X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fdatamodel%2Ffeatures%2FFeatureStore.java;h=7651016f6283a390fe21934ae627a4ef750f7b93;hb=af259f508805faf2da90585ee9a67cd7853bf5aa;hp=ddee921fa0d46f88572d0ec0fa5bddee77c3f3aa;hpb=523b07fc514fc3936fac484dd6cfce6563ef0e83;p=jalview.git

diff --git a/src/jalview/datamodel/features/FeatureStore.java b/src/jalview/datamodel/features/FeatureStore.java
index ddee921..7651016 100644
--- a/src/jalview/datamodel/features/FeatureStore.java
+++ b/src/jalview/datamodel/features/FeatureStore.java
@@ -1,14 +1,37 @@
+/*
+ * Jalview - A Sequence Alignment Editor and Viewer ($$Version-Rel$$)
+ * Copyright (C) $$Year-Rel$$ The Jalview Authors
+ * 
+ * This file is part of Jalview.
+ * 
+ * Jalview is free software: you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License 
+ * as published by the Free Software Foundation, either version 3
+ * of the License, or (at your option) any later version.
+ *  
+ * Jalview is distributed in the hope that it will be useful, but 
+ * WITHOUT ANY WARRANTY; without even the implied warranty 
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR 
+ * PURPOSE.  See the GNU General Public License for more details.
+ * 
+ * You should have received a copy of the GNU General Public License
+ * along with Jalview.  If not, see <http://www.gnu.org/licenses/>.
+ * The Jalview Authors are detailed in the 'AUTHORS' file.
+ */
 package jalview.datamodel.features;
 
-import jalview.datamodel.SequenceFeature;
-
 import java.util.ArrayList;
 import java.util.Collections;
-import java.util.Comparator;
 import java.util.HashSet;
 import java.util.List;
 import java.util.Set;
 
+import intervalstore.api.IntervalStoreI;
+import intervalstore.impl.BinarySearcher;
+import intervalstore.impl.BinarySearcher.Compare;
+import intervalstore.impl.IntervalStore;
+import jalview.datamodel.SequenceFeature;
+
 /**
  * A data store for a set of sequence features that supports efficient lookup of
  * features overlapping a given range. Intended for (but not limited to) storage
@@ -19,63 +42,6 @@ import java.util.Set;
  */
 public class FeatureStore
 {
-  /**
-   * a class providing criteria for performing a binary search of a list
-   */
-  abstract static class SearchCriterion
-  {
-    /**
-     * Answers true if the entry passes the search criterion test
-     * 
-     * @param entry
-     * @return
-     */
-    abstract boolean compare(SequenceFeature entry);
-
-    static SearchCriterion byStart(final long target)
-    {
-      return new SearchCriterion() {
-
-        @Override
-        boolean compare(SequenceFeature entry)
-        {
-          return entry.getBegin() >= target;
-        }
-      };
-    }
-
-    static SearchCriterion byEnd(final long target)
-    {
-      return new SearchCriterion()
-      {
-
-        @Override
-        boolean compare(SequenceFeature entry)
-        {
-          return entry.getEnd() >= target;
-        }
-      };
-    }
-
-    static SearchCriterion byFeature(final ContiguousI to,
-            final Comparator<ContiguousI> rc)
-    {
-      return new SearchCriterion()
-      {
-
-        @Override
-        boolean compare(SequenceFeature entry)
-        {
-          return rc.compare(entry, to) >= 0;
-        }
-      };
-    }
-  }
-
-  Comparator<ContiguousI> startOrdering = new RangeComparator(true);
-
-  Comparator<ContiguousI> endOrdering = new RangeComparator(false);
-
   /*
    * Non-positional features have no (zero) start/end position.
    * Kept as a separate list in case this criterion changes in future.
@@ -83,14 +49,6 @@ public class FeatureStore
   List<SequenceFeature> nonPositionalFeatures;
 
