X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fdatamodel%2FAlignedCodonFrame.java;h=3c33063faf0dfd2008a743740f0ceb0f0dd81946;hb=b1f3e3bac931d7f309f60e50b18c435c65586de8;hp=5d00b6b3c5b5450beb9e05f6953454506df5eec0;hpb=cbdd0bde03d210649623b4576dd75e6c25fe4582;p=jalview.git

diff --git a/src/jalview/datamodel/AlignedCodonFrame.java b/src/jalview/datamodel/AlignedCodonFrame.java
index 5d00b6b..3c33063 100644
--- a/src/jalview/datamodel/AlignedCodonFrame.java
+++ b/src/jalview/datamodel/AlignedCodonFrame.java
@@ -60,14 +60,15 @@ public class AlignedCodonFrame
     }
 
     /**
-     * Returns a hashCode derived from the hashcodes of the mappings
+     * Returns a hashCode derived from the hashcodes of the mappings and fromSeq
      * 
      * @see SequenceToSequenceMapping#hashCode()
      */
     @Override
     public int hashCode()
     {
-      return mappings.hashCode();
+      return (fromSeq == null ? 0 : fromSeq.hashCode() * 31)
+              + mapping.hashCode();
     }
 
     /**
@@ -88,7 +89,13 @@ public class AlignedCodonFrame
       {
         return that.mapping == null;
       }
-      return this.mapping.equals(that.mapping);
+      // TODO: can simplify by asserting fromSeq is a dataset sequence
+      return (this.fromSeq == that.fromSeq
+              || (this.fromSeq != null && that.fromSeq != null
+                      && this.fromSeq.getDatasetSequence() != null
+                      && this.fromSeq.getDatasetSequence() == that.fromSeq
+                              .getDatasetSequence()))
+              && this.mapping.equals(that.mapping);
     }
 
     public SequenceI getFromSeq()
@@ -100,6 +107,143 @@ public class AlignedCodonFrame
     {
       return mapping;
     }
+    
+    /**
+     * Returns true if the mapping covers the full length of the given sequence.
+     * This allows us to distinguish the CDS that codes for a protein from
+     * another overlapping CDS in the parent dna sequence.
+     * 
+     * @param seq
+     * @return
+     */
+    public boolean covers(SequenceI seq)
+    {
+      return covers(seq,false,false);
+    }
+    /**
+     * 
+     * @param seq
+     * @param localCover - when true - compare extent of seq's dataset sequence rather than the local extent
+     * @param either - when true coverage is required for either seq or the mapped sequence 
+     * @return true if mapping covers full length of given sequence (or the other if either==true)
+     */
+    public boolean covers(SequenceI seq, boolean localCover,boolean either)
+    {
+      List<int[]> mappedRanges = null,otherRanges=null;
+      MapList mapList = mapping.getMap();
+      int mstart=seq.getStart(),mend=seq.getEnd(),ostart,oend;
+              ;
+      if (fromSeq == seq || fromSeq == seq.getDatasetSequence())
+      {
+        if (localCover && fromSeq !=seq)
+        {
+          mstart=fromSeq.getStart();
+          mend=fromSeq.getEnd();
+        }
+        mappedRanges = mapList.getFromRanges();
+        otherRanges=mapList.getToRanges();
+        ostart=mapping.to.getStart();
+        oend=mapping.to.getEnd();
+      }
+      else if (mapping.to == seq || mapping.to == seq.getDatasetSequence())
+      {
+        if (localCover && mapping.to !=seq)
+        {
+          mstart=mapping.to.getStart();
+          mend=mapping.to.getEnd();
+        }
+        mappedRanges = mapList.getToRanges();
+        otherRanges=mapList.getFromRanges();
+        ostart=fromSeq.getStart();
+        oend=fromSeq.getEnd();
+      }
+      else
+      {
+        return false;
+      }
+
+      /*
+       * check that each mapped range lies within the sequence range
+       * (necessary for circular CDS - example EMBL:J03321:AAA91567)
+       * and mapped length covers (at least) sequence length
+       */
+      int length = countRange(mappedRanges,mstart,mend);
+
+      if (length != -1)
+      {
+        // add 3 to mapped length to allow for a mapped stop codon
+        if (length + 3 >= (mend - mstart + 1))
+        {
+          return true;
+        }
+      }
+      if (either)
+      {
+        // also check coverage of the other range
+        length = countRange(otherRanges, ostart, oend);
+        if (length != -1)
+        {
+          if (length + 1 >= (oend - ostart + 1))
+          {
+            return true;
+          }
+        }
+      }
+      return false;
+    }
+    private int countRange(List<int[]> mappedRanges,int mstart,int mend)
+    {
+      int length=0;
+      for (int[] range : mappedRanges)
+      {
+        int from = Math.min(range[0], range[1]);
+        int to = Math.max(range[0], range[1]);
+        if (from < mstart || to > mend)
+        {
+          return -1;
+        }
+        length += (to - from + 1);
+      }
+      return length;
+    }
+
+    /**
+     * Adds any regions mapped to or from position {@code pos} in sequence
+     * {@code seq} to the given search results
+     * 
+     * @param seq
+     * @param pos
+     * @param sr
+     */
+    public void markMappedRegion(SequenceI seq, int pos, SearchResultsI sr)
+    {
+      int[] codon = null;
+      SequenceI mappedSeq = null;
+      SequenceI ds = seq.getDatasetSequence();
+      if (ds == null)
+      {
+        ds = seq;
+      }
+      
+      if (this.fromSeq == seq || this.fromSeq == ds)
+      {
+        codon = this.mapping.map.locateInTo(pos, pos);
+        mappedSeq = this.mapping.to;
+      }
+      else if (this.mapping.to == seq || this.mapping.to == ds)
+      {
+        codon = this.mapping.map.locateInFrom(pos, pos);
+        mappedSeq = this.fromSeq;
+      }
+
+      if (codon != null)
+      {
+        for (int i = 0; i < codon.length; i += 2)
+        {
+          sr.addResult(mappedSeq, codon[i], codon[i + 1]);
+        }
+      }
+    }
   }
 
