From b2824493add6517beee16bf7dc9066642c9bd26e Mon Sep 17 00:00:00 2001
From: jprocter <jprocter@compbio.dundee.ac.uk>
Date: Thu, 8 Dec 2011 17:28:54 +0000
Subject: [PATCH] (JAL-1013) generalise binary sequence encoding scheme for
 nucleotide and protein sequences with arbitrary
 substitution matrices

---
 src/jalview/datamodel/BinarySequence.java |   88 ++++++++++++++++++-----------
 1 file changed, 56 insertions(+), 32 deletions(-)

diff --git a/src/jalview/datamodel/BinarySequence.java b/src/jalview/datamodel/BinarySequence.java
index 9d6f5e8..2b4ce6b 100755
--- a/src/jalview/datamodel/BinarySequence.java
+++ b/src/jalview/datamodel/BinarySequence.java
@@ -20,58 +20,83 @@ package jalview.datamodel;
 import jalview.schemes.*;
 
 /**
- * DOCUMENT ME!
+ * Encode a sequence as a numeric vector using either classic residue binary
+ * encoding or convolved with residue substitution matrix.
  * 
  * @author $author$
  * @version $Revision$
  */
 public class BinarySequence extends Sequence
 {
+  public class InvalidSequenceTypeException extends Exception
+  {
+
+    public InvalidSequenceTypeException(String string)
+    {
+      super(string);
+    }
+
+  }
+
   int[] binary;
 
   double[] dbinary;
 
+  boolean isNa=false;
   /**
    * Creates a new BinarySequence object.
    * 
    * @param s
    *          DOCUMENT ME!
    */
-  public BinarySequence(String s)
+  public BinarySequence(String s, boolean isNa)
   {
     super("", s, 0, s.length());
+    this.isNa=isNa;
   }
 
   /**
-   * DOCUMENT ME!
+   * clear the dbinary matrix 
+   * @return nores - dimension of sequence symbol encoding for this sequence
    */
-  public void encode()
+  private int initMatrixGetNoRes()
   {
+    int nores=(isNa) ? ResidueProperties.maxNucleotideIndex : ResidueProperties.maxProteinIndex;
     // Set all matrix to 0
-    dbinary = new double[getSequence().length * 21];
-
-    int nores = 21;
+    dbinary = new double[getSequence().length * nores];
 
     for (int i = 0; i < dbinary.length; i++)
     {
       dbinary[i] = 0.0;
     }
-
+    return nores;
+  }
+  private int[] getSymbolmatrix()
+  {
+    return (isNa) ? ResidueProperties.nucleotideIndex : ResidueProperties.aaIndex;
+  }
+  /**
+   * DOCUMENT ME!
+   */
+  public void encode()
+  {
+    int nores=initMatrixGetNoRes();
+    final int[] sindex=getSymbolmatrix();
     for (int i = 0; i < getSequence().length; i++)
     {
-      int aanum = 20;
+      int aanum = nores-1;
 
       try
       {
-        aanum = ResidueProperties.aaIndex[getCharAt(i)];
+        aanum = sindex[getCharAt(i)];
       } catch (NullPointerException e)
       {
-        aanum = 20;
+        aanum = nores-1;
       }
 
-      if (aanum > 20)
+      if (aanum >= nores)
       {
-        aanum = 20;
+        aanum = nores-1;
       }
 
       dbinary[(i * nores) + aanum] = 1.0;
@@ -83,50 +108,49 @@ public class BinarySequence extends Sequence
    * 
    * @param matrix
    */
-  public void matrixEncode(ScoreMatrix matrix)
+  public void matrixEncode(final ScoreMatrix matrix) throws InvalidSequenceTypeException
   {
+    if (isNa!=matrix.isDNA())
+    {
+      throw new InvalidSequenceTypeException("matrix "
+              + matrix.getClass().getCanonicalName()
+              + " is not a valid matrix for "
+              + (isNa ? "nucleotide" : "protein") + "sequences");
+    }
     matrixEncode(matrix.isDNA() ? ResidueProperties.nucleotideIndex
             : ResidueProperties.aaIndex, matrix.getMatrix());
   }
 
-  /**
-   * DOCUMENT ME!
-   */
-  public void blosumEncode()
-  {
-    matrixEncode(ResidueProperties.aaIndex, ResidueProperties.getBLOSUM62());
-  }
-
-  private void matrixEncode(int[] aaIndex, int[][] matrix)
+  private void matrixEncode(final int[] aaIndex, final int[][] matrix)
   {
     // Set all matrix to 0
-    dbinary = new double[getSequence().length * 21];
+    // dbinary = new double[getSequence().length * 21];
 
-    int nores = 21;
+    int nores = initMatrixGetNoRes();
 
     // for (int i = 0; i < dbinary.length; i++) {
     // dbinary[i] = 0.0;
     // }
-    for (int i = 0; i < getSequence().length; i++)
+    for (int i = 0,iSize=getSequence().length; i<iSize; i++)
     {
-      int aanum = 20;
+      int aanum = nores-1;
 
       try
       {
         aanum = aaIndex[getCharAt(i)];
       } catch (NullPointerException e)
       {
-        aanum = 20;
+        aanum = nores-1;
       }
 
-      if (aanum > 20)
+      if (aanum >=nores)
       {
-        aanum = 20;
+        aanum = nores-1;
       }
 
-      // Do the blosum thing
+      // Do the blosum^H^H^H^H^H score matrix summation thing
 
-      for (int j = 0; j < 20; j++)
+      for (int j = 0; j < nores; j++)
       {
         dbinary[(i * nores) + j] = matrix[aanum][j];
       }
-- 
1.7.10.2