   /*
-   * An ordered list of features, with the promise that no feature in the list 
-   * properly contains any other. This constraint allows bounded linear search
-   * of the list for features overlapping a region.
-   * Contact features are not included in this list.
-   */
-  List<SequenceFeature> nonNestedFeatures;
-
-  /*
    * contact features ordered by first contact position
    */
   List<SequenceFeature> contactFeatureStarts;
@@ -101,13 +59,10 @@ public class FeatureStore
   List<SequenceFeature> contactFeatureEnds;
 
   /*
-   * Nested Containment List is used to hold any features that are nested 
-   * within (properly contained by) any other feature. This is a recursive tree
-   * which supports depth-first scan for features overlapping a range.
-   * It is used here as a 'catch-all' fallback for features that cannot be put
-   * into a simple ordered list without invalidating the search methods.
+   * IntervalStore holds remaining features and provides efficient
+   * query for features overlapping any given interval
    */
-  NCList<SequenceFeature> nestedFeatures;
+  IntervalStoreI<SequenceFeature> features;
 
   /*
    * Feature groups represented in stored positional features 
@@ -121,16 +76,34 @@ public class FeatureStore
    */
   Set<String> nonPositionalFeatureGroups;
 
+  /*
+   * the total length of all positional features; contact features count 1 to
+   * the total and 1 to size(), consistent with an average 'feature length' of 1
+   */
+  int totalExtent;
+
+  float positionalMinScore;
+
+  float positionalMaxScore;
+
+  float nonPositionalMinScore;
+
+  float nonPositionalMaxScore;
+
   /**
    * Constructor
    */
   public FeatureStore()
   {
-    nonNestedFeatures = new ArrayList<SequenceFeature>();
-    positionalFeatureGroups = new HashSet<String>();
-
-    // we only construct nonPositionalFeatures, contactFeatures
-    // or the NCList if we need to
+    features = new IntervalStore<>();
+    positionalFeatureGroups = new HashSet<>();
+    nonPositionalFeatureGroups = new HashSet<>();
+    positionalMinScore = Float.NaN;
+    positionalMaxScore = Float.NaN;
+    nonPositionalMinScore = Float.NaN;
+    nonPositionalMaxScore = Float.NaN;
+
+    // we only construct nonPositionalFeatures, contactFeatures if we need to
   }
 
   /**
@@ -143,6 +116,11 @@ public class FeatureStore
    */
   public boolean addFeature(SequenceFeature feature)
   {
+    if (contains(feature))
+    {
+      return false;
+    }
+
     /*
      * keep a record of feature groups
      */
@@ -151,157 +129,134 @@ public class FeatureStore
       positionalFeatureGroups.add(feature.getFeatureGroup());
     }
 
-    boolean added = false;
-
     if (feature.isContactFeature())
     {
-      added = addContactFeature(feature);
+      addContactFeature(feature);
     }
     else if (feature.isNonPositional())
     {
-      added = addNonPositionalFeature(feature);
+      addNonPositionalFeature(feature);
     }
     else
     {
-      if (!nonNestedFeatures.contains(feature))
+      addNestedFeature(feature);
+    }
+
+    /*
+     * record the total extent of positional features, to make
+     * getTotalFeatureLength possible; we count the length of a 
+     * contact feature as 1
+     */
+    totalExtent += getFeatureLength(feature);
+
+    /*
+     * record the minimum and maximum score for positional
+     * and non-positional features
+     */
+    float score = feature.getScore();
+    if (!Float.isNaN(score))
+    {
+      if (feature.isNonPositional())
       {
-        added = addNonNestedFeature(feature);
-        if (!added)
-        {
-          /*
-           * detected a nested feature - put it in the NCList structure
-           */
-          added = addNestedFeature(feature);
-        }
+        nonPositionalMinScore = min(nonPositionalMinScore, score);
+        nonPositionalMaxScore = max(nonPositionalMaxScore, score);
+      }
+      else
+      {
+        positionalMinScore = min(positionalMinScore, score);
+        positionalMaxScore = max(positionalMaxScore, score);
       }
     }
 
