JAL-3700 fix for propagation of sort commands

[jalview.git] / src / jalview / datamodel / AlignedCodonFrame.java

diff --git a/src/jalview/datamodel/AlignedCodonFrame.java b/src/jalview/datamodel/AlignedCodonFrame.java
index ec11fc1..6103df5 100644 (file)
--- a/src/jalview/datamodel/AlignedCodonFrame.java
+++ b/src/jalview/datamodel/AlignedCodonFrame.java
@@ -107,6 +107,55 @@ 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)
+    {
+      List<int[]> mappedRanges = null;
+      MapList mapList = mapping.getMap();
+      if (fromSeq == seq || fromSeq == seq.getDatasetSequence())
+      {
+        mappedRanges = mapList.getFromRanges();
+      }
+      else if (mapping.to == seq || mapping.to == seq.getDatasetSequence())
+      {
+        mappedRanges = mapList.getToRanges();
+      }
+      else
+      {
+        return false;
+      }
+
+      /*
+       * check that each mapped range lieS with the sequence range
+       * (necessary for circular CDS - example EMBL:J03321:AAA91567)
+       * and mapped length covers (at least) sequence length
+       */
+      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 < seq.getStart() || to > seq.getEnd())
+        {
+          return false;
+        }
+        length += (to - from + 1);
+      }
+      // add 1 to mapped length to allow for a mapped stop codon
+      if (length + 1 < (seq.getEnd() - seq.getStart() + 1))
+      {
+        return false;
+      }
+      return true;
+    }
   }
 
   private List<SequenceToSequenceMapping> mappings;
@@ -116,7 +165,7 @@ public class AlignedCodonFrame
    */
   public AlignedCodonFrame()
   {
-    mappings = new ArrayList<SequenceToSequenceMapping>();
+    mappings = new ArrayList<>();
   }
 
   /**
@@ -179,7 +228,7 @@ 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);
@@ -190,7 +239,7 @@ 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);
@@ -200,7 +249,7 @@ 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);
@@ -210,7 +259,7 @@ 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);
@@ -220,7 +269,7 @@ 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
@@ -261,9 +310,12 @@ public class AlignedCodonFrame
   }
 
   /**
+   * Return the corresponding aligned or dataset dna sequence for given amino
+   * acid sequence, or null if not found. returns the sequence from the first
+   * mapping found that involves the protein sequence.
    * 
-   * @param sequenceRef
-   * @return null or corresponding aaSeq entry for dnaSeq entry
+   * @param aaSeqRef
+   * @return
    */
   public SequenceI getDnaForAaSeq(SequenceI aaSeqRef)
   {
@@ -485,7 +537,7 @@ public class AlignedCodonFrame
   {
     MapList ml = null;
     SequenceI dnaSeq = null;
-    List<char[]> result = new ArrayList<char[]>();
+    List<char[]> result = new ArrayList<>();
 
     for (SequenceToSequenceMapping ssm : mappings)
     {
@@ -524,8 +576,8 @@ 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;
 
@@ -767,7 +819,7 @@ public class AlignedCodonFrame
    * Two AlignedCodonFrame objects are equal if they hold the same ordered list
    * of mappings
    * 
-   * @see SequenceToSequenceMapping#
+   * @see SequenceToSequenceMapping#equals
    */
   @Override
   public boolean equals(Object obj)
@@ -783,4 +835,25 @@ public class AlignedCodonFrame
   {
     return mappings;
   }
+
+  /**
+   * Returns the first mapping found which is between the two given sequences,
+   * and covers the full extent of both.
+   * 
+   * @param seq1
+   * @param seq2
+   * @return
+   */
+  public SequenceToSequenceMapping getCoveringMapping(SequenceI seq1,
+          SequenceI seq2)
+  {
+    for (SequenceToSequenceMapping mapping : mappings)
+    {
+      if (mapping.covers(seq2) && mapping.covers(seq1))
+      {
+        return mapping;
+      }
+    }
+    return null;
+  }
 }