   private List<SequenceToSequenceMapping> mappings;
@@ -109,7 +253,7 @@ public class AlignedCodonFrame
    */
   public AlignedCodonFrame()
   {
-    mappings = new ArrayList<SequenceToSequenceMapping>();
+    mappings = new ArrayList<>();
   }
 
   /**
@@ -122,14 +266,29 @@ public class AlignedCodonFrame
    */
   public void addMap(SequenceI dnaseq, SequenceI aaseq, MapList map)
   {
+    addMap(dnaseq, aaseq, map, null);
+  }
+
+  /**
+   * Adds a mapping between the dataset sequences for the associated dna and
+   * protein sequence objects
+   * 
+   * @param dnaseq
+   * @param aaseq
+   * @param map
+   * @param mapFromId
+   */
+  public void addMap(SequenceI dnaseq, SequenceI aaseq, MapList map,
+          String mapFromId)
+  {
     // JBPNote DEBUG! THIS !
     // dnaseq.transferAnnotation(aaseq, mp);
     // aaseq.transferAnnotation(dnaseq, new Mapping(map.getInverse()));
 
     SequenceI fromSeq = (dnaseq.getDatasetSequence() == null) ? dnaseq
             : dnaseq.getDatasetSequence();
-    SequenceI toSeq = (aaseq.getDatasetSequence() == null) ? aaseq : aaseq
-            .getDatasetSequence();
+    SequenceI toSeq = (aaseq.getDatasetSequence() == null) ? aaseq
+            : aaseq.getDatasetSequence();
 
     /*
      * if we already hold a mapping between these sequences, just add to it 
@@ -149,6 +308,7 @@ public class AlignedCodonFrame
      * otherwise, add a new sequence mapping
      */
     Mapping mp = new Mapping(toSeq, map);
+    mp.setMappedFromId(mapFromId);
     mappings.add(new SequenceToSequenceMapping(fromSeq, mp));
   }
 
@@ -156,7 +316,7 @@ public class AlignedCodonFrame
   {
     // TODO return a list instead?
     // return dnaSeqs;
-    List<SequenceI> seqs = new ArrayList<SequenceI>();
+    List<SequenceI> seqs = new ArrayList<>();
     for (SequenceToSequenceMapping ssm : mappings)
     {
       seqs.add(ssm.fromSeq);
@@ -167,7 +327,7 @@ public class AlignedCodonFrame
   public SequenceI[] getAaSeqs()
   {
     // TODO not used - remove?
-    List<SequenceI> seqs = new ArrayList<SequenceI>();
+    List<SequenceI> seqs = new ArrayList<>();
     for (SequenceToSequenceMapping ssm : mappings)
     {
       seqs.add(ssm.mapping.to);
@@ -177,7 +337,7 @@ public class AlignedCodonFrame
 
   public MapList[] getdnaToProt()
   {
-    List<MapList> maps = new ArrayList<MapList>();
+    List<MapList> maps = new ArrayList<>();
     for (SequenceToSequenceMapping ssm : mappings)
     {
       maps.add(ssm.mapping.map);
@@ -187,7 +347,7 @@ public class AlignedCodonFrame
 
   public Mapping[] getProtMappings()
   {
-    List<Mapping> maps = new ArrayList<Mapping>();
+    List<Mapping> maps = new ArrayList<>();
     for (SequenceToSequenceMapping ssm : mappings)
     {
       maps.add(ssm.mapping);
@@ -197,7 +357,7 @@ public class AlignedCodonFrame
 