-    return added;
+    return true;
   }
 
   /**
-   * Adds the feature to the list of non-positional features (with lazy
-   * instantiation of the list if it is null), and returns true. If the
-   * non-positional features already include the new feature (by equality test),
-   * then it is not added, and this method returns false. The feature group is
-   * added to the set of distinct feature groups for non-positional features.
+   * Answers true if this store contains the given feature (testing by
+   * SequenceFeature.equals), else false
    * 
    * @param feature
+   * @return
    */
-  protected boolean addNonPositionalFeature(SequenceFeature feature)
+  public boolean contains(SequenceFeature feature)
   {
-    if (nonPositionalFeatures == null)
+    if (feature.isNonPositional())
     {
-      nonPositionalFeatures = new ArrayList<SequenceFeature>();
-      nonPositionalFeatureGroups = new HashSet<String>();
+      return nonPositionalFeatures == null ? false : nonPositionalFeatures
+              .contains(feature);
     }
-    if (nonPositionalFeatures.contains(feature))
+
+    if (feature.isContactFeature())
     {
-      return false;
+      return contactFeatureStarts == null ? false : listContains(
+              contactFeatureStarts, feature);
     }
 
-    nonPositionalFeatures.add(feature);
-
-    nonPositionalFeatureGroups.add(feature.getFeatureGroup());
-
-    return true;
+    return features == null ? false : features
+            .contains(feature);
   }
 
   /**
-   * Adds one feature to the NCList that can manage nested features (creating
-   * the NCList if necessary), and returns true. If the feature is already
-   * stored in the NCList (by equality test), then it is not added, and this
-   * method returns false.
+   * Answers the 'length' of the feature, counting 0 for non-positional features
+   * and 1 for contact features
+   * 
+   * @param feature
+   * @return
    */
-  protected synchronized boolean addNestedFeature(SequenceFeature feature)
+  protected static int getFeatureLength(SequenceFeature feature)
   {
-    if (nestedFeatures == null)
+    if (feature.isNonPositional())
     {
-      nestedFeatures = new NCList<SequenceFeature>(feature);
-      return true;
+      return 0;
     }
-    return nestedFeatures.add(feature, false);
+    if (feature.isContactFeature())
+    {
+      return 1;
+    }
+    return 1 + feature.getEnd() - feature.getBegin();
   }
 
   /**
-   * Add a feature to the list of non-nested features, maintaining the ordering
-   * of the list. A check is made for whether the feature is nested in (properly
-   * contained by) an existing feature. If there is no nesting, the feature is
-   * added to the list and the method returns true. If nesting is found, the
-   * feature is not added and the method returns false.
+   * Adds the feature to the list of non-positional features (with lazy
+   * instantiation of the list if it is null), and returns true. The feature
+   * group is added to the set of distinct feature groups for non-positional
+   * features. This method allows duplicate features, so test before calling to
+   * prevent this.
    * 
    * @param feature
-   * @return
    */
-  protected boolean addNonNestedFeature(SequenceFeature feature)
+  protected boolean addNonPositionalFeature(SequenceFeature feature)
   {
-    synchronized (nonNestedFeatures)
+    if (nonPositionalFeatures == null)
     {
-      /*
-       * find the first stored feature which doesn't precede the new one
-       */
-      int insertPosition = binarySearch(nonNestedFeatures,
-              SearchCriterion.byFeature(feature, startOrdering));
+      nonPositionalFeatures = new ArrayList<>();
+    }
 
-      /*
-       * fail if we detect feature enclosure - of the new feature by
-       * the one preceding it, or of the next feature by the new one
-       */
-      if (insertPosition > 0)
-      {
-        if (encloses(nonNestedFeatures.get(insertPosition - 1), feature))
-        {
-          return false;
-        }
-      }
-      if (insertPosition < nonNestedFeatures.size())
-      {
-        // FIXME may have to test more than one feature here
-        // e.g. add [40-60] to [20-42, 30-50, 45-55]
-        if (encloses(feature, nonNestedFeatures.get(insertPosition)))
-        {
-          return false;
-        }
-      }
+    nonPositionalFeatures.add(feature);
 
-      /*
-       * checks passed - add the feature
-       */
-      nonNestedFeatures.add(insertPosition, feature);
+    nonPositionalFeatureGroups.add(feature.getFeatureGroup());
 
