X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fanalysis%2Fscoremodels%2FFeatureDistanceModel.java;fp=src%2Fjalview%2Fanalysis%2Fscoremodels%2FFeatureDistanceModel.java;h=b44b4cfff7042606a5531e342d9948a2b184b757;hb=693e17575680c94d9bdc5797faa0b328bb6f7c05;hp=636c19b3b545760a4d6018dddd7f43da7ee51e25;hpb=f56a081f0c0174cf6a5b8799a5531ddbdb71f20f;p=jalview.git

diff --git a/src/jalview/analysis/scoremodels/FeatureDistanceModel.java b/src/jalview/analysis/scoremodels/FeatureDistanceModel.java
index 636c19b..b44b4cf 100644
--- a/src/jalview/analysis/scoremodels/FeatureDistanceModel.java
+++ b/src/jalview/analysis/scoremodels/FeatureDistanceModel.java
@@ -57,17 +57,28 @@ public class FeatureDistanceModel implements DistanceScoreModelI,
    * features each sequence pair has at each column, ignore feature types they
    * have in common, and count the rest. The totals are normalised by the number
    * of columns processed.
+   * <p>
+   * The parameters argument provides settings for treatment of gap-residue
+   * aligned positions, and whether the score is over the longer or shorter of
+   * each pair of sequences
+   * 
+   * @param seqData
+   * @param params
    */
   @Override
   public MatrixI findDistances(AlignmentView seqData,
-          SimilarityParamsI options)
+          SimilarityParamsI params)
   {
-    List<String> dft = fr.getDisplayedFeatureTypes();
     SeqCigar[] seqs = seqData.getSequences();
     int noseqs = seqs.length;
     int cpwidth = 0;// = seqData.getWidth();
     double[][] distances = new double[noseqs][noseqs];
-    if (dft.isEmpty())
+    List<String> dft = null;
+    if (fr != null)
+    {
+      dft = fr.getDisplayedFeatureTypes();
+    }
+    if (dft == null || dft.isEmpty())
     {
       return new Matrix(distances);
     }
@@ -86,7 +97,7 @@ public class FeatureDistanceModel implements DistanceScoreModelI,
         cpwidth++;
 
         /*
-         * first pass: record features types in column for each sequence
+         * first record feature types in this column for each sequence
          */
         Map<SeqCigar, Set<String>> sfap = findFeatureTypesAtColumn(
                 seqs, cpos);
@@ -99,9 +110,23 @@ public class FeatureDistanceModel implements DistanceScoreModelI,
         {
           for (int j = i + 1; j < noseqs; j++)
           {
-            int seqDistance = SetUtils.countDisjunction(sfap.get(seqs[i]),
-                    sfap.get(seqs[j]));
-            distances[i][j] += seqDistance;
+            SeqCigar sc1 = seqs[i];
+            SeqCigar sc2 = seqs[j];
+            Set<String> set1 = sfap.get(sc1);
+            Set<String> set2 = sfap.get(sc2);
+            boolean gap1 = set1 == null;
+            boolean gap2 = set2 == null;
+
+            /*
+             * gap-gap always scores zero
+             * residue-residue is always scored
+             * include gap-residue score if params say to do so
+             */
+            if ((!gap1 && !gap2) || params.includeGaps())
+            {
+              int seqDistance = SetUtils.countDisjunction(set1, set2);
+              distances[i][j] += seqDistance;
+            }
           }
         }
       }
@@ -125,8 +150,9 @@ public class FeatureDistanceModel implements DistanceScoreModelI,
   }
 
   /**
-   * Builds and returns a list (one per SeqCigar) of visible feature types at
-   * the given column position
+   * Builds and returns a map containing a (possibly empty) list (one per
+   * SeqCigar) of visible feature types at the given column position. The map
+   * has no entry for sequences which are gapped at the column position.
    * 
    * @param seqs
    * @param columnPosition
@@ -138,18 +164,18 @@ public class FeatureDistanceModel implements DistanceScoreModelI,
     Map<SeqCigar, Set<String>> sfap = new HashMap<SeqCigar, Set<String>>();
     for (SeqCigar seq : seqs)
     {
-      Set<String> types = new HashSet<String>();
       int spos = seq.findPosition(columnPosition);
       if (spos != -1)
       {
+        Set<String> types = new HashSet<String>();
         List<SequenceFeature> sfs = fr.findFeaturesAtRes(seq.getRefSeq(),
                 spos);
         for (SequenceFeature sf : sfs)
         {
           types.add(sf.getType());
         }
+        sfap.put(seq, types);
       }
-      sfap.put(seq, types);
     }
     return sfap;
   }