Merge pull request #47 from EBI-IntAct/export-datasets-on-mitab-format

Export datasets on mitab format
EBI-IntAct · Sep 27, 2024 · 5d7fb74 · 5d7fb74
2 parents cd95a06 + 0da9d8a
commit 5d7fb74
Show file tree

Hide file tree

Showing 33 changed files with 61 additions and 41 deletions.
diff --git a/complex-go-export/pom.xml b/complex-go-export/pom.xml
@@ -5,7 +5,7 @@
     <parent>
         <groupId>uk.ac.ebi.intact.dataexchange</groupId>
         <artifactId>intact-dataexchange-master</artifactId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <artifactId>complex-go-export</artifactId>

diff --git a/complex-pdb-export/pom.xml b/complex-pdb-export/pom.xml
@@ -5,7 +5,7 @@
     <parent>
         <groupId>uk.ac.ebi.intact.dataexchange</groupId>
         <artifactId>intact-dataexchange-master</artifactId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <artifactId>complex-pdb-export</artifactId>

diff --git a/complex-tab-export/pom.xml b/complex-tab-export/pom.xml
@@ -5,7 +5,7 @@
     <parent>
         <groupId>uk.ac.ebi.intact.dataexchange</groupId>
         <artifactId>intact-dataexchange-master</artifactId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
 

diff --git a/complex-uniprot-dr-export/pom.xml b/complex-uniprot-dr-export/pom.xml
@@ -5,7 +5,7 @@
     <parent>
         <groupId>uk.ac.ebi.intact.dataexchange</groupId>
         <artifactId>intact-dataexchange-master</artifactId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <artifactId>complex-uniprot-dr-export</artifactId>

diff --git a/cttv-exporter/pom.xml b/cttv-exporter/pom.xml
@@ -3,7 +3,7 @@
     <parent>
         <groupId>uk.ac.ebi.intact.dataexchange</groupId>
         <artifactId>intact-dataexchange-master</artifactId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <modelVersion>4.0.0</modelVersion>

diff --git a/imex-id-update/pom.xml b/imex-id-update/pom.xml
@@ -4,7 +4,7 @@
     <parent>
         <groupId>uk.ac.ebi.intact.dataexchange</groupId>
         <artifactId>intact-dataexchange-master</artifactId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <groupId>uk.ac.ebi.intact.dataexchange.imex</groupId>

diff --git a/intact-cvutils/pom.xml b/intact-cvutils/pom.xml
@@ -3,7 +3,7 @@
     <parent>
         <groupId>uk.ac.ebi.intact.dataexchange</groupId>
         <artifactId>intact-dataexchange-master</artifactId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <modelVersion>4.0.0</modelVersion>

diff --git a/intact-db-importer/pom.xml b/intact-db-importer/pom.xml
@@ -4,7 +4,7 @@
     <parent>
         <groupId>uk.ac.ebi.intact.dataexchange</groupId>
         <artifactId>intact-dataexchange-master</artifactId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <artifactId>intact-db-importer</artifactId>

diff --git a/intact-enricher/pom.xml b/intact-enricher/pom.xml
@@ -4,7 +4,7 @@
     <parent>
         <artifactId>intact-dataexchange-master</artifactId>
         <groupId>uk.ac.ebi.intact.dataexchange</groupId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <modelVersion>4.0.0</modelVersion>

diff --git a/intact-mi-cluster-score/pom.xml b/intact-mi-cluster-score/pom.xml
@@ -4,7 +4,7 @@
     <parent>
         <artifactId>intact-dataexchange-master</artifactId>
         <groupId>uk.ac.ebi.intact.dataexchange</groupId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <artifactId>intact-mi-cluster-score</artifactId>

diff --git a/intact-mutation-export/pom.xml b/intact-mutation-export/pom.xml
@@ -3,7 +3,7 @@
     <parent>
         <groupId>uk.ac.ebi.intact.dataexchange</groupId>
         <artifactId>intact-dataexchange-master</artifactId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <modelVersion>4.0.0</modelVersion>

diff --git a/intact-orthology-import/pom.xml b/intact-orthology-import/pom.xml
@@ -8,7 +8,7 @@
     <parent>
         <groupId>uk.ac.ebi.intact.dataexchange</groupId>
         <artifactId>intact-dataexchange-master</artifactId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <artifactId>intact-orthology-import</artifactId>

