Merge pull request #30 from waterflow80/sorted-name-length-pair

added the sorted-name-length-pair attribute to the model
EBIvariation · Aug 17, 2023 · 6af09f5 · 6af09f5
2 parents 2b5e1fb + 1438fb2
commit 6af09f5
Show file tree

Hide file tree

Showing 8 changed files with 142 additions and 10 deletions.
diff --git a/src/main/java/uk/ac/ebi/eva/evaseqcol/entities/SeqColExtendedDataEntity.java b/src/main/java/uk/ac/ebi/eva/evaseqcol/entities/SeqColExtendedDataEntity.java
@@ -6,6 +6,7 @@
 import org.hibernate.annotations.TypeDef;
 import org.hibernate.annotations.TypeDefs;
 
+import uk.ac.ebi.eva.evaseqcol.model.NameLengthPairEntity;
 import uk.ac.ebi.eva.evaseqcol.refget.SHA512ChecksumCalculator;
 import uk.ac.ebi.eva.evaseqcol.utils.JSONExtData;
 
@@ -18,6 +19,7 @@
 import javax.persistence.Transient;
 import java.util.ArrayList;
 import java.util.Arrays;
+import java.util.Comparator;
 import java.util.LinkedList;
 import java.util.List;
 
@@ -47,7 +49,7 @@ public class SeqColExtendedDataEntity {
     private SeqColEntity.NamingConvention namingConvention;
 
     public enum AttributeType {
-        names, sequences, md5DigestsOfSequences, lengths
+        names, sequences, md5DigestsOfSequences, lengths, sortedNameLengthPairs
     }
 
     public SeqColExtendedDataEntity setAttributeType(AttributeType attributeType) {
@@ -110,7 +112,8 @@ public static SeqColExtendedDataEntity constructSeqColLengthsObject(AssemblyEnti
 
     /**
      * Return the seqCol sequences array object*/
-    public static SeqColExtendedDataEntity constructSeqColSequencesObject(AssemblySequenceEntity assemblySequenceEntity) {
+    public static SeqColExtendedDataEntity constructSeqColSequencesObject(
+            AssemblySequenceEntity assemblySequenceEntity) {
         SeqColExtendedDataEntity seqColSequencesObject = new SeqColExtendedDataEntity().setAttributeType(
                 SeqColExtendedDataEntity.AttributeType.sequences);
         JSONExtData seqColSequencesArray = new JSONExtData();
@@ -128,7 +131,8 @@ public static SeqColExtendedDataEntity constructSeqColSequencesObject(AssemblySe
 
     /**
      * Return the seqCol sequences array object*/
-    public static SeqColExtendedDataEntity constructSeqColSequencesMd5Object(AssemblySequenceEntity assemblySequenceEntity) {
+    public static SeqColExtendedDataEntity constructSeqColSequencesMd5Object(
+            AssemblySequenceEntity assemblySequenceEntity) {
         SeqColExtendedDataEntity seqColSequencesObject = new SeqColExtendedDataEntity().setAttributeType(
                 AttributeType.md5DigestsOfSequences);
         JSONExtData seqColSequencesArray = new JSONExtData();
@@ -144,6 +148,63 @@ public static SeqColExtendedDataEntity constructSeqColSequencesMd5Object(Assembl
         return seqColSequencesObject;
     }
 
