - patient stable ids to remove
+     * @throws DaoException
+     */
+    public static void deletePatients(int internalStudyId, Set<String> patientStableIds) throws DaoException
     {
-        if (internalPatientIds == null || internalPatientIds.isEmpty()) {
+        if (patientStableIds == null || patientStableIds.isEmpty()) {
+            log.info("No patients specified to remove for study with internal id={}. Skipping.", internalStudyId);
             return;
         }
+        log.info("Removing {} patients from study with internal id={} ...", patientStableIds, internalStudyId);
+
+        Set<Integer> internalPatientIds = findInternalPatientIdsInStudy(internalStudyId, patientStableIds);
+        Set<String> patientsSampleStableIds = internalPatientIds.stream().flatMap(internalPatientId ->
+                DaoSample.getSamplesByPatientId(internalPatientId).stream().map(Sample::getStableId))
+                .collect(Collectors.toSet());
+        DaoSample.deleteSamples(internalStudyId, patientsSampleStableIds);
+
         Connection con = null;
         PreparedStatement pstmt = null;
         try {
@@ -257,6 +260,7 @@ public static void deletePatients(Collection<Integer> internalPatientIds) throws
         finally {
             JdbcUtil.closeAll(DaoPatient.class, con, pstmt, null);
         }
+        log.info("Removing {} patients from study with internal id={} done.", patientStableIds, internalStudyId);
     }
 
     public static Set<Integer> findInternalPatientIdsInStudy(Integer internalStudyId, Set<String> patientStableIds) {

diff --git a/src/main/java/org/mskcc/cbio/portal/dao/DaoSample.java b/src/main/java/org/mskcc/cbio/portal/dao/DaoSample.java
@@ -32,6 +32,8 @@
 
 package org.mskcc.cbio.portal.dao;
 
+import org.mskcc.cbio.portal.model.GeneticAlterationType;
+import org.mskcc.cbio.portal.model.GeneticProfile;
 import org.mskcc.cbio.portal.model.Patient;
 import org.mskcc.cbio.portal.model.Sample;
 import org.mskcc.cbio.portal.util.ProgressMonitor;
@@ -52,6 +54,7 @@
 import java.util.Map;
 import java.util.NoSuchElementException;
 import java.util.Set;
+import java.util.stream.Collectors;
 
 /**
  * DAO to `sample`.
@@ -279,11 +282,25 @@ public static void deleteAllRecords() throws DaoException
 
         clearCache();
     }
-    public static void deleteSamples(Collection<Integer> internalSampleIds) throws DaoException
+
+    /**
+     * Remove set of samples from the study
+     * @param internalStudyId - id of the study that contains the samples
+     * @param sampleStableIds - sample stable ids of samples to remove
+     * @throws DaoException
+     */
+    public static void deleteSamples(int internalStudyId, Set<String> sampleStableIds) throws DaoException
     {
-        if (internalSampleIds == null || internalSampleIds.isEmpty()) {
+        if (sampleStableIds == null || sampleStableIds.isEmpty()) {
+            log.info("No samples specified to remove for study with internal id={}. Skipping.", internalStudyId);
             return;
         }
+
+        log.info("Removing {} samples from study with internal id={} ...", sampleStableIds, internalStudyId);
+
+        Set<Integer> internalSampleIds = findInternalSampleIdsInStudy(internalStudyId, sampleStableIds);
+        removeSamplesInGeneticAlterationsForStudy(internalStudyId, internalSampleIds);
+
         Connection con = null;
         PreparedStatement pstmt = null;
         try {
@@ -303,6 +320,7 @@ public static void deleteSamples(Collection<Integer> internalSampleIds) throws D
         finally {
             JdbcUtil.closeAll(DaoSample.class, con, pstmt, null);
         }
+        log.info("Removing {} samples from study with internal id={} done.", sampleStableIds, internalStudyId);
     }
 
     private static Sample extractSample(ResultSet rs) throws SQLException
@@ -312,12 +330,59 @@ private static Sample extractSample(ResultSet rs) throws SQLException
                           rs.getInt("PATIENT_ID"));
     }
 
