*/
package jalview.analysis.scoremodels;
-import jalview.api.analysis.DistanceModelI;
-import jalview.api.analysis.ViewBasedAnalysisI;
+import jalview.api.AlignmentViewPanel;
+import jalview.api.FeatureRenderer;
+import jalview.api.analysis.ScoreModelI;
+import jalview.api.analysis.SimilarityParamsI;
import jalview.datamodel.AlignmentView;
import jalview.datamodel.SeqCigar;
import jalview.datamodel.SequenceFeature;
+import jalview.math.Matrix;
+import jalview.math.MatrixI;
import jalview.util.SetUtils;
import java.util.HashMap;
import java.util.Map;
import java.util.Set;
-public class FeatureDistanceModel implements DistanceModelI, ViewBasedAnalysisI
+public class FeatureDistanceModel extends DistanceScoreModel
{
- jalview.api.FeatureRenderer fr;
+ private static final String NAME = "Sequence Feature Similarity";
+
+ private String description;
+
+ FeatureRenderer fr;
+
+ /**
+ * Constructor
+ */
+ public FeatureDistanceModel()
+ {
+ }
@Override
- public boolean configureFromAlignmentView(
- jalview.api.AlignmentViewPanel view)
+ public ScoreModelI getInstance(AlignmentViewPanel view)
+ {
+ FeatureDistanceModel instance;
+ try
+ {
+ instance = this.getClass().getDeclaredConstructor().newInstance();
+ instance.configureFromAlignmentView(view);
+ return instance;
+ } catch (InstantiationException | IllegalAccessException e)
+ {
+ System.err.println("Error in " + getClass().getName()
+ + ".getInstance(): " + e.getMessage());
+ return null;
+ } catch (ReflectiveOperationException roe)
+ {
+ return null;
+ }
+ }
+
+ boolean configureFromAlignmentView(AlignmentViewPanel view)
+
{
fr = view.cloneFeatureRenderer();
return true;
* 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 float[][] findDistances(AlignmentView seqData)
+ public MatrixI findDistances(AlignmentView seqData,
+ SimilarityParamsI params)
{
- List<String> dft = fr.getDisplayedFeatureTypes();
SeqCigar[] seqs = seqData.getSequences();
int noseqs = seqs.length;
int cpwidth = 0;// = seqData.getWidth();
- float[][] distance = new float[noseqs][noseqs];
- if (dft.isEmpty())
+ double[][] distances = new double[noseqs][noseqs];
+ List<String> dft = null;
+ if (fr != null)
{
- return distance;
+ dft = fr.getDisplayedFeatureTypes();
+ }
+ if (dft == null || dft.isEmpty())
+ {
+ return new Matrix(distances);
}
// need to get real position for view position
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);
+ Map<SeqCigar, Set<String>> sfap = findFeatureTypesAtColumn(seqs,
+ cpos);
/*
* count feature types on either i'th or j'th sequence but not both
{
for (int j = i + 1; j < noseqs; j++)
{
- int seqDistance = SetUtils.countDisjunction(sfap.get(seqs[i]),
- sfap.get(seqs[j]));
- distance[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;
+ }
}
}
}
/*
* normalise the distance scores (summed over columns) by the
* number of visible columns used in the calculation
+ * and fill in the bottom half of the matrix
*/
+ // TODO JAL-2424 cpwidth may be out by 1 - affects scores but not tree shape
for (int i = 0; i < noseqs; i++)
{
for (int j = i + 1; j < noseqs; j++)
{
- distance[i][j] /= cpwidth;
- distance[j][i] = distance[i][j];
+ distances[i][j] /= cpwidth;
+ distances[j][i] = distances[i][j];
}
}
- return distance;
+ return new Matrix(distances);
}
/**
- * 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
+ * does not include entries for features which straddle a gapped column
+ * positions.
*
* @param seqs
* @param columnPosition
+ * (0..)
* @return
*/
protected Map<SeqCigar, Set<String>> findFeatureTypesAtColumn(
SeqCigar[] seqs, int columnPosition)
{
- Map<SeqCigar, Set<String>> sfap = new HashMap<SeqCigar, Set<String>>();
+ Map<SeqCigar, Set<String>> sfap = new HashMap<>();
for (SeqCigar seq : seqs)
{
- Set<String> types = new HashSet<String>();
int spos = seq.findPosition(columnPosition);
if (spos != -1)
{
- List<SequenceFeature> sfs = fr.findFeaturesAtRes(seq.getRefSeq(),
- spos);
+ /*
+ * position is not a gap
+ */
+ Set<String> types = new HashSet<>();
+ List<SequenceFeature> sfs = fr.findFeaturesAtResidue(
+ seq.getRefSeq(), spos, spos);
for (SequenceFeature sf : sfs)
{
types.add(sf.getType());
}
+ sfap.put(seq, types);
}
- sfap.put(seq, types);
}
return sfap;
}
@Override
public String getName()
{
- return "Sequence Feature Similarity";
+ return NAME;
+ }
+
+ @Override
+ public String getDescription()
+ {
+ return description;
}
@Override
git://source.jalview.org / jalview.git / blobdiff