Merge branch 'develop' into feature/JAL-2759

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

diff --git a/src/jalview/datamodel/Sequence.java b/src/jalview/datamodel/Sequence.java
index d254a17..9c2fa25 100755 (executable)
--- a/src/jalview/datamodel/Sequence.java
+++ b/src/jalview/datamodel/Sequence.java
@@ -408,7 +408,7 @@ public class Sequence extends ASequence implements SequenceI
   {
     if (pdbIds == null)
     {
-      pdbIds = new Vector<PDBEntry>();
+      pdbIds = new Vector<>();
       pdbIds.add(entry);
       return true;
     }
@@ -1071,6 +1071,27 @@ public class Sequence extends ASequence implements SequenceI
     return map;
   }
 
+  /**
+   * Build a bitset corresponding to sequence gaps
+   * 
+   * @return a BitSet where set values correspond to gaps in the sequence
+   */
+  @Override
+  public BitSet gapBitset()
+  {
+    BitSet gaps = new BitSet(sequence.length);
+    int j = 0;
+    while (j < sequence.length)
+    {
+      if (jalview.util.Comparison.isGap(sequence[j]))
+      {
+        gaps.set(j);
+      }
+      j++;
+    }
+    return gaps;
+  }
+
   @Override
   public int[] findPositionMap()
   {
@@ -1094,7 +1115,7 @@ public class Sequence extends ASequence implements SequenceI
   @Override
   public List<int[]> getInsertions()
   {
-    ArrayList<int[]> map = new ArrayList<int[]>();
+    ArrayList<int[]> map = new ArrayList<>();
     int lastj = -1, j = 0;
     int pos = start;
     int seqlen = sequence.length;
@@ -1160,7 +1181,7 @@ public class Sequence extends ASequence implements SequenceI
   }
 
   @Override
-  public void deleteChars(int i, int j)
+  public void deleteChars(final int i, final int j)
   {
     int newstart = start, newend = end;
     if (i >= sequence.length || i < 0)
@@ -1172,62 +1193,75 @@ public class Sequence extends ASequence implements SequenceI
     boolean createNewDs = false;
     // TODO: take a (second look) at the dataset creation validation method for
     // the very large sequence case
-    int eindex = -1, sindex = -1;
-    boolean ecalc = false, scalc = false;
+    int startIndex = findIndex(start) - 1;
+    int endIndex = findIndex(end) - 1;
+    int startDeleteColumn = -1; // for dataset sequence deletions
+    int deleteCount = 0;
+
     for (int s = i; s < j; s++)
     {
-      if (jalview.schemes.ResidueProperties.aaIndex[sequence[s]] != 23)
+      if (Comparison.isGap(sequence[s]))
+      {
+        continue;
+      }
+      deleteCount++;
+      if (startDeleteColumn == -1)
+      {
+        startDeleteColumn = findPosition(s) - start;
+      }
+      if (createNewDs)
       {
-        if (createNewDs)
+        newend--;
+      }
+      else
+      {
+        if (startIndex == s)
         {
-          newend--;
+          /*
+           * deleting characters from start of sequence; new start is the
+           * sequence position of the next column (position to the right
+           * if the column position is gapped)
+           */
+          newstart = findPosition(j);
+          break;
         }
         else
         {
-          if (!scalc)
-          {
-            sindex = findIndex(start) - 1;
-            scalc = true;
-          }
-          if (sindex == s)
+          if (endIndex < j)
           {
-            // delete characters including start of sequence
-            newstart = findPosition(j);
-            break; // don't need to search for any more residue characters.
+            /*
+             * deleting characters at end of sequence; new end is the sequence
+             * position of the column before the deletion; subtract 1 if this is
+             * gapped since findPosition returns the next sequence position
+             */
+            newend = findPosition(i - 1);
+            if (Comparison.isGap(sequence[i - 1]))
+            {
+              newend--;
+            }
+            break;
           }
           else
           {
-            // delete characters after start.
-            if (!ecalc)
-            {
-              eindex = findIndex(end) - 1;
-              ecalc = true;
-            }
-            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
-            }
+            createNewDs = true;
+            newend--;
           }
         }
       }
     }
-    // deletion occured in the middle of the sequence
+
     if (createNewDs && this.datasetSequence != null)
     {
-      // construct a new sequence
+      /*
+       * if deletion occured in the middle of the sequence,
+       * construct a new dataset sequence and delete the residues
+       * that were deleted from the aligned sequence
+       */
       Sequence ds = new Sequence(datasetSequence);
+      ds.deleteChars(startDeleteColumn, startDeleteColumn + deleteCount);
+      datasetSequence = ds;
       // 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;
@@ -1392,7 +1426,7 @@ public class Sequence extends ASequence implements SequenceI
   {
     if (this.annotation == null)
     {
-      this.annotation = new Vector<AlignmentAnnotation>();
+      this.annotation = new Vector<>();
     }
     if (!this.annotation.contains(annotation))
     {
@@ -1559,7 +1593,7 @@ public class Sequence extends ASequence implements SequenceI
       return null;
     }
 
-    Vector<AlignmentAnnotation> subset = new Vector<AlignmentAnnotation>();
+    Vector<AlignmentAnnotation> subset = new Vector<>();
     Enumeration<AlignmentAnnotation> e = annotation.elements();
     while (e.hasMoreElements())
     {
@@ -1693,7 +1727,7 @@ public class Sequence extends ASequence implements SequenceI
   public List<AlignmentAnnotation> getAlignmentAnnotations(String calcId,
           String label)
   {
-    List<AlignmentAnnotation> result = new ArrayList<AlignmentAnnotation>();
+    List<AlignmentAnnotation> result = new ArrayList<>();
     if (this.annotation != null)
     {
       for (AlignmentAnnotation ann : annotation)
@@ -1749,7 +1783,7 @@ public class Sequence extends ASequence implements SequenceI
     }
     synchronized (dbrefs)
     {
-      List<DBRefEntry> primaries = new ArrayList<DBRefEntry>();
+      List<DBRefEntry> primaries = new ArrayList<>();
       DBRefEntry[] tmp = new DBRefEntry[1];
       for (DBRefEntry ref : dbrefs)
       {