-    public static void removeSamplesEverywhereInStudy(int internalStudyId, Set<String> sampleStableIds) throws DaoException {
-        log.info("Removing {} samples from study with internal id={} ...", sampleStableIds, internalStudyId);
-        Set<Integer> internalSampleIds = findInternalSampleIdsInStudy(internalStudyId, sampleStableIds);
-        DaoGeneticAlteration.getInstance().removeSamplesInGeneticAlterationsForStudy(internalStudyId, internalSampleIds);
-        DaoSample.deleteSamples(internalSampleIds);
-        log.info("Removing {} samples from study with internal id={} done.", sampleStableIds, internalStudyId);
+    /**
+     * Removes sample in genetic alterations' data for a study
+     * @param internalStudyId - internal id of study to remove samples in genetic alterations data
+     * @param internalSampleIdsToRemove - internal ids of samples to remove
+     * @throws DaoException
+     */
+    private static void removeSamplesInGeneticAlterationsForStudy(int internalStudyId, Set<Integer> internalSampleIdsToRemove) throws DaoException {
+        List<GeneticProfile> geneticProfiles = DaoGeneticProfile.getAllGeneticProfiles(internalStudyId);
+        for (GeneticProfile geneticProfile : geneticProfiles) {
+            Set<Integer> removedInternalSampleIds = removeSamplesInGeneticAlterationsForGeneticProfile(geneticProfile, internalSampleIdsToRemove);
+            log.debug("Genetic alterations data for {} sample ids ouf of {} requested have been removed for genetic profile with stable id={}",
+                    removedInternalSampleIds, internalSampleIdsToRemove, geneticProfile.getStableId());
+        }
+    }
+
+    /**
+     * Removes sample in genetic alterations' data for a genetic profile
+     * @param geneticProfile - genetic profile to remove samples in genetic alteration data
+     * @param internalSampleIdsToRemove - internal ids of samples to remove
+     * @return set of sample internal ids that were actually removed
+     * @throws DaoException
+     */
+    private static Set<Integer> removeSamplesInGeneticAlterationsForGeneticProfile(GeneticProfile geneticProfile, Set<Integer> internalSampleIdsToRemove) throws DaoException {
+        int geneticProfileId = geneticProfile.getGeneticProfileId();
+        List<Integer> orderedSampleList = DaoGeneticProfileSamples.getOrderedSampleList(geneticProfileId);
+        Set<Integer> actualInternalSampleIdsToRemove = orderedSampleList.stream()
+                .filter(internalSampleIdsToRemove::contains).collect(Collectors.toUnmodifiableSet());
+        if (!actualInternalSampleIdsToRemove.isEmpty()) {
+            if (GeneticAlterationType.GENESET_SCORE.equals(geneticProfile.getGeneticAlterationType())) {
+                List<String> sampleStableIds = actualInternalSampleIdsToRemove.stream()
+                        .map(internalSampleID ->
+                                DaoSample.getSampleById(internalSampleID).getStableId())
+                        .toList();
+                throw new RuntimeException("Sample(s) with stable id "
+                        + String.join(", ", sampleStableIds)
+                        + " can't be removed as it contains GSVA data." +
+                        " Consider dropping and re-uploading the whole study.");
+            }
+            orderedSampleList.removeAll(actualInternalSampleIdsToRemove);
+            HashMap<Integer, HashMap<Integer, String>> geneticAlterationMapForEntityIds = DaoGeneticAlteration.getInstance().getGeneticAlterationMapForEntityIds(geneticProfileId, null);
+            DaoGeneticAlteration.getInstance().deleteAllRecordsInGeneticProfile(geneticProfileId);
+            if (!orderedSampleList.isEmpty()) {
+                for (Map.Entry<Integer, HashMap<Integer, String>> entry : geneticAlterationMapForEntityIds.entrySet()) {
+                    String[] values = orderedSampleList.stream().map(isid -> entry.getValue().get(isid)).toArray(String[]::new);
+                    DaoGeneticAlteration.getInstance().addGeneticAlterationsForGeneticEntity(geneticProfileId, entry.getKey(), values);
+                }
+            }
+            DaoGeneticProfileSamples.deleteAllSamplesInGeneticProfile(geneticProfileId);
+            if (!orderedSampleList.isEmpty()) {
+                DaoGeneticProfileSamples.addGeneticProfileSamples(geneticProfileId, orderedSampleList);
+            }
+        }
+        return actualInternalSampleIdsToRemove;
     }
 
     public static Set<Integer> findInternalSampleIdsInStudy(Integer internalStudyId, Set<String> sampleStableIds) {

diff --git a/src/main/java/org/mskcc/cbio/portal/scripts/RemovePatients.java b/src/main/java/org/mskcc/cbio/portal/scripts/RemovePatients.java
@@ -75,7 +75,7 @@ private void doRun() {
                 ProgressMonitor.setCurrentMessage("Removing patient with stable id(s) ("
                         + String.join(", ", patientIds)
                         + ") from study with stable id=" + cancerStudy.getCancerStudyStableId() + " ...");
-                DaoPatient.removePatientsEverywhereInStudy(cancerStudy.getInternalId(), patientIds);
+                DaoPatient.deletePatients(cancerStudy.getInternalId(), patientIds);
             }
         } catch (DaoException e) {
             throw new RuntimeException(e);

diff --git a/src/main/java/org/mskcc/cbio/portal/scripts/RemoveSamples.java b/src/main/java/org/mskcc/cbio/portal/scripts/RemoveSamples.java
@@ -24,6 +24,7 @@
 import org.jetbrains.annotations.NotNull;
 import org.mskcc.cbio.portal.dao.DaoCancerStudy;
 import org.mskcc.cbio.portal.dao.DaoException;
+import org.mskcc.cbio.portal.dao.DaoSample;
 import org.mskcc.cbio.portal.dao.JdbcUtil;
 import org.mskcc.cbio.portal.model.CancerStudy;
 import org.mskcc.cbio.portal.util.ProgressMonitor;
@@ -33,8 +34,6 @@
 import java.util.Set;
 import java.util.stream.Collectors;
 
-import static org.mskcc.cbio.portal.dao.DaoSample.removeSamplesEverywhereInStudy;
-
 /**
  * Command Line Tool to Remove Samples in Cancer Studies
  */
@@ -76,7 +75,7 @@ private void doRun() {
                 ProgressMonitor.setCurrentMessage("Removing sample with stable id(s) ("
                         + String.join(", ", sampleIds)
                         + ") from study with stable id=" + cancerStudy.getCancerStudyStableId() + " ...");
-                removeSamplesEverywhereInStudy(cancerStudy.getInternalId(), sampleIds);
+                DaoSample.deleteSamples(cancerStudy.getInternalId(), sampleIds);
             }
         } catch (DaoException e) {
             throw new RuntimeException(e);