-      return true;
-    }
+    return true;
   }
 
   /**
-   * Answers true if range1 properly encloses range2, else false
-   * 
-   * @param range1
-   * @param range2
-   * @return
+   * Adds one feature to the IntervalStore that can manage nested features
+   * (creating the IntervalStore if necessary)
    */
-  protected boolean encloses(ContiguousI range1, ContiguousI range2)
+  protected synchronized void addNestedFeature(SequenceFeature feature)
   {
-    int begin1 = range1.getBegin();
-    int begin2 = range2.getBegin();
-    int end1 = range1.getEnd();
-    int end2 = range2.getEnd();
-    if (begin1 == begin2 && end1 > end2)
+    if (features == null)
     {
-      return true;
+      features = new IntervalStore<>();
     }
-    if (begin1 < begin2 && end1 >= end2)
-    {
-      return true;
-    }
-    return false;
+    features.add(feature);
   }
 
   /**
    * Add a contact feature to the lists that hold them ordered by start (first
    * contact) and by end (second contact) position, ensuring the lists remain
-   * ordered, and returns true. If the contact feature lists already contain the
-   * given feature (by test for equality), does not add it and returns false.
+   * ordered, and returns true. This method allows duplicate features to be
+   * added, so test before calling to avoid this.
    * 
    * @param feature
    * @return
@@ -310,25 +265,72 @@ public class FeatureStore
   {
     if (contactFeatureStarts == null)
     {
-      contactFeatureStarts = new ArrayList<SequenceFeature>();
+      contactFeatureStarts = new ArrayList<>();
     }
     if (contactFeatureEnds == null)
     {
-      contactFeatureEnds = new ArrayList<SequenceFeature>();
+      contactFeatureEnds = new ArrayList<>();
     }
 
-    // TODO binary search for insertion points!
-    if (contactFeatureStarts.contains(feature))
+    /*
+     * insert into list sorted by start (first contact position):
+     * binary search the sorted list to find the insertion point
+     */
+    int insertPosition = BinarySearcher.findFirst(contactFeatureStarts,
+            true, Compare.GE, feature.getBegin());
+    contactFeatureStarts.add(insertPosition, feature);
+
+
+    /*
+     * insert into list sorted by end (second contact position):
+     * binary search the sorted list to find the insertion point
+     */
+    insertPosition = BinarySearcher.findFirst(contactFeatureEnds,
+            false, Compare.GE, feature.getEnd());
+    contactFeatureEnds.add(insertPosition, feature);
+
+    return true;
+  }
+
+  /**
+   * Answers true if the list contains the feature, else false. This method is
+   * optimised for the condition that the list is sorted on feature start
+   * position ascending, and will give unreliable results if this does not hold.
+   * 
+   * @param features
+   * @param feature
+   * @return
+   */
+  protected static boolean listContains(List<SequenceFeature> features,
+          SequenceFeature feature)
+  {
+    if (features == null || feature == null)
     {
       return false;
     }
 
-    contactFeatureStarts.add(feature);
-    Collections.sort(contactFeatureStarts, startOrdering);
-    contactFeatureEnds.add(feature);
-    Collections.sort(contactFeatureEnds, endOrdering);
-
-    return true;
+    /*
+     * locate the first entry in the list which does not precede the feature
+     */
+    // int pos = binarySearch(features,
+    // SearchCriterion.byFeature(feature, RangeComparator.BY_START_POSITION));
+    int pos = BinarySearcher.findFirst(features, true, Compare.GE,
+            feature.getBegin());
+    int len = features.size();
+    while (pos < len)
+    {
+      SequenceFeature sf = features.get(pos);
+      if (sf.getBegin() > feature.getBegin())
+      {
+        return false; // no match found
+      }
+      if (sf.equals(feature))
+      {
+        return true;
+      }
+      pos++;
+    }
+    return false;
   }
 
   /**
@@ -344,15 +346,13 @@ public class FeatureStore
    */
   public List<SequenceFeature> findOverlappingFeatures(long start, long end)
   {
-    List<SequenceFeature> result = new ArrayList<SequenceFeature>();
-
-    findNonNestedFeatures(start, end, result);
+    List<SequenceFeature> result = new ArrayList<>();
 
     findContactFeatures(start, end, result);
 
