Merge branch 'tair-fix' into 'devel_13.1.x'

Tair fix

See merge request minerva/core!774

CHANGELOG

+minerva (12.3.1~beta.1) unstable; urgency=low
+  * Bug fix: tair locus identifiers were used improperly - instead of id the
+    name was used
+
 minerva (13.1.0~beta.0) unstable; urgency=low
   * Feature: annotators are more flexible - you can define set of input and
     outputs used by annotator (#617)
   * Feature: changes in admin panel doesn't require saving - they are saved
     automatically (#676)
   * Feature: elements can have custom glyphs used for visualization (#487)
-  * Small improvement: all bio entities have z-index associated with them 
+  * Small improvement: all bio entities have z-index associated with them
     (#487)
   * Small improvement: validation of the organism and disease id on map upload
     added (#618)
@@ -51,7 +55,7 @@ minerva (13.1.0~beta.0) unstable; urgency=low
   * Small improvement: allow admin to disable CORS check (#802)
   * Small improvement: TransparencyZoomLevelVisibility parameter renamed to
     SemanticZoomLevelTransparency (#801)
-  * Small improvement: export/import from SBML support z-index in LAYOUT 
+  * Small improvement: export/import from SBML support z-index in LAYOUT
     extension
   * Bug fix: progress bar of gene genome mapping upload is refreshing properly
     (#728)
@@ -68,7 +72,7 @@ minerva (13.1.0~beta.0) unstable; urgency=low
   * Bug fix: plugin contect element width is adjusted when link to tabs are
     wrapped in more than one line (#758)
   * Bug fix: export to CellDesigner preserve font size (#803)
-  * Bug fix: layout data was ignored for some reactions when importing from 
+  * Bug fix: layout data was ignored for some reactions when importing from
     SBML (#812)
 
  -- Piotr Gawron <piotr.gawron@uni.lu>  Mon, 13 May 2019 19:00:00 +0200

annotation/src/main/java/lcsb/mapviewer/annotation/services/annotators/KeggAnnotator.java

@@ -152,6 +152,9 @@ public class KeggAnnotator extends ElementAnnotator implements IExternalService
         logger.warn("Cannot find kegg data for id: " + identifier);
       } catch (IOException exception) {
         throw new AnnotatorException(exception);
+      } catch (UniprotSearchException e) {
+    	  logger.warn(e, e);
+    	  return false;
       }
     }
     object.addMiriamData(annotations);
@@ -181,8 +184,9 @@ public class KeggAnnotator extends ElementAnnotator implements IExternalService
    *          organisms names. If the value has not been set by the user, null
    *          will be passed.
    * @return {@link MiriamType#PUBMED}s found on the page
+   * @throws UniprotSearchException 
    */
-  private Collection<MiriamData> parseKegg(String pageContent, AnnotatorData params) {
+  private Collection<MiriamData> parseKegg(String pageContent, AnnotatorData params) throws UniprotSearchException {
 
     // Retrieve Pubmeds
     Collection<MiriamData> result = new HashSet<>();
@@ -201,10 +205,13 @@ public class KeggAnnotator extends ElementAnnotator implements IExternalService
         } else {
           m = athOrthologMatcher.matcher(pageContent);
           if (m.find()) {
-            String[] tairCodes = m.group(1).trim().split(" ");
-            for (String tairCode : tairCodes) {
-              tairCode = tairCode.split("\\(")[0]; // some codes are in the form AT1G08510(FATB)
-              result.add(new MiriamData(MiriamType.TAIR_LOCUS, tairCode));
+            String[] tairLocusNames = m.group(1).trim().split(" ");
+            for (String tairLocusName : tairLocusNames) {
+            	tairLocusName = tairLocusName.split("\\(")[0]; // some codes are in the form AT1G08510(FATB)
+            	MiriamData md = uniprotAnnotator.uniprotTairLocusNameToId(tairLocusName);
+            	if (!md.equals(new MiriamData())){
+            		result.add(md);            		
+            	}
             }
           }
         }

annotation/src/main/java/lcsb/mapviewer/annotation/services/annotators/StringAnnotator.java

