import jalview.datamodel.SequenceFeature;
import java.util.ArrayList;
-import java.util.Arrays;
-import java.util.BitSet;
+import java.util.Collection;
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.IntervalStore;
+import intervalstore.impl.BinarySearcher.Compare;
-
-/**
- * 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
- * of features for one sequence and feature type.
- *
- * @author gmcarstairs
- *
- */
public class FeatureStore
{
+ public enum IntervalStoreType
+ {
+ /**
+ * original NCList-based IntervalStore
+ */
+ INTERVAL_STORE_NCLIST,
+
+ /**
+ * linked-list IntervalStore
+ */
+ INTERVAL_STORE_LINKED_LIST,
+
+ /**
+ * NCList as array buffer IntervalStore
+ */
+ INTERVAL_STORE_NCARRAY
+ }
+
+ /*
+ * track largest start for quick insertion of ordered features
+ */
+ protected int maxStart = -1;
+
+ protected int maxContactStart = -1;
+
/*
* Non-positional features have no (zero) start/end position.
* Kept as a separate list in case this criterion changes in future.
float nonPositionalMaxScore;
- private ArrayList<SequenceFeature> featuresList;
+ /**
+ * Answers the 'length' of the feature, counting 0 for non-positional features
+ * and 1 for contact features
+ *
+ * @param feature
+ * @return
+ */
+ protected static int getFeatureLength(SequenceFeature feature)
+ {
+ if (feature.isNonPositional())
+ {
+ return 0;
+ }
+ if (feature.isContactFeature())
+ {
+ return 1;
+ }
+ return 1 + feature.getEnd() - feature.getBegin();
+ }
+
+ /**
+ * 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 list
+ * @param feature
+ * @return
+ */
+ public static boolean listContains(List<SequenceFeature> list,
+ SequenceFeature feature)
+ {
+ if (list == null || feature == null)
+ {
+ return false;
+ }
+
+ /*
+ * locate the first entry in the list which does not precede the feature
+ */
+ int begin = feature.begin;
+ int pos = BinarySearcher.findFirst(list, true, Compare.GE, begin);
+ int len = list.size();
+ while (pos < len)
+ {
+ SequenceFeature sf = list.get(pos);
+ if (sf.begin > begin)
+ {
+ return false; // no match found
+ }
+ if (sf.equals(feature))
+ {
+ return true;
+ }
+ pos++;
+ }
+ return 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 f1
+ * @param f2
+ */
+ protected static float max(float f1, float f2)
+ {
+ if (Float.isNaN(f1))
+ {
+ return Float.isNaN(f2) ? f1 : f2;
+ }
+ else
+ {
+ return Float.isNaN(f2) ? f1 : Math.max(f1, f2);
+ }
+ }
+
+ /**
+ * 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 f1
+ * @param f2
+ */
+ protected static float min(float f1, float f2)
+ {
+ if (Float.isNaN(f1))
+ {
+ return Float.isNaN(f2) ? f1 : f2;
+ }
+ else
+ {
+ return Float.isNaN(f2) ? f1 : Math.min(f1, f2);
+ }
+ }
/**
- * Constructor
+ * Constructor that defaults to using NCList IntervalStore
*/
public FeatureStore()
{
- features = new IntervalStore<>();
- featuresList = new ArrayList<>();
+ this(IntervalStoreType.INTERVAL_STORE_NCLIST);
+ }
+
+ /**
+ * Constructor that allows an alternative IntervalStore implementation to be
+ * chosen
+ */
+ public FeatureStore(IntervalStoreType intervalStoreType)
+ {
+ features = getIntervalStore(intervalStoreType);
positionalFeatureGroups = new HashSet<>();
nonPositionalFeatureGroups = new HashSet<>();
positionalMinScore = Float.NaN;
nonPositionalMinScore = Float.NaN;
nonPositionalMaxScore = Float.NaN;
- // we only construct nonPositionalFeatures, contactFeatures if we need to
+ // only construct nonPositionalFeatures or contactFeatures if needed
}
/**
- * Adds one sequence feature to the store, and returns true, unless the
- * feature is already contained in the store, in which case this method
- * returns false. Containment is determined by SequenceFeature.equals()
- * comparison.
+ * Returns a new instance of IntervalStoreI of implementation as selected by
+ * the type parameter
*
- * @param feature
+ * @param type
+ * @return
*/
+ private IntervalStoreI<SequenceFeature> getIntervalStore(
+ IntervalStoreType type)
+ {
+ switch (type)
+ {
+ default:
+ case INTERVAL_STORE_NCLIST:
+ return new intervalstore.impl.IntervalStore<>();
+ case INTERVAL_STORE_NCARRAY:
+ return new intervalstore.nonc.IntervalStore<>();
+ case INTERVAL_STORE_LINKED_LIST:
+ return new intervalstore.nonc.IntervalStore0<>();
+ }
+ }
- public boolean addFeature(SequenceFeature 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. This method allows duplicate features to be
+ * added, so test before calling to avoid this.
