Method for percentage of identity of a sequence pair biojava#491

lafita · lafita · commit f57ede4a5dca · 2016-06-24T16:20:35.000+02:00
diff --git a/biojava-core/src/main/java/org/biojava/nbio/core/alignment/SimpleAlignedSequence.java b/biojava-core/src/main/java/org/biojava/nbio/core/alignment/SimpleAlignedSequence.java
@@ -59,6 +59,7 @@ public class SimpleAlignedSequence<S extends Sequence<C>, C extends Compound> im
 
 	// cached (lazily initialized)
 	private int numGaps = -1;
+	private int numGapPositions = -1;
 	private int[] alignmentFromSequence, sequenceFromAlignment;
 
 	/**
@@ -167,6 +168,7 @@ public Location getLocationInAlignment() {
 	public int getNumGaps() {
 		if (numGaps == -1) {
 			numGaps = 0;
+			numGapPositions = 0;
 			C cGap = getCompoundSet().getCompoundForString(gap);
 			boolean inGap = false;
 			for (C compound : getAsList()) {
@@ -175,6 +177,7 @@ public int getNumGaps() {
 						numGaps++;
 						inGap = true;
 					}
+					numGapPositions++;
 				} else {
 					inGap = false;
 				}
@@ -382,4 +385,18 @@ private void setLocation(List<Step> steps) {
 	public SequenceView<C> getInverse() {
 		throw new UnsupportedOperationException("Not supported yet.");
 	}
+
+	@Override
+	public int getNumGapPositions() {
+		if (numGapPositions == -1)
+			getNumGaps();
+		return numGapPositions;
+	}
+
+	@Override
+	public double getCoverage() {
+		
+		double coverage = getLength() - getNumGapPositions();
+		return coverage / getOriginalSequence().getLength();
+	}
 }
diff --git a/biojava-core/src/main/java/org/biojava/nbio/core/alignment/template/AlignedSequence.java b/biojava-core/src/main/java/org/biojava/nbio/core/alignment/template/AlignedSequence.java
@@ -74,12 +74,28 @@ enum Step { COMPOUND, GAP }
 	Location getLocationInAlignment();
 
 	/**
-	 * Returns number of gaps in the sequence.  This could be determined from the {@link Location} information or from
+	 * Returns number of gaps (gap openings) in the sequence. This could be determined from the {@link Location} information or from
 	 * gap {@link Compound}s, which may not necessarily result in the same number.
 	 *
 	 * @return number of gaps in the sequence
 	 */
 	int getNumGaps();
+	
+	/**
+	 * Returns number of gap positions (gap openings and extensions) in the sequence. This could be determined from the {@link Location} 
+	 * information or from gap {@link Compound}s, which may not necessarily result in the same number.
+	 *
+	 * @return number of gap positions in the sequence
+	 */
+	int getNumGapPositions();
+	
+	/**
+	 * Returns the coverage, as a fraction between 0 and 1, of this {@link AlignedSequence} with respect to the original sequence. 
+	 * This is equivalent to ({@link #getLength()} - {@link #getNumGapPositions()}) / getOriginalSequence().getLength().
+	 *
+	 * @return coverage of the original sequence by the aligned sequence
+	 */
+	double getCoverage();
 
 	/**
 	 * Returns the original {@link Sequence} before alignment.