@@ -32,6 +32,12 @@ public class StringAnnotator extends ElementAnnotator implements IExternalServic
   /**
    * Service used for annotation of entities using {@link MiriamType#TAIR_LOCUS
    * TAIR}.
+   * Note that STRING annotation process will annotate only
+ * records which have a TAIR ID assigned by a human annotator.
+ * Otherwise, it would generate UniProt miriam records also for
+ * TAIR IDs generated from, e.g., KEGG annotator, i.e. for homologues
+ * and these UniProt IDs would be indistinguishable from the
+ * UniProt IDs describing the molecule.
    */
   private TairAnnotator tairAnnotator;
 
@@ -54,7 +60,9 @@ public class StringAnnotator extends ElementAnnotator implements IExternalServic
       throws AnnotatorException {
     List<MiriamData> mdUniprots = new ArrayList<>();
     if (identifier.getDataType().equals(MiriamType.TAIR_LOCUS)) {
-      mdUniprots.addAll(tairAnnotator.tairToUniprot(identifier));
+    	if (identifier.getAnnotator() == null) {
+    		mdUniprots.addAll(tairAnnotator.tairToUniprot(identifier));
+    	}      
     } else if (identifier.getDataType().equals(MiriamType.UNIPROT)) {
       mdUniprots.add(identifier);
     } else {

annotation/src/main/java/lcsb/mapviewer/annotation/services/annotators/TairAnnotator.java

@@ -5,6 +5,8 @@ import java.util.Arrays;
 import java.util.Collection;
 import java.util.HashSet;
 import java.util.List;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
 
 import org.apache.log4j.Logger;
 import org.springframework.stereotype.Service;
@@ -28,6 +30,13 @@ import lcsb.mapviewer.model.user.annotator.AnnotatorOutputParameter;
 
 /**
  * This is a class that implements a backend to TAIR.
+ * Note that TAIR annotation process will annotate only
+ * records which have a TAIR ID assigned by a human annotator.
+ * Otherwise, it would generate UniProt miriam records also for
+ * TAIR IDs generated from, e.g., KEGG annotator, i.e. for homologues
+ * and these UniProt IDs would be indistinguishable from the
+ * UniProt IDs describing the molecule.
+ *    
  * 
  * @author David Hoksza
  * 
@@ -39,6 +48,14 @@ public class TairAnnotator extends ElementAnnotator implements IExternalService
    * Default class logger.
    */
   private static Logger logger = Logger.getLogger(TairAnnotator.class);
+  
+  /**
+   * Pattern used for getting Tair Locus ID symbol from UniProt result page.
+   */
+  private Pattern getUniprotIdParsePattern(String tairId) {
+	  return Pattern.compile("(\\w*)\\tlocus:" + tairId);
+  }
+      
 
   /**
    * Default constructor.
@@ -74,7 +91,13 @@ public class TairAnnotator extends ElementAnnotator implements IExternalService
   @Override
   public boolean annotateElement(BioEntityProxy object, MiriamData identifier, AnnotatorData parameters)
       throws AnnotatorException {
+	  
+	  if (identifier.getAnnotator() != null) {
+		  return false;
+	  }
     if (identifier.getDataType().equals(MiriamType.TAIR_LOCUS)) {
+    	//UniProt are only obained from TAIR's which were provided by the annotator (otherwise we would get
+    	//also UniProt IDs for, e.g., homologous genes' TAIR IDs obtained from KEGG
       Collection<MiriamData> collection = tairToUniprot(identifier);
       if (collection.size() > 0) {
         object.addMiriamData(collection);
@@ -97,8 +120,8 @@ public class TairAnnotator extends ElementAnnotator implements IExternalService
    * @return URL to UniProt result page with the TAIR mapping
    */
   private String getUniProtUrl(String tairId) {
-    return "https://www.uniprot.org/uniprot/?query=database%3A%28type%3Atair%29+database%3A%28type%3Atair+" + tairId
-        + "%29&format=list&columns=id";
+	  return "https://www.uniprot.org/uniprot/?query=database%3A%28type%3Atair%29+" + tairId + "&format=tab&columns=id,database(tair)";
+//	  return "https://www.uniprot.org/uniprot/?query=database%3A%28type%3Atair+"+tairId+"%29&format=list&columns=id";
   }
 