-    if (nestedFeatures != null)
+    if (features != null)
     {
-      result.addAll(nestedFeatures.findOverlaps(start, end));
+      result.addAll(features.findOverlaps(start, end));
     }
 
     return result;
@@ -371,11 +371,11 @@ public class FeatureStore
   {
     if (contactFeatureStarts != null)
     {
-      findContactStartFeatures(from, to, result);
+      findContactStartOverlaps(from, to, result);
     }
     if (contactFeatureEnds != null)
     {
-      findContactEndFeatures(from, to, result);
+      findContactEndOverlaps(from, to, result);
     }
   }
 
@@ -388,22 +388,24 @@ public class FeatureStore
    * @param to
    * @param result
    */
-  protected void findContactEndFeatures(long from, long to,
+  protected void findContactEndOverlaps(long from, long to,
           List<SequenceFeature> result)
   {
     /*
-     * find the first contact feature (if any) that does not lie 
-     * entirely before the target range
+     * find the first contact feature (if any) 
+     * whose end point is not before the target range
      */
-    int startPosition = binarySearch(contactFeatureEnds,
-            SearchCriterion.byEnd(from));
-    for (; startPosition < contactFeatureEnds.size(); startPosition++)
+    int index = BinarySearcher.findFirst(contactFeatureEnds,
+            false, Compare.GE, (int) from);
+
+    while (index < contactFeatureEnds.size())
     {
-      SequenceFeature sf = contactFeatureEnds.get(startPosition);
+      SequenceFeature sf = contactFeatureEnds.get(index);
       if (!sf.isContactFeature())
       {
         System.err.println("Error! non-contact feature type "
                 + sf.getType() + " in contact features list");
+        index++;
         continue;
       }
 
@@ -414,71 +416,25 @@ public class FeatureStore
          * this feature's first contact position lies in the search range
          * so we don't include it in results a second time
          */
+        index++;
         continue;
       }
 
-      int end = sf.getEnd();
-      if (end >= from && end <= to)
-      {
-        result.add(sf);
-      }
-      if (end > to)
+      if (sf.getEnd() > to)
       {
+        /*
+         * this feature (and all following) has end point after the target range
+         */
         break;
       }
-    }
-  }
 
-  /**
-   * Adds non-nested features to the result list that lie within the target
-   * range. Non-positional features (start=end=0), contact features and nested
-   * features are excluded.
-   * 
-   * @param from
-   * @param to
-   * @param result
-   */
-  protected void findNonNestedFeatures(long from, long to,
-          List<SequenceFeature> result)
-  {
-    int startIndex = binarySearch(nonNestedFeatures,
-            SearchCriterion.byEnd(from));
-
-    findNonNestedFeatures(startIndex, from, to, result);
-  }
-
-  /**
-   * Scans the list of non-nested features, starting from startIndex, to find
-   * those that overlap the from-to range, and adds them to the result list.
-   * Returns the index of the first feature whose start position is after the
-   * target range (or the length of the whole list if none such feature exists).
-   * 
-   * @param startIndex
-   * @param from
-   * @param to
-   * @param result
-   * @return
-   */
-  protected int findNonNestedFeatures(final int startIndex, long from,
-          long to, List<SequenceFeature> result)
-  {
-    int i = startIndex;
-    while (i < nonNestedFeatures.size())
-    {
-      SequenceFeature sf = nonNestedFeatures.get(i);
-      if (sf.getBegin() > to)
-      {
-        break;
-      }
-      int start = sf.getBegin();
-      int end = sf.getEnd();
-      if (start <= to && end >= from)
-      {
-        result.add(sf);
-      }
-      i++;
+      /*
+       * feature has end >= from and end <= to
+       * i.e. contact end point lies within overlap search range
+       */
+      result.add(sf);
+      index++;
     }
-    return i;
   }
 
   /**
@@ -489,26 +445,36 @@ public class FeatureStore
    * @param to
    * @param result
    */
-  protected void findContactStartFeatures(long from, long to,
+  protected void findContactStartOverlaps(long from, long to,
           List<SequenceFeature> result)
   {
-    int startPosition = binarySearch(contactFeatureStarts,
-            SearchCriterion.byStart(from));
+    int index = BinarySearcher.findFirst(contactFeatureStarts,
+            true, Compare.GE, (int) from);
 