   /**
    * Returns the first mapping found which is to or from the given sequence, or
-   * null.
+   * null if none is found
    * 
    * @param seq
    * @return
@@ -279,7 +439,7 @@ public class AlignedCodonFrame
    *          where highlighted regions go
    */
   public void markMappedRegion(SequenceI seq, int index,
-          SearchResults results)
+          SearchResultsI results)
   {
     int[] codon;
     SequenceI ds = seq.getDatasetSequence();
@@ -412,8 +572,8 @@ public class AlignedCodonFrame
   {
     SequenceI targetDs = target.getDatasetSequence() == null ? target
             : target.getDatasetSequence();
-    SequenceI queryDs = query.getDatasetSequence() == null ? query : query
-            .getDatasetSequence();
+    SequenceI queryDs = query.getDatasetSequence() == null ? query
+            : query.getDatasetSequence();
     if (targetDs == null || queryDs == null /*|| dnaToProt == null*/)
     {
       return null;
@@ -462,11 +622,12 @@ public class AlignedCodonFrame
   {
     MapList ml = null;
     SequenceI dnaSeq = null;
-    List<char[]> result = new ArrayList<char[]>();
+    List<char[]> result = new ArrayList<>();
 
     for (SequenceToSequenceMapping ssm : mappings)
     {
-      if (ssm.mapping.to == protein)
+      if (ssm.mapping.to == protein
+              && ssm.mapping.getMap().getFromRatio() == 3)
       {
         ml = ssm.mapping.map;
         dnaSeq = ssm.fromSeq;
@@ -481,10 +642,11 @@ public class AlignedCodonFrame
          * Read off the mapped nucleotides (converting to position base 0)
          */
         codonPos = MappingUtils.flattenRanges(codonPos);
-        char[] dna = dnaSeq.getSequence();
         int start = dnaSeq.getStart();
-        result.add(new char[] { dna[codonPos[0] - start],
-            dna[codonPos[1] - start], dna[codonPos[2] - start] });
+        char c1 = dnaSeq.getCharAt(codonPos[0] - start);
+        char c2 = dnaSeq.getCharAt(codonPos[1] - start);
+        char c3 = dnaSeq.getCharAt(codonPos[2] - start);
+        result.add(new char[] { c1, c2, c3 });
       }
     }
     return result.isEmpty() ? null : result;
@@ -499,8 +661,8 @@ public class AlignedCodonFrame
    */
   public List<Mapping> getMappingsFromSequence(SequenceI seq)
   {
-    List<Mapping> result = new ArrayList<Mapping>();
-    List<SequenceI> related = new ArrayList<SequenceI>();
+    List<Mapping> result = new ArrayList<>();
+    List<SequenceI> related = new ArrayList<>();
     SequenceI seqDs = seq.getDatasetSequence();
     seqDs = seqDs != null ? seqDs : seq;
 
@@ -556,8 +718,9 @@ public class AlignedCodonFrame
    */
   protected int realiseWith(SequenceI seq, boolean doUpdate)
   {
-    SequenceI ds = seq.getDatasetSequence() != null ? seq
-            .getDatasetSequence() : seq;
+    SequenceI ds = seq.getDatasetSequence() != null
+            ? seq.getDatasetSequence()
+            : seq;
     int count = 0;
 
     /*
@@ -631,8 +794,8 @@ public class AlignedCodonFrame
     {
       int start = replacement.getStart();
       int end = replacement.getEnd();
-      boolean mappingOverlapsSequence = (mapStart >= start && mapStart <= end)
-              || (mapEnd >= start && mapEnd <= end);
+      boolean mappingOverlapsSequence = (mapStart >= start
+              && mapStart <= end) || (mapEnd >= start && mapEnd <= end);
       if (mappingOverlapsSequence)
       {
         return true;
@@ -696,8 +859,8 @@ public class AlignedCodonFrame
   }
 
   /**
-   * Returns the first mapping found that is from 'fromSeq' to 'toSeq', or null
-   * if none found
+   * Returns the first mapping found that is between 'fromSeq' and 'toSeq', or
+   * null if none found
    * 
    * @param fromSeq
    *          aligned or dataset sequence
@@ -707,12 +870,17 @@ public class AlignedCodonFrame
    */
   public Mapping getMappingBetween(SequenceI fromSeq, SequenceI toSeq)
   {
+    SequenceI dssFrom = fromSeq.getDatasetSequence() == null ? fromSeq
+            : fromSeq.getDatasetSequence();
+    SequenceI dssTo = toSeq.getDatasetSequence() == null ? toSeq
+            : toSeq.getDatasetSequence();
+
     for (SequenceToSequenceMapping mapping : mappings)
     {
       SequenceI from = mapping.fromSeq;
       SequenceI to = mapping.mapping.to;
-      if ((from == fromSeq || from == fromSeq.getDatasetSequence())
-              && (to == toSeq || to == toSeq.getDatasetSequence()))
+      if ((from == dssFrom && to == dssTo)
+              || (from == dssTo && to == dssFrom))
       {
         return mapping.mapping;
       }