   /**
@@ -110,11 +133,14 @@ public class TairAnnotator extends ElementAnnotator implements IExternalService
    *          uniprot REST API result page
    * @return uniprot identifier found on the page
    */
-  private Collection<MiriamData> parseUniprotUniprot(String pageContent) {
+  private Collection<MiriamData> parseUniprotUniprot(String pageContent, String tairId) {
     Collection<MiriamData> result = new HashSet<MiriamData>();
     if (!pageContent.isEmpty()) {
-      String[] sPageContent = pageContent.split("\\R");
-      result.add(new MiriamData(MiriamType.UNIPROT, sPageContent[0]));
+    	//the query returns a list of possible matches which needs to be pruned
+    	Matcher m = getUniprotIdParsePattern(tairId).matcher(pageContent);
+    	if (m.find()) {
+    		result.add(new MiriamData(MiriamType.UNIPROT, m.group(1)));  		
+    	} 
     }
     return result;
   }
@@ -144,7 +170,7 @@ public class TairAnnotator extends ElementAnnotator implements IExternalService
     try {
       String accessUrl = getUniProtUrl(tair.getResource());
       String pageContent = getWebPageContent(accessUrl);
-      return parseUniprotUniprot(pageContent);
+      return parseUniprotUniprot(pageContent, tair.getResource());
     } catch (WrongResponseCodeIOException exception) {
       logger.warn("Wrong reponse code when accessing tair data with id: " + tair.getResource());
       return null;
@@ -185,7 +211,7 @@ public class TairAnnotator extends ElementAnnotator implements IExternalService
 
   @Override
   public MiriamData getExampleValidAnnotation() {
-    return new MiriamData(MiriamType.TAIR_LOCUS, "AT1G01030");
+    return new MiriamData(MiriamType.TAIR_LOCUS, "2200950");
   }
 
 }

annotation/src/main/java/lcsb/mapviewer/annotation/services/annotators/UniprotAnnotator.java

@@ -59,6 +59,12 @@ public class UniprotAnnotator extends ElementAnnotator implements IExternalServi
    */
   private Pattern uniprotToEC = Pattern
       .compile("EC=((\\d+\\.-\\.-\\.-)|(\\d+\\.\\d+\\.-\\.-)|(\\d+\\.\\d+\\.\\d+\\.-)|(\\d+\\.\\d+\\.\\d+\\.\\d+))");
+  
+  /**
+   * Pattern used for getting Tair Locus ID symbol from UniProt result page.
+   */
+  private Pattern uniprotTairLocusToId = Pattern
+      .compile("locus:(\\d*)");
 
   /**
    * Default constructor.
@@ -136,6 +142,18 @@ public class UniprotAnnotator extends ElementAnnotator implements IExternalServi
   private String getUniprotUrl(String uniprotId) {
     return "https://www.uniprot.org/uniprot/" + uniprotId + ".txt";
   }
+  
+  /**
+   * Returns URL to UniProt result containing mapped UniProtIds for submitted TAIR
+   * entry.
+   * 
+   * @param tairId
+   *          TAIR identifier
+   * @return URL to UniProt result page with the TAIR mapping
+   */
+  private String getUniProtTairLocus2IdUrl(String tairLocusName) {
+    return "https://www.uniprot.org/uniprot/?query=database%3A%28type%3Atair+" + tairLocusName + "%29&format=tab&columns=id,database(tair)";
+  }
 
   /**
    * Parse uniprot webpage to find information about {@link MiriamType#ENTREZ} and
@@ -250,7 +268,34 @@ public class UniprotAnnotator extends ElementAnnotator implements IExternalServi
     }
 
