added aligned codon frames, and dataset reference ids for sequenceSet and sequences...

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

diff --git a/src/jalview/datamodel/Sequence.java b/src/jalview/datamodel/Sequence.java
index d64e133..257f88a 100755 (executable)
--- a/src/jalview/datamodel/Sequence.java
+++ b/src/jalview/datamodel/Sequence.java
@@ -24,7 +24,8 @@ import java.util.*;
 import jalview.analysis.*;
 
 /**
- * DOCUMENT ME!
+ * 
+ * Implements the SequenceI interface for a char[] based sequence object.
  *
  * @author $author$
  * @version $Revision$
@@ -46,17 +47,17 @@ public class Sequence
    * positions are tied to the residues of this sequence */
   Vector annotation;
 
-  /** DOCUMENT ME!! */
+  /** array of seuqence features - may not be null for a valid sequence object */
   public SequenceFeature[] sequenceFeatures;
 
 
   /**
    * Creates a new Sequence object.
    *
-   * @param name DOCUMENT ME!
-   * @param sequence DOCUMENT ME!
-   * @param start DOCUMENT ME!
-   * @param end DOCUMENT ME!
+   * @param name display name string
+   * @param sequence string to form a possibly gapped sequence out of
+   * @param start first position of non-gap residue in the sequence
+   * @param end last position of ungapped residues (nearly always only used for display purposes)
    */
   public Sequence(String name, String sequence, int start, int end)
   {
@@ -85,6 +86,11 @@ public class Sequence
 
   void parseId()
   {
+    if (name==null)
+    {
+      System.err.println("POSSIBLE IMPLEMENTATION ERROR: null sequence name passed to constructor.");
+      name = "";
+    }
     // Does sequence have the /start-end signiature?
     if (limitrx.search(name))
     {
@@ -165,11 +171,15 @@ public class Sequence
       }
     }
     setDatasetSequence(seq.getDatasetSequence());
-    if (seq.getAnnotation()!=null && alAnnotation!=null) {
+    if (seq.getAnnotation()!=null) {
       AlignmentAnnotation[] sqann = seq.getAnnotation();
       for (int i=0;i<sqann.length; i++)
       {
-        boolean found = sqann == alAnnotation;
+        if (sqann[i]==null)
+        {
+          continue;
+        }
+        boolean found = (alAnnotation==null);
         if (!found)
         {
           for (int apos = 0; !found && apos<alAnnotation.length; apos++)
@@ -177,8 +187,12 @@ public class Sequence
             found = (alAnnotation[apos] == sqann[i]);
           }
         }
-        AlignmentAnnotation newann = new AlignmentAnnotation(sqann[i]);
-        addAlignmentAnnotation(newann);
+        if (found)
+        {
+          // only copy the given annotation
+          AlignmentAnnotation newann = new AlignmentAnnotation(sqann[i]);
+          addAlignmentAnnotation(newann);
+        }
       }
     }
     if (seq.getPDBId()!=null) {
@@ -607,14 +621,12 @@ public class Sequence
     return map;
   }
 
-  /**
-   * DOCUMENT ME!
-   *
-   * @param i DOCUMENT ME!
-   * @param j DOCUMENT ME!
+  /* (non-Javadoc)
+   * @see jalview.datamodel.SequenceI#deleteChars(int, int)
    */
   public void deleteChars(int i, int j)
   {
+    int newstart=start,newend=end;
     if (i >= sequence.length)
     {
       return;
@@ -633,29 +645,50 @@ public class Sequence
       System.arraycopy(sequence,0,tmp,0,i);
       System.arraycopy(sequence,j,tmp,i,sequence.length-j);
     }
-
-    if (this.datasetSequence != null)
+    boolean createNewDs=false;
+    for (int s = i; s < j; s++)
     {
-      for (int s = i; s < j; s++)
+      if (jalview.schemes.ResidueProperties.aaIndex[sequence[s]] != 23)
       {
-        if (jalview.schemes.ResidueProperties.aaIndex[sequence[s]] != 23)
+        if (createNewDs)
         {
-
-          Sequence ds = new Sequence(name,
-                                     AlignSeq.extractGaps(
-                                         jalview.util.Comparison.GapChars,
-                                         this.getSequenceAsString()
-                                     ),
-                                     start,
-                                     end);
-          ds.setDescription(description);
+          newend--;
+        } else {
+          int sindex = findIndex(start)-1;
+          if (sindex==s)
+        {
+          // delete characters including start of sequence
+          newstart = findPosition(j);
+          break; // don't need to search for any more residue characters.
+        } else {
+          // delete characters after start.
+          int eindex = findIndex(end)-1;
+          if (eindex<j)
+          {
+            // delete characters at end of sequence
+            newend = findPosition(i-1);
+            break; // don't need to search for any more residue characters.
+          } else {
+            createNewDs=true;
+            newend--; // decrease end position by one for the deleted residue and search further
+          }
+        }
         }
-        break;
       }
     }
-
+    // deletion occured in the middle of the sequence
+    if (createNewDs && this.datasetSequence != null)
+    {
+      // construct a new sequence
+      Sequence ds = new Sequence(datasetSequence);
+      // TODO: remove any non-inheritable properties ?
+      // TODO: create a sequence mapping (since there is a relation here ?)
+      ds.deleteChars(i, j);
+      datasetSequence = ds;
+    }
+    start = newstart;
+    end = newend;
     sequence = tmp;
-
   }
 
 
@@ -732,8 +765,16 @@ public class Sequence
 
     for(i=0; i<iSize; i++)
     {
-      if(dbrefs[i].equals(entry))
+      if(dbrefs[i].equalRef(entry))
       {
+        if (entry.getMap()!=null)
+        {
+          if (dbrefs[i].getMap()==null)
+          {
+            // overwrite with 'superior' entry that contains a mapping.
+            dbrefs[i] = entry;
+          }
+        }
         return;
       }
     }
@@ -835,6 +876,7 @@ public class Sequence
         SequenceI ds = seq;
         ds.setSequence(AlignSeq.extractGaps(jalview.util.Comparison.GapChars, new String(sequence)));
         setDatasetSequence(ds);
+        ds.setSequenceFeatures(getSequenceFeatures());
         seq = this; // and return this sequence as the derived sequence.
       }
     }