diff --git a/intact-pdbe-import/pom.xml b/intact-pdbe-import/pom.xml
@@ -4,7 +4,7 @@
     <parent>
         <groupId>uk.ac.ebi.intact.dataexchange</groupId>
         <artifactId>intact-dataexchange-master</artifactId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <modelVersion>4.0.0</modelVersion>

diff --git a/intact-tasks/pom.xml b/intact-tasks/pom.xml
@@ -4,7 +4,7 @@
     <parent>
         <groupId>uk.ac.ebi.intact.dataexchange</groupId>
         <artifactId>intact-dataexchange-master</artifactId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <modelVersion>4.0.0</modelVersion>

diff --git a/intact-tasks/src/main/java/uk/ac/ebi/intact/task/mitab/pmid/PublicationFileEntry.java b/intact-tasks/src/main/java/uk/ac/ebi/intact/task/mitab/pmid/PublicationFileEntry.java
@@ -1,6 +1,7 @@
 package uk.ac.ebi.intact.task.mitab.pmid;
 
 import java.util.Date;
+import java.util.List;
 
 /**
  * The publication entry is an mitab inputstream with a few information about the publication. A publicationEntry only contains interactions of a same publication.
@@ -29,8 +30,14 @@ public class PublicationFileEntry implements Comparable<PublicationFileEntry>{
     private boolean isNegative;
 
     private String publicationDate;
-
-    public PublicationFileEntry(Date createdDate, String pubId, StringBuffer mitab, boolean isNegative, String publicationDate){
+    private List<String> datasets;
+
+    public PublicationFileEntry(Date createdDate,
+                                String pubId,
+                                StringBuffer mitab,
+                                boolean isNegative,
+                                String publicationDate,
+                                List<String> datasets) {
         super();
         this.createdDate = createdDate;
         this.entryName = pubId;
@@ -42,6 +49,7 @@ public PublicationFileEntry(Date createdDate, String pubId, StringBuffer mitab,
         }
         this.isNegative = isNegative;
         this.publicationDate = publicationDate;
+        this.datasets = datasets;
     }
 
     public Date getCreatedDate() {
@@ -72,6 +80,14 @@ public void setPublicationDate(String publicationDate) {
         this.publicationDate = publicationDate;
     }
 
+    public List<String> getDatasets() {
+        return datasets;
+    }
+
+    public void setDatasets(List<String> datasets) {
+        this.datasets = datasets;
+    }
+
     @Override
     public int compareTo(PublicationFileEntry o) {
         final int BEFORE = -1;

diff --git a/...t-tasks/src/main/java/uk/ac/ebi/intact/task/mitab/pmid/PublicationMitabItemProcessor.java b/...t-tasks/src/main/java/uk/ac/ebi/intact/task/mitab/pmid/PublicationMitabItemProcessor.java
@@ -29,8 +29,6 @@
 import uk.ac.ebi.intact.task.mitab.InteractionExpansionCompositeProcessor;
 import uk.ac.ebi.intact.task.util.FileNameGenerator;
 
-import java.text.DateFormat;
-import java.text.SimpleDateFormat;
 import java.util.*;
 import java.util.regex.Pattern;
 
@@ -134,25 +132,30 @@ public SortedSet<PublicationFileEntry> process(Publication item) throws Exceptio
         }
 
         String publicationDate = null;
+        List<String> datasets = new ArrayList<>();
         Collection<Annotation> pubAnnotations = AnnotatedObjectUtils.getPublicAnnotations(publication);
-        for (Annotation annot : pubAnnotations){
+        for (Annotation annot : pubAnnotations) {
             if (annot.getCvTopic() != null){
                 // date
                 if (CvTopic.PUBLICATION_YEAR_MI_REF.equals(annot.getCvTopic().getIdentifier())) {
                     if (PUBLICATION_DATE_REGEX.matcher(annot.getAnnotationText()).matches()) {
                         publicationDate = annot.getAnnotationText();
                     }
                 }
+                // dataset
+                if (CvTopic.DATASET_MI_REF.equals(annot.getCvTopic().getIdentifier())) {
+                    datasets.add(annot.getAnnotationText());
+                }
             }
         }
 
         if (this.currentNegativeStringBuilder.length() > 0){
             createPublicationEntry(
-                    publicationEntries, publication.getCreated(), publication.getShortLabel(), this.currentNegativeStringBuilder, true, publicationDate);
+                    publicationEntries, publication.getCreated(), publication.getShortLabel(), this.currentNegativeStringBuilder, true, publicationDate, datasets);
         }
         if (this.currentStringBuilder.length() > 0){
             createPublicationEntry(
-                    publicationEntries, publication.getCreated(), publication.getShortLabel(), this.currentStringBuilder, false, publicationDate);
+                    publicationEntries, publication.getCreated(), publication.getShortLabel(), this.currentStringBuilder, false, publicationDate, datasets);
         }
 