   }
-
+  
+  /**
+   * Transform TAIR Locus name into TAIR Locus identifier.
+   * UniProt is used for this task because TAIR i) does not have
+   * an API and ii) restricts the number of accesses.
+   * 
+   * @param tairLocus
+   *          String with the TAIR Locus name.
+   * @return {@link MiriamData} with TAIR Locus ID
+   * @throws UniprotSearchException
+   *           thrown when there is a problem with accessing external database
+   */
+  public MiriamData uniprotTairLocusNameToId(String tairLocus) throws UniprotSearchException {
+	  String accessUrl = getUniProtTairLocus2IdUrl(tairLocus);
+	    try {
+	    	String pageContent = getWebPageContent(accessUrl);
+	    	Matcher m = uniprotTairLocusToId.matcher(pageContent);
+	    	if (m.find()) {
+	    		return new MiriamData(MiriamType.TAIR_LOCUS, m.group(1));	    		
+	    	} else {
+	    		logger.warn("No TAIR ID found for locus: " + tairLocus);
+	    		return new MiriamData();	    		
+	    	}	      
+	    } catch (IOException e) {
+	    	throw new UniprotSearchException("Problem with accessing uniprot webpage", e);
+	    }	  
+  }
+  
   @Override
   public String getCommonName() {
     return MiriamType.UNIPROT.getCommonName();

annotation/src/test/java/lcsb/mapviewer/annotation/services/annotators/KeggAnnotatorTest.java

@@ -189,7 +189,7 @@ public class KeggAnnotatorTest extends AnnotationTestFunctions {
     try {
       Species protein = new GenericProtein("id");
       protein.setName("bla");
-      protein.addMiriamData(new MiriamData(MiriamType.TAIR_LOCUS, "AT3G25110"));
+      protein.addMiriamData(new MiriamData(MiriamType.TAIR_LOCUS, "2090285")); //TAIR locus AT3G25110
 
       keggAnnotator.annotateElement(protein);
 

annotation/src/test/java/lcsb/mapviewer/annotation/services/annotators/StringAnnotatorTest.java

@@ -57,12 +57,12 @@ public class StringAnnotatorTest extends AnnotationTestFunctions {
   }
 
   @Test
-  @Ignore("TAIR DB restricts queries by IP")
+//  @Ignore("TAIR DB restricts queries by IP")
   public void testAnnotateTair() throws Exception {
     try {
 
       Species bioEntity = new GenericProtein("id");
-      bioEntity.addMiriamData(new MiriamData(MiriamType.TAIR_LOCUS, "AT1G01030"));
+      bioEntity.addMiriamData(new MiriamData(MiriamType.TAIR_LOCUS, "2200950"));
 
       testedAnnotator.annotateElement(bioEntity);
 
@@ -70,18 +70,35 @@ public class StringAnnotatorTest extends AnnotationTestFunctions {
 
       for (MiriamData md : bioEntity.getMiriamData()) {
         if (md.getDataType().equals(MiriamType.STRING)) {
-          mdString = md; // there should be only one EC number for that TAIR<->UNIPROT record
+          mdString = md; 
         }
       }
 
       assertTrue("No STRING annotation extracted from STRING annotator", mdString != null);
-      assertTrue("Wrong number of annotations extract from STRING annotator", bioEntity.getMiriamData().size() == 3);
+      assertTrue("Wrong number of annotations extract from STRING annotator", bioEntity.getMiriamData().size() == 2);
 
     } catch (Exception e) {
       e.printStackTrace();
       throw e;
     }
   }
+  
+  @Test
+//@Ignore("TAIR DB restricts queries by IP")
+public void testAnnotateTairOnlyFromHumanAnnotator() throws Exception {
+  try {
+
+    Species bioEntity = new GenericProtein("id");
+    bioEntity.addMiriamData(new MiriamData(MiriamType.TAIR_LOCUS, "2200950", KeggAnnotator.class));
+
+    testedAnnotator.annotateElement(bioEntity);
+
+    assertTrue(bioEntity.getMiriamData().size() == 1);
+  } catch (Exception e) {
+    e.printStackTrace();
+    throw e;
+  }
+}
 
