Merge branch 'features/pca_jaxb_datasetrefs_JAL-3171_JAL-3063_JAL-1767' into develop

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

diff --git a/src/jalview/datamodel/SequenceI.java b/src/jalview/datamodel/SequenceI.java
index 03fc545..48615f0 100755 (executable)
--- a/src/jalview/datamodel/SequenceI.java
+++ b/src/jalview/datamodel/SequenceI.java
@@ -24,6 +24,7 @@ import jalview.datamodel.features.SequenceFeaturesI;
 import jalview.util.MapList;
 
 import java.util.BitSet;
+import java.util.Iterator;
 import java.util.List;
 import java.util.Vector;
 
@@ -193,22 +194,26 @@ public interface SequenceI extends ASequenceI
   public int findIndex(int pos);
 
   /**
-   * Returns the sequence position for an alignment position.
+   * Returns the sequence position for an alignment (column) position. If at a
+   * gap, returns the position of the next residue to the right. If beyond the
+   * end of the sequence, returns 1 more than the last residue position.
    * 
    * @param i
    *          column index in alignment (from 0..<length)
    * 
-   * @return TODO: JAL-2562 - residue number for residue (left of and) nearest
-   *         ith column
+   * @return
    */
   public int findPosition(int i);
 
   /**
-   * Returns the from-to sequence positions (start..) for the given column
-   * positions (1..), or null if no residues are included in the range
+   * Returns the sequence positions for first and last residues lying within the
+   * given column positions [fromColum,toColumn] (where columns are numbered
+   * from 1), or null if no residues are included in the range
    * 
    * @param fromColum
+   *          - first column base 1
    * @param toColumn
+   *          - last column, base 1
    * @return
    */
   public Range findPositions(int fromColum, int toColumn);
@@ -223,6 +228,13 @@ public interface SequenceI extends ASequenceI
   public int[] gapMap();
 
   /**
+   * Build a bitset corresponding to sequence gaps
+   * 
+   * @return a BitSet where set values correspond to gaps in the sequence
+   */
+  public BitSet gapBitset();
+
+  /**
    * Returns an int array where indices correspond to each position in sequence
    * char array and the element value gives the result of findPosition for that
    * index in the sequence.
@@ -450,17 +462,6 @@ public interface SequenceI extends ASequenceI
   public void transferAnnotation(SequenceI entry, Mapping mp);
 
   /**
-   * @param index
-   *          The sequence index in the MSA
-   */
-  public void setIndex(int index);
-
-  /**
-   * @return The index of the sequence in the alignment
-   */
-  public int getIndex();
-
-  /**
    * @return The RNA of the sequence in the alignment
    */
 
@@ -536,8 +537,43 @@ public interface SequenceI extends ASequenceI
    */
   public int replace(char c1, char c2);
 
+  /**
+   * Answers the GeneLociI, or null if not known
+   * 
+   * @return
+   */
   GeneLociI getGeneLoci();
 
+  /**
+   * Sets the mapping to gene loci for the sequence
+   * 
+   * @param speciesId
+   * @param assemblyId
+   * @param chromosomeId
+   * @param map
+   */
   void setGeneLoci(String speciesId, String assemblyId,
           String chromosomeId, MapList map);
+
+
+  /**
+   * Returns the sequence string constructed from the substrings of a sequence
+   * defined by the int[] ranges provided by an iterator. E.g. the iterator
+   * could iterate over all visible regions of the alignment
+   * 
+   * @param it
+   *          the iterator to use
+   * @return a String corresponding to the sequence
+   */
+  public String getSequenceStringFromIterator(Iterator<int[]> it);
+
+  /**
+   * Locate the first position in this sequence which is not contained in an
+   * iterator region. If no such position exists, return 0
+   * 
+   * @param it
+   *          iterator over regions
+   * @return first residue not contained in regions
+   */
+  public int firstResidueOutsideIterator(Iterator<int[]> it);
 }