         IntactContext.getCurrentInstance().getDaoFactory().getEntityManager().clear();
@@ -228,15 +231,16 @@ private void createPublicationEntry(Set<PublicationFileEntry> publicationEntries
                                         String publicationName,
                                         StringBuffer mitab,
                                         boolean isNegative,
-                                        String publicationDate) {
+                                        String publicationDate,
+                                        List<String> datasets) {
         log.info("create publication entry : " + publicationName);
 
         // create a publication name
         // name of the entry = publicationId_experimentLabel_chunkNumber
         String entryName = publicationNameGenerator.createPublicationName(publicationName, null, isNegative);
 
         // create a publication entry
-        PublicationFileEntry publicationEntry = new PublicationFileEntry(date, entryName, mitab, isNegative, publicationDate);
+        PublicationFileEntry publicationEntry = new PublicationFileEntry(date, entryName, mitab, isNegative, publicationDate, datasets);
         // add the publication entry to the list of publication entries
         publicationEntries.add(publicationEntry);
 

diff --git a/intact-tasks/src/test/java/uk/ac/ebi/intact/task/pmid/PublicationMitabItemWriterTest.java b/intact-tasks/src/test/java/uk/ac/ebi/intact/task/pmid/PublicationMitabItemWriterTest.java
@@ -137,11 +137,11 @@ public void test_write_publications() throws Exception {
 
         PublicationFileEntry oldEntryWithNoPublicationDate = pubWithNoPublicationDateEntries.iterator().next();
         PublicationFileEntry newEntryWithNoPublicationDate = new PublicationFileEntry(
-                dateFormatForEntry.parse(secondCreatedDate), "12345_10", oldEntryWithNoPublicationDate.getBinaryInteractions(), false, null);
+                dateFormatForEntry.parse(secondCreatedDate), "12345_10", oldEntryWithNoPublicationDate.getBinaryInteractions(), false, null, new ArrayList<>());
         pubWithNoPublicationDateEntries2.add(newEntryWithNoPublicationDate);
         PublicationFileEntry oldEntryWithPublicationDate = pubWithPublicationDateEntries.iterator().next();
         PublicationFileEntry newEntryWithPublicationDate = new PublicationFileEntry(
-                dateFormatForEntry.parse(secondCreatedDate), "98765_10", oldEntryWithPublicationDate.getBinaryInteractions(), false, secondPublicationDate);
+                dateFormatForEntry.parse(secondCreatedDate), "98765_10", oldEntryWithPublicationDate.getBinaryInteractions(), false, secondPublicationDate, new ArrayList<>());
         pubWithPublicationDateEntries2.add(newEntryWithPublicationDate);
 
         DateFormat format = new SimpleDateFormat("yyyy");

diff --git a/intact-uniprot-export/pom.xml b/intact-uniprot-export/pom.xml
@@ -3,7 +3,7 @@
     <parent>
         <groupId>uk.ac.ebi.intact.dataexchange</groupId>
         <artifactId>intact-dataexchange-master</artifactId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <modelVersion>4.0.0</modelVersion>

diff --git a/pom.xml b/pom.xml
@@ -12,7 +12,7 @@
     <groupId>uk.ac.ebi.intact.dataexchange</groupId>
     <artifactId>intact-dataexchange-master</artifactId>
     <packaging>pom</packaging>
-    <version>4.2.0-SNAPSHOT</version>
+    <version>4.3.0-SNAPSHOT</version>
 
     <name>IntAct Data Exchange</name>
     <description>Data Exchange Master POM</description>

diff --git a/psimi/intact-psimi-exporter/pom.xml b/psimi/intact-psimi-exporter/pom.xml
@@ -4,7 +4,7 @@
     <parent>
         <groupId>uk.ac.ebi.intact.dataexchange.psimi</groupId>
         <artifactId>intact-psimi-master</artifactId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <artifactId>intact-psimi-exporter</artifactId>

diff --git a/psimi/pom.xml b/psimi/pom.xml
@@ -4,7 +4,7 @@
     <parent>
         <groupId>uk.ac.ebi.intact.dataexchange</groupId>
         <artifactId>intact-dataexchange-master</artifactId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <modelVersion>4.0.0</modelVersion>