+    /**
+     * Return the seqCol sorted-name-length-pairs extended object*/
+    public static SeqColExtendedDataEntity constructSeqColSortedNameLengthPairs(
+            SeqColExtendedDataEntity extendedNames, SeqColExtendedDataEntity extendedLengths) {
+        if (extendedNames.getExtendedSeqColData().getObject().size() != extendedLengths.getExtendedSeqColData().getObject().size()) {
+            return null; // Names and Lengths entities are not compatible
+        }
+        SeqColExtendedDataEntity SeqColSortedNameLengthPairsObject = new SeqColExtendedDataEntity().setAttributeType(
+                AttributeType.sortedNameLengthPairs);
+        JSONExtData seqColSortedNameLengthPairsArray = new JSONExtData();
+
+        // Get the plain name-length pairs
+        List<NameLengthPairEntity> nameLengthPairList = constructNameLengthPairList(extendedNames, extendedLengths);
+        // Get the sorted list
+        List<String> sortedNameLengthPairsList = constructSortedNameLengthPairs(nameLengthPairList);
+
+        SHA512ChecksumCalculator sha512ChecksumCalculator = new SHA512ChecksumCalculator();
+        seqColSortedNameLengthPairsArray.setObject(sortedNameLengthPairsList);
+        SeqColSortedNameLengthPairsObject.setExtendedSeqColData(seqColSortedNameLengthPairsArray);
+        SeqColSortedNameLengthPairsObject.setDigest(sha512ChecksumCalculator.calculateChecksum(
+                seqColSortedNameLengthPairsArray.toString()));
+        return SeqColSortedNameLengthPairsObject;
+    }
+
+    /**
+     * Retrieve and construct the list of name-length pairs*/
+    private static List<NameLengthPairEntity> constructNameLengthPairList(
+            SeqColExtendedDataEntity extendedNames, SeqColExtendedDataEntity extendedLengths) {
+        List<NameLengthPairEntity> nameLengthPairList = new ArrayList<>();
+        for (int i=0; i<extendedNames.getExtendedSeqColData().getObject().size(); i++) {
+            String name = extendedNames.getExtendedSeqColData().getObject().get(i);
+            String length = extendedLengths.getExtendedSeqColData().getObject().get(i);
+            nameLengthPairList.add(new NameLengthPairEntity(name, length));
+        }
+        return nameLengthPairList;
+    }
+
+    /**
+     * Return the sorted-name-length-pair list for the given list of nameLengthPairEntity*/
+    public static List<String> constructSortedNameLengthPairs(List<NameLengthPairEntity> nameLengthPairList) {
+        SHA512ChecksumCalculator sha512ChecksumCalculator = new SHA512ChecksumCalculator();
+        List<String> sortedNameLengthPairs = new ArrayList<>();
+        for (NameLengthPairEntity entity: nameLengthPairList) {
+            String nameLengthHash = sha512ChecksumCalculator.calculateChecksum(entity.toString());
+            sortedNameLengthPairs.add(nameLengthHash);
+        }
+        // Sorting the name-length-pair list according to the elements' natural order (alphanumerically)
+        Comparator<String> nameLengthComparator = new Comparator<String>() {
+            @Override
+            public int compare(String o1, String o2) {
+                return o1.toLowerCase().compareTo(o2.toLowerCase());
+            }
+        };
+        sortedNameLengthPairs.sort(nameLengthComparator);
+        return sortedNameLengthPairs;
+    }
+
     /**
      * Return the list of extended data entities that are the same across multiple seqCol objects under
      * the same assembly accession. These entities are "sequences", "md5Sequences" and "lengths". */
@@ -159,7 +220,8 @@ public static List<SeqColExtendedDataEntity> constructSameValueExtendedSeqColDat
     /**
      * Return a list of seqCol sequences' names with all possible naming convention that can be extracted
      * from the given assemblyEntity*/
-    public static List<SeqColExtendedDataEntity> constructAllPossibleExtendedNamesSeqColData(AssemblyEntity assemblyEntity) {
+    public static List<SeqColExtendedDataEntity> constructAllPossibleExtendedNamesSeqColData(
+            AssemblyEntity assemblyEntity) {
         List<SeqColEntity.NamingConvention> existingNamingConventions = new ArrayList<>();
         if (assemblyEntity.getChromosomes().get(0).getEnaSequenceName() != null) {
             existingNamingConventions.add(SeqColEntity.NamingConvention.ENA);

diff --git a/src/main/java/uk/ac/ebi/eva/evaseqcol/entities/SeqColLevelTwoEntity.java b/src/main/java/uk/ac/ebi/eva/evaseqcol/entities/SeqColLevelTwoEntity.java
@@ -17,6 +17,8 @@ public class SeqColLevelTwoEntity extends SeqColEntity{
     private List<String> lengths;
     @JsonProperty("md5-sequences")
     private List<String> md5DigestsOfSequences;
+    @JsonProperty("sorted-name-length-pairs")
+    private List<String> sortedNameLengthPairs;
 
     public SeqColLevelTwoEntity setDigest(String digest) {
         this.digest = digest;

diff --git a/src/main/java/uk/ac/ebi/eva/evaseqcol/model/NameLengthPairEntity.java b/src/main/java/uk/ac/ebi/eva/evaseqcol/model/NameLengthPairEntity.java
@@ -0,0 +1,25 @@
+package uk.ac.ebi.eva.evaseqcol.model;
+
+import lombok.AllArgsConstructor;
+import lombok.Data;
+import lombok.NoArgsConstructor;
+
+@Data
+@NoArgsConstructor
+public class NameLengthPairEntity {
+    private String name;
+    private String length;
+
+    public NameLengthPairEntity(String name, String length) {
+        this.name = name;
+        this.length = length;
+    }
+
+    @Override
+    public String toString() {
+        return "{" +
+                "    \"name\":\""+ name +"\"," +
+                "    \"length\":\""+ length +"\"," +
+                "}";
+    }
+}
diff --git a/src/main/java/uk/ac/ebi/eva/evaseqcol/service/SeqColExtendedDataService.java b/src/main/java/uk/ac/ebi/eva/evaseqcol/service/SeqColExtendedDataService.java
@@ -92,14 +92,22 @@ public Optional<SeqColExtendedDataEntity> getExtendedAttributeByDigest(String di
     }
 