-    for (; startPosition < contactFeatureStarts.size(); startPosition++)
+    while (index < contactFeatureStarts.size())
     {
-      SequenceFeature sf = contactFeatureStarts.get(startPosition);
+      SequenceFeature sf = contactFeatureStarts.get(index);
       if (!sf.isContactFeature())
       {
-        System.err.println("Error! non-contact feature type "
-                + sf.getType() + " in contact features list");
+        System.err.println("Error! non-contact feature " + sf.toString()
+                + " in contact features list");
+        index++;
         continue;
       }
-      int begin = sf.getBegin();
-      if (begin >= from && begin <= to)
+      if (sf.getBegin() > to)
       {
-        result.add(sf);
+        /*
+         * this feature's start (and all following) follows the target range
+         */
+        break;
       }
+
+      /*
+       * feature has begin >= from and begin <= to
+       * i.e. contact start point lies within overlap search range
+       */
+      result.add(sf);
+      index++;
     }
   }
 
@@ -519,11 +485,7 @@ public class FeatureStore
    */
   public List<SequenceFeature> getPositionalFeatures()
   {
-    /*
-     * add non-nested features (may be all features for many cases)
-     */
-    List<SequenceFeature> result = new ArrayList<SequenceFeature>();
-    result.addAll(nonNestedFeatures);
+    List<SequenceFeature> result = new ArrayList<>();
 
     /*
      * add any contact features - from the list by start position
@@ -536,9 +498,9 @@ public class FeatureStore
     /*
      * add any nested features
      */
-    if (nestedFeatures != null)
+    if (features != null)
     {
-      result.addAll(nestedFeatures.getEntries());
+      result.addAll(features);
     }
 
     return result;
@@ -556,7 +518,7 @@ public class FeatureStore
     {
       return Collections.emptyList();
     }
-    return new ArrayList<SequenceFeature>(contactFeatureStarts);
+    return new ArrayList<>(contactFeatureStarts);
   }
 
   /**
@@ -571,7 +533,7 @@ public class FeatureStore
     {
       return Collections.emptyList();
     }
-    return new ArrayList<SequenceFeature>(nonPositionalFeatures);
+    return new ArrayList<>(nonPositionalFeatures);
   }
 
   /**
@@ -584,13 +546,10 @@ public class FeatureStore
    */
   public synchronized boolean delete(SequenceFeature sf)
   {
-    /*
-     * try the non-nested positional features first
-     */
-    boolean removed = nonNestedFeatures.remove(sf);
+    boolean removed = false;
 
     /*
-     * if not found, try contact positions (and if found, delete
+     * try contact positions (and if found, delete
      * from both lists of contact positions)
      */
     if (!removed && contactFeatureStarts != null)
@@ -616,73 +575,96 @@ public class FeatureStore
     /*
      * if not found, try nested features
      */
-    if (!removed && nestedFeatures != null)
+    if (!removed && features != null)
     {
-      removed = nestedFeatures.delete(sf);
+      removed = features.remove(sf);
     }
 
     if (removed)
     {
-      rebuildFeatureGroups(sf.getFeatureGroup(), removedNonPositional);
+      rescanAfterDelete();
     }
 