   @Test
   public void testAnnotateInvalidEmpty() throws Exception {

annotation/src/test/java/lcsb/mapviewer/annotation/services/annotators/TairAnnotatorTest.java

 package lcsb.mapviewer.annotation.services.annotators;
 
 import static org.junit.Assert.assertEquals;
+import static org.junit.Assert.assertFalse;
 import static org.junit.Assert.assertNull;
 import static org.junit.Assert.assertTrue;
 import static org.junit.Assert.fail;
@@ -45,7 +46,7 @@ public class TairAnnotatorTest extends AnnotationTestFunctions {
 
       Species protein = new GenericProtein("id");
       protein.setName("bla");
-      protein.addMiriamData(new MiriamData(MiriamType.TAIR_LOCUS, "AT1G01030"));
+      protein.addMiriamData(new MiriamData(MiriamType.TAIR_LOCUS, "2200950")); //AT1G01030
 
       tairAnnotator.annotateElement(protein);
 
@@ -74,7 +75,7 @@ public class TairAnnotatorTest extends AnnotationTestFunctions {
 
       Species protein = new GenericProtein("id");
       protein.setName("bla");
-      protein.addMiriamData(new MiriamData(MiriamType.TAIR_LOCUS, "AT1G15950"));
+      protein.addMiriamData(new MiriamData(MiriamType.TAIR_LOCUS, "2200427"));
       protein.addMiriamData(new MiriamData(MiriamType.UNIPROT, "P32246")); // Human version of the protein
 
       tairAnnotator.annotateElement(protein);
@@ -156,13 +157,29 @@ public class TairAnnotatorTest extends AnnotationTestFunctions {
   // @Ignore("TAIR DB restricts queries by IP")
   public void testTairToUniprot() throws Exception {
     try {
-      assertTrue(tairAnnotator.tairToUniprot(new MiriamData(MiriamType.TAIR_LOCUS, "AT1G01030"))
+      assertTrue(tairAnnotator.tairToUniprot(new MiriamData(MiriamType.TAIR_LOCUS, "2200950")) // TAIR locus AT1G01030
           .contains(new MiriamData(MiriamType.UNIPROT, "Q9MAN1")));
     } catch (Exception e) {
       e.printStackTrace();
       throw e;
     }
   }
+  
+  @Test
+  public void testTairToUniprotFromKEGG() throws Exception {
+	  //TAIR Loci comming from annotators should be ignored by TAIR (only TAIR LOCI provided by the human annotator should be considered)
+    try {
+    	Species protein = new GenericProtein("id");
+        protein.setName("bla");
+        protein.addMiriamData(new MiriamData(MiriamType.TAIR_LOCUS, "2200427", KeggAnnotator.class));
+        tairAnnotator.annotateElement(protein);
+        assertTrue(  protein.getMiriamData().size() == 1);
+    } catch (Exception e) {
+      e.printStackTrace();
+      throw e;
+    }
+  }
+  
 
   @Test
   // @Ignore("TAIR DB restricts queries by IP")

annotation/src/test/java/lcsb/mapviewer/annotation/services/annotators/UniprotAnnotatorTest.java

@@ -278,6 +278,30 @@ public class UniprotAnnotatorTest extends AnnotationTestFunctions {
       throw e;
     }
   }
+  
+  @Test
+  public void testUniprotTairLocusToId() throws Exception {
+    try {
+      assertEquals(new MiriamData(MiriamType.TAIR_LOCUS, "2201786"),
+    		  uniprotAnnotator.uniprotTairLocusNameToId("AT1G08510") );
+    } catch (Exception e) {
+      e.printStackTrace();
+      throw e;
+    }
+  }
+  
+  @Test
+  public void testUniprotWrongTairLocusToId() throws Exception {
+    try {
+    	assertEquals(new MiriamData(),
+    	          uniprotAnnotator.uniprotTairLocusNameToId("bla"));
+    } catch (Exception e) {
+      e.printStackTrace();
+      throw e;
+    }
+  }
+  
+  
 
   @Test
   public void testStatus() throws Exception {