     /**
-     * Return the 3 extended data objects (names, lengths, sequences and sequencesMD5) of the given naming convention*/
-    public List<SeqColExtendedDataEntity> constructExtendedSeqColDataList(AssemblyEntity assemblyEntity, AssemblySequenceEntity assemblySequenceEntity,
-                                                            SeqColEntity.NamingConvention convention) throws IOException {
+     * Return the 5 seqCol extended data objects (names, lengths, sequences, sequencesMD5 and sorted-name-length-pair)
+     * of the given assembly and naming convention*/
+    public List<SeqColExtendedDataEntity> constructExtendedSeqColDataList(
+            AssemblyEntity assemblyEntity, AssemblySequenceEntity assemblySequenceEntity,
+            SeqColEntity.NamingConvention convention) {
+        SeqColExtendedDataEntity extendedLengthsEntity = SeqColExtendedDataEntity
+                .constructSeqColLengthsObject(assemblyEntity);
+        SeqColExtendedDataEntity extendedNamesEntity = SeqColExtendedDataEntity
+                .constructSeqColNamesObjectByNamingConvention(assemblyEntity, convention);
+
         return Arrays.asList(
                 SeqColExtendedDataEntity.constructSeqColSequencesObject(assemblySequenceEntity),
                 SeqColExtendedDataEntity.constructSeqColSequencesMd5Object(assemblySequenceEntity),
-                SeqColExtendedDataEntity.constructSeqColNamesObjectByNamingConvention(assemblyEntity, convention),
-                SeqColExtendedDataEntity.constructSeqColLengthsObject(assemblyEntity)
+                extendedNamesEntity,
+                extendedLengthsEntity,
+                SeqColExtendedDataEntity.constructSeqColSortedNameLengthPairs(extendedNamesEntity, extendedLengthsEntity)
         );
     }
 

diff --git a/src/main/java/uk/ac/ebi/eva/evaseqcol/service/SeqColLevelOneService.java b/src/main/java/uk/ac/ebi/eva/evaseqcol/service/SeqColLevelOneService.java
@@ -79,6 +79,9 @@ public SeqColLevelOneEntity constructSeqColLevelOne(List<SeqColExtendedDataEntit
                 case md5DigestsOfSequences:
                     jsonLevelOne.setMd5DigestsOfSequences(dataEntity.getDigest());
                     break;
+                case sortedNameLengthPairs:
+                    jsonLevelOne.setSortedNameLengthPairs(dataEntity.getDigest());
+                    break;
             }
         }
         levelOneEntity.setSeqColLevel1Object(jsonLevelOne);
@@ -97,6 +100,7 @@ public SeqColLevelOneEntity constructSeqColLevelOne(
         JSONExtData lengthsExtData = new JSONExtData(levelTwoEntity.getLengths());
         JSONExtData namesExtData = new JSONExtData(levelTwoEntity.getNames());
         JSONExtData md5SequencesExtData = new JSONExtData(levelTwoEntity.getMd5DigestsOfSequences());
+        JSONExtData sortedNameLengthPairsData = new JSONExtData(levelTwoEntity.getSortedNameLengthPairs());
 
         // Sequences
         SeqColExtendedDataEntity sequencesExtEntity = new SeqColExtendedDataEntity();
@@ -118,12 +122,18 @@ public SeqColLevelOneEntity constructSeqColLevelOne(
         namesExtEntity.setAttributeType(SeqColExtendedDataEntity.AttributeType.names);
         namesExtEntity.setExtendedSeqColData(namesExtData);
         namesExtEntity.setDigest(sha512Calculator.calculateChecksum(namesExtData.toString()));
+        //sorted-name-length-pairs
+        SeqColExtendedDataEntity sortedNameLengthPairsExtEntity = new SeqColExtendedDataEntity();
+        sortedNameLengthPairsExtEntity.setAttributeType(SeqColExtendedDataEntity.AttributeType.sortedNameLengthPairs);
+        sortedNameLengthPairsExtEntity.setExtendedSeqColData(sortedNameLengthPairsData);
+        sortedNameLengthPairsExtEntity.setDigest(sha512Calculator.calculateChecksum(sortedNameLengthPairsData.toString()));
 