     return removed;
   }
 
   /**
-   * Check whether the given feature group is still represented, in either
-   * positional or non-positional features, and if not, remove it from the set
-   * of feature groups
+   * Rescan all features to recompute any cached values after an entry has been
+   * deleted. This is expected to be an infrequent event, so performance here is
+   * not critical.
+   */
+  protected synchronized void rescanAfterDelete()
+  {
+    positionalFeatureGroups.clear();
+    nonPositionalFeatureGroups.clear();
+    totalExtent = 0;
+    positionalMinScore = Float.NaN;
+    positionalMaxScore = Float.NaN;
+    nonPositionalMinScore = Float.NaN;
+    nonPositionalMaxScore = Float.NaN;
+
+    /*
+     * scan non-positional features for groups and scores
+     */
+    for (SequenceFeature sf : getNonPositionalFeatures())
+    {
+      nonPositionalFeatureGroups.add(sf.getFeatureGroup());
+      float score = sf.getScore();
+      nonPositionalMinScore = min(nonPositionalMinScore, score);
+      nonPositionalMaxScore = max(nonPositionalMaxScore, score);
+    }
+
+    /*
+     * scan positional features for groups, scores and extents
+     */
+    for (SequenceFeature sf : getPositionalFeatures())
+    {
+      positionalFeatureGroups.add(sf.getFeatureGroup());
+      float score = sf.getScore();
+      positionalMinScore = min(positionalMinScore, score);
+      positionalMaxScore = max(positionalMaxScore, score);
+      totalExtent += getFeatureLength(sf);
+    }
+  }
+
+  /**
+   * A helper method to return the minimum of two floats, where a non-NaN value
+   * is treated as 'less than' a NaN value (unlike Math.min which does the
+   * opposite)
    * 
-   * @param featureGroup
-   * @param nonPositional
+   * @param f1
+   * @param f2
    */
-  protected void rebuildFeatureGroups(String featureGroup,
-          boolean nonPositional)
+  protected static float min(float f1, float f2)
   {
-    if (nonPositional && nonPositionalFeatures != null)
+    if (Float.isNaN(f1))
     {
-      boolean found = false;
-      for (SequenceFeature sf : nonPositionalFeatures)
-      {
-        String group = sf.getFeatureGroup();
-        if (featureGroup == group
-                || (featureGroup != null && featureGroup.equals(group)))
-        {
-          found = true;
-          break;
-        }
-      }
-      if (!found)
-      {
-        nonPositionalFeatureGroups.remove(featureGroup);
-      }
+      return Float.isNaN(f2) ? f1 : f2;
     }
-    else if (!findFeatureGroup(featureGroup))
+    else
     {
-      positionalFeatureGroups.remove(featureGroup);
+      return Float.isNaN(f2) ? f1 : Math.min(f1, f2);
     }
   }
 
   /**
-   * Scans all positional features to check whether the given feature group is
-   * found, and returns true if found, else false
+   * A helper method to return the maximum of two floats, where a non-NaN value
+   * is treated as 'greater than' a NaN value (unlike Math.max which does the
+   * opposite)
    * 
-   * @param featureGroup
-   * @return
+   * @param f1
+   * @param f2
    */
-  protected boolean findFeatureGroup(String featureGroup)
+  protected static float max(float f1, float f2)
   {
-    for (SequenceFeature sf : getPositionalFeatures())
+    if (Float.isNaN(f1))
     {
-      String group = sf.getFeatureGroup();
-      if (group == featureGroup
-              || (group != null && group.equals(featureGroup)))
-      {
-        return true;
-      }
+      return Float.isNaN(f2) ? f1 : f2;
+    }
+    else
+    {
+      return Float.isNaN(f2) ? f1 : Math.max(f1, f2);
     }
-    return false;
   }
 
   /**
@@ -692,12 +674,12 @@ public class FeatureStore
    */
   public boolean isEmpty()
   {
-    boolean hasFeatures = !nonNestedFeatures.isEmpty()
-            || (contactFeatureStarts != null && !contactFeatureStarts
+    boolean hasFeatures = (contactFeatureStarts != null
+            && !contactFeatureStarts
                     .isEmpty())
             || (nonPositionalFeatures != null && !nonPositionalFeatures
                     .isEmpty())
-            || (nestedFeatures != null && nestedFeatures.size() > 0);
+            || (features != null && features.size() > 0);
 
     return !hasFeatures;
   }
@@ -725,37 +707,148 @@ public class FeatureStore
   }
 
   /**
-   * Performs a binary search of the (sorted) list to find the index of the
-   * first entry which returns true for the given comparator function. Returns
-   * the length of the list if there is no such entry.
+   * Answers the number of positional (or non-positional) features stored.
+   * Contact features count as 1.
    * 
-   * @param features
-   * @param sc
+   * @param positional
    * @return
    */
-  protected int binarySearch(List<SequenceFeature> features,
-          SearchCriterion sc)
+  public int getFeatureCount(boolean positional)
   {
-    int start = 0;
-    int end = features.size() - 1;
-    int matched = features.size();
+    if (!positional)
+    {
+      return nonPositionalFeatures == null ? 0 : nonPositionalFeatures
+              .size();
+    }
 