diff --git a/psimi/psimitab/intact-calimocho-converters/pom.xml b/psimi/psimitab/intact-calimocho-converters/pom.xml
@@ -4,7 +4,7 @@
     <parent>
         <artifactId>psimitab-master</artifactId>
         <groupId>uk.ac.ebi.intact.dataexchange.psimi</groupId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <artifactId>intact-calimocho-converters</artifactId>

diff --git a/psimi/psimitab/intact-jami-mitab/pom.xml b/psimi/psimitab/intact-jami-mitab/pom.xml
@@ -4,7 +4,7 @@
     <parent>
         <groupId>uk.ac.ebi.intact.dataexchange.psimi</groupId>
         <artifactId>psimitab-master</artifactId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <artifactId>intact-jami-mitab</artifactId>

diff --git a/psimi/psimitab/intact-psimitab-converters/pom.xml b/psimi/psimitab/intact-psimitab-converters/pom.xml
@@ -3,7 +3,7 @@
     <parent>
         <artifactId>psimitab-master</artifactId>
         <groupId>uk.ac.ebi.intact.dataexchange.psimi</groupId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <modelVersion>4.0.0</modelVersion>

diff --git a/psimi/psimitab/intact-solr-home/pom.xml b/psimi/psimitab/intact-solr-home/pom.xml
@@ -3,7 +3,7 @@
     <parent>
         <artifactId>psimitab-master</artifactId>
         <groupId>uk.ac.ebi.intact.dataexchange.psimi</groupId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
     <modelVersion>4.0.0</modelVersion>
 

diff --git a/psimi/psimitab/intact-solr/pom.xml b/psimi/psimitab/intact-solr/pom.xml
@@ -6,7 +6,7 @@
     <parent>
         <artifactId>psimitab-master</artifactId>
         <groupId>uk.ac.ebi.intact.dataexchange.psimi</groupId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <modelVersion>4.0.0</modelVersion>

diff --git a/psimi/psimitab/pom.xml b/psimi/psimitab/pom.xml
@@ -4,7 +4,7 @@
     <parent>
         <groupId>uk.ac.ebi.intact.dataexchange.psimi</groupId>
         <artifactId>intact-psimi-master</artifactId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <modelVersion>4.0.0</modelVersion>

diff --git a/psimi/psixml/intact-jami-xml/pom.xml b/psimi/psixml/intact-jami-xml/pom.xml
@@ -4,7 +4,7 @@
     <parent>
         <groupId>uk.ac.ebi.intact.dataexchange.psimi</groupId>
         <artifactId>intact-psixml</artifactId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <artifactId>intact-jami-xml</artifactId>

diff --git a/psimi/psixml/intact-psixml-converters/pom.xml b/psimi/psixml/intact-psixml-converters/pom.xml
@@ -4,7 +4,7 @@
     <parent>
         <artifactId>intact-psixml</artifactId>
         <groupId>uk.ac.ebi.intact.dataexchange.psimi</groupId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <modelVersion>4.0.0</modelVersion>

diff --git a/psimi/psixml/intact-psixml-dbimporter/pom.xml b/psimi/psixml/intact-psixml-dbimporter/pom.xml
@@ -6,7 +6,7 @@
     <parent>
         <artifactId>intact-psixml</artifactId>
         <groupId>uk.ac.ebi.intact.dataexchange.psimi</groupId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <artifactId>intact-psixml-dbimporter</artifactId>

diff --git a/psimi/psixml/intact-psixml-exchange/pom.xml b/psimi/psixml/intact-psixml-exchange/pom.xml
@@ -3,7 +3,7 @@
     <parent>
         <groupId>uk.ac.ebi.intact.dataexchange.psimi</groupId>
         <artifactId>intact-psixml</artifactId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <modelVersion>4.0.0</modelVersion>

diff --git a/psimi/psixml/pom.xml b/psimi/psixml/pom.xml
@@ -4,7 +4,7 @@
     <parent>
         <groupId>uk.ac.ebi.intact.dataexchange.psimi</groupId>
         <artifactId>intact-psimi-master</artifactId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <modelVersion>4.0.0</modelVersion>

diff --git a/structured-abstract/pom.xml b/structured-abstract/pom.xml
@@ -4,7 +4,7 @@
     <parent>
         <groupId>uk.ac.ebi.intact.dataexchange</groupId>
         <artifactId>intact-dataexchange-master</artifactId>
-        <version>4.2.0-SNAPSHOT</version>
+        <version>4.3.0-SNAPSHOT</version>
     </parent>
 
     <artifactId>structured-abstract</artifactId>