         List<SeqColExtendedDataEntity> extendedDataEntities = Arrays.asList(
                 sequencesExtEntity,
                 md5SequencesExtEntity,
                 lengthsExtEntity,
-                namesExtEntity
+                namesExtEntity,
+                sortedNameLengthPairsExtEntity
         );
         return constructSeqColLevelOne(extendedDataEntities, convention);
     }

diff --git a/src/main/java/uk/ac/ebi/eva/evaseqcol/service/SeqColLevelTwoService.java b/src/main/java/uk/ac/ebi/eva/evaseqcol/service/SeqColLevelTwoService.java
@@ -106,6 +106,9 @@ public SeqColLevelTwoEntity constructSeqColL2(String level0Digest, List<SeqColEx
                 case md5DigestsOfSequences:
                     levelTwoEntity.setMd5DigestsOfSequences(extendedData.getExtendedSeqColData().getObject());
                     break;
+                case sortedNameLengthPairs:
+                    levelTwoEntity.setSortedNameLengthPairs(extendedData.getExtendedSeqColData().getObject());
+                    break;
             }
         }
         return levelTwoEntity;

diff --git a/src/main/java/uk/ac/ebi/eva/evaseqcol/service/SeqColService.java b/src/main/java/uk/ac/ebi/eva/evaseqcol/service/SeqColService.java
@@ -82,10 +82,16 @@ public Optional<? extends SeqColEntity> getSeqColByDigestAndLevel(String digest,
            String namesDigest = seqColLevelOne.get().getSeqColLevel1Object().getNames();
            JSONExtData extendedNames = extendedDataService.getSeqColExtendedDataEntityByDigest(namesDigest).get().getExtendedSeqColData();
 
+           // Retrieving sortedNameLengthPairs
+           String sortedNameLengthPairsDigest = seqColLevelOne.get().getSeqColLevel1Object().getSortedNameLengthPairs();
+           JSONExtData extendedSortedNameLengthPairs = extendedDataService.
+                   getSeqColExtendedDataEntityByDigest(sortedNameLengthPairsDigest).get().getExtendedSeqColData();
+
            levelTwoEntity.setSequences(extendedSequences.getObject());
            levelTwoEntity.setMd5DigestsOfSequences(extendedMd5Sequnces.getObject());
            levelTwoEntity.setLengths(extendedLengths.getObject());
            levelTwoEntity.setNames(extendedNames.getObject());
+           levelTwoEntity.setSortedNameLengthPairs(extendedSortedNameLengthPairs.getObject());
 
            return Optional.of(levelTwoEntity);
        } else {
@@ -128,7 +134,11 @@ public List<String> fetchAndInsertAllSeqColByAssemblyAccession(
         List<SeqColExtendedDataEntity> sameValueAttributeList = seqColDataMap.get().get("sameValueAttributes");
         for (SeqColExtendedDataEntity extendedNamesEntity: possibleSequencesNamesList) {
             List<SeqColExtendedDataEntity> seqColExtendedDataEntities = new ArrayList<>(sameValueAttributeList);
+            SeqColExtendedDataEntity extendedLengthsEntity = retrieveExtendedLengthEntity(seqColExtendedDataEntities);
+            SeqColExtendedDataEntity seqColSortedNameLengthPairEntity = SeqColExtendedDataEntity.
+                    constructSeqColSortedNameLengthPairs(extendedNamesEntity, extendedLengthsEntity);
             seqColExtendedDataEntities.add(extendedNamesEntity);
+            seqColExtendedDataEntities.add(seqColSortedNameLengthPairEntity);
             SeqColLevelOneEntity levelOneEntity = levelOneService.constructSeqColLevelOne(seqColExtendedDataEntities, extendedNamesEntity.getNamingConvention());
             Optional<String> seqColDigest = insertSeqColL1AndL2(levelOneEntity, seqColExtendedDataEntities);
             if (seqColDigest.isPresent()) {
@@ -142,6 +152,16 @@ public List<String> fetchAndInsertAllSeqColByAssemblyAccession(
         return insertedSeqColDigests;
     }
 
+    /**
+     * Return the extended data entity that corresponds to the seqCol lengths attribute*/
+    public SeqColExtendedDataEntity retrieveExtendedLengthEntity(List<SeqColExtendedDataEntity> extendedDataEntities) {
+        for (SeqColExtendedDataEntity entity: extendedDataEntities) {
+            if (entity.getAttributeType() == SeqColExtendedDataEntity.AttributeType.lengths) {
+                return entity;
+            }
+        }
+        return null;
+    }
     @Transactional
     /**
      * Insert the given Level 1 seqCol entity and its corresponding extended level 2 data (names, lengths, sequences, ...)

diff --git a/src/main/java/uk/ac/ebi/eva/evaseqcol/utils/JSONLevelOne.java b/src/main/java/uk/ac/ebi/eva/evaseqcol/utils/JSONLevelOne.java
@@ -16,6 +16,8 @@ public class JSONLevelOne implements Serializable {
     private String md5DigestsOfSequences;
     private String names;
     private String lengths;
+    @JsonProperty("sorted-name-length-pairs")
+    private String sortedNameLengthPairs;
 
     public JSONLevelOne setSequences(String sequences) {
         this.sequences = sequences;