-    while (start <= end)
+    int size = 0;
+
+    if (contactFeatureStarts != null)
+    {
+      // note a contact feature (start/end) counts as one
+      size += contactFeatureStarts.size();
+    }
+
+    if (features != null)
     {
-      int mid = (start + end) / 2;
-      SequenceFeature entry = features.get(mid);
-      boolean compare = sc.compare(entry);
-      if (compare)
+      size += features.size();
+    }
+
+    return size;
+  }
+
+  /**
+   * Answers the total length of positional features (or zero if there are
+   * none). Contact features contribute a value of 1 to the total.
+   * 
+   * @return
+   */
+  public int getTotalFeatureLength()
+  {
+    return totalExtent;
+  }
+
+  /**
+   * Answers the minimum score held for positional or non-positional features.
+   * This may be Float.NaN if there are no features, are none has a non-NaN
+   * score.
+   * 
+   * @param positional
+   * @return
+   */
+  public float getMinimumScore(boolean positional)
+  {
+    return positional ? positionalMinScore : nonPositionalMinScore;
+  }
+
+  /**
+   * Answers the maximum score held for positional or non-positional features.
+   * This may be Float.NaN if there are no features, are none has a non-NaN
+   * score.
+   * 
+   * @param positional
+   * @return
+   */
+  public float getMaximumScore(boolean positional)
+  {
+    return positional ? positionalMaxScore : nonPositionalMaxScore;
+  }
+
+  /**
+   * Answers a list of all either positional or non-positional features whose
+   * feature group matches the given group (which may be null)
+   * 
+   * @param positional
+   * @param group
+   * @return
+   */
+  public List<SequenceFeature> getFeaturesForGroup(boolean positional,
+          String group)
+  {
+    List<SequenceFeature> result = new ArrayList<>();
+
+    /*
+     * if we know features don't include the target group, no need
+     * to inspect them for matches
+     */
+    if (positional && !positionalFeatureGroups.contains(group)
+            || !positional && !nonPositionalFeatureGroups.contains(group))
+    {
+      return result;
+    }
+
+    List<SequenceFeature> sfs = positional ? getPositionalFeatures()
+            : getNonPositionalFeatures();
+    for (SequenceFeature sf : sfs)
+    {
+      String featureGroup = sf.getFeatureGroup();
+      if (group == null && featureGroup == null || group != null
+              && group.equals(featureGroup))
       {
-        matched = mid;
-        end = mid - 1;
+        result.add(sf);
       }
-      else
+    }
+    return result;
+  }
+
+  /**
+   * Adds the shift amount to the start and end of all positional features whose
+   * start position is at or after fromPosition. Returns true if at least one
+   * feature was shifted, else false.
+   * 
+   * @param fromPosition
+   * @param shiftBy
+   * @return
+   */
+  public synchronized boolean shiftFeatures(int fromPosition, int shiftBy)
+  {
+    /*
+     * Because begin and end are final fields (to ensure the data store's
+     * integrity), we have to delete each feature and re-add it as amended.
+     * (Although a simple shift of all values would preserve data integrity!)
+     */
+    boolean modified = false;
+    for (SequenceFeature sf : getPositionalFeatures())
+    {
+      if (sf.getBegin() >= fromPosition)
       {
-        start = mid + 1;
+        modified = true;
+        int newBegin = sf.getBegin() + shiftBy;
+        int newEnd = sf.getEnd() + shiftBy;
+
+        /*
+         * sanity check: don't shift left of the first residue
+         */
+        if (newEnd > 0)
+        {
+          newBegin = Math.max(1, newBegin);
+          SequenceFeature sf2 = new SequenceFeature(sf, newBegin, newEnd,
+                  sf.getFeatureGroup(), sf.getScore());
+          addFeature(sf2);
+        }
+        delete(sf);
       }
     }
-
-    return matched;
+    return modified;
   }
 }