+ *
+ * @param feature
+ * @return
+ */
+ protected synchronized boolean addContactFeature(SequenceFeature feature)
{
- if (contains(feature))
+ if (contactFeatureStarts == null)
{
- return false;
+ contactFeatureStarts = new ArrayList<>();
+ contactFeatureEnds = new ArrayList<>();
}
/*
- * keep a record of feature groups
+ * insert into list sorted by start (first contact position):
+ * binary search the sorted list to find the insertion point
+ */
+ int insertAt = BinarySearcher.findFirst(contactFeatureStarts, true,
+ Compare.GE, feature.begin);
+ contactFeatureStarts.add(insertAt, feature);
+ /*
+ * insert into list sorted by end (second contact position):
+ * binary search the sorted list to find the insertion point
*/
- if (!feature.isNonPositional())
- {
- positionalFeatureGroups.add(feature.getFeatureGroup());
- }
+ contactFeatureEnds.add(findFirstEnd(contactFeatureEnds, feature.end),
+ feature);
+ return true;
+ }
+
+ /**
+ * Adds one sequence feature to the store, and returns true, unless the
+ * feature is already contained in the store, in which case this method
+ * returns false. Containment is determined by SequenceFeature.equals()
+ * comparison.
+ *
+ * @param feature
+ */
+ public boolean addFeature(SequenceFeature feature)
+ {
if (feature.isContactFeature())
{
+ if (containsContactFeature(feature))
+ {
+ return false;
+ }
+ positionalFeatureGroups.add(feature.getFeatureGroup());
+ if (feature.begin > maxContactStart)
+ {
+ maxContactStart = feature.begin;
+ }
addContactFeature(feature);
}
else if (feature.isNonPositional())
{
+ if (containsNonPositionalFeature(feature))
+ {
+ return false;
+ }
+
addNonPositionalFeature(feature);
}
else
{
- addNestedFeature(feature);
+ if (!features.add(feature, false))
+ {
+ return false;
+ }
+ positionalFeatureGroups.add(feature.getFeatureGroup());
+ if (feature.begin > maxStart)
+ {
+ maxStart = feature.begin;
+ }
}
/*
}
/**
- * Answers true if this store contains the given feature (testing by
- * SequenceFeature.equals), else false
- *
- * @param feature
- * @return
- */
- public boolean contains(SequenceFeature feature)
- {
- if (feature.isNonPositional())
- {
- return nonPositionalFeatures == null ? false
- : nonPositionalFeatures.contains(feature);
- }
-
- if (feature.isContactFeature())
- {
- return contactFeatureStarts == null ? false
- : listContains(contactFeatureStarts, feature);
- }
-
- return features == null ? false : features.contains(feature);
- }
-
- /**
- * Answers the 'length' of the feature, counting 0 for non-positional features
- * and 1 for contact features
+ * A helper method that adds to the result list any features from the
+ * collection provided whose feature group matches the specified group
*
- * @param feature
- * @return
+ * @param group
+ * @param sfs
+ * @param result
*/
- protected static int getFeatureLength(SequenceFeature feature)
+ private void addFeaturesForGroup(String group,
+ Collection<SequenceFeature> sfs, List<SequenceFeature> result)
{
- if (feature.isNonPositional())
+ if (sfs == null)
{
- return 0;
+ return;
}
- if (feature.isContactFeature())
+ for (SequenceFeature sf : sfs)
{
- return 1;
+ String featureGroup = sf.getFeatureGroup();
+ if (group == null && featureGroup == null
+ || group != null && group.equals(featureGroup))
+ {
+ result.add(sf);
+ }
}
- return 1 + feature.getEnd() - feature.getBegin();
}
/**
}
/**
- * Adds one feature to the IntervalStore that can manage nested features
- * (creating the IntervalStore if necessary)
- */
- protected synchronized void addNestedFeature(SequenceFeature feature)
- {
- if (features == null)
- {
- features = new IntervalStore<>();
- }
- features.add(feature);
- featuresList.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. This method allows duplicate features to be
- * added, so test before calling to avoid this.
+ * Answers true if this store contains the given feature (testing by
+ * SequenceFeature.equals), else false
*
* @param feature
* @return
*/
- protected synchronized boolean addContactFeature(SequenceFeature feature)
+ public boolean contains(SequenceFeature feature)
{
- if (contactFeatureStarts == null)
+ if (feature.isNonPositional())
{
- contactFeatureStarts = new ArrayList<>();
+ return containsNonPositionalFeature(feature);
}
- if (contactFeatureEnds == null)
+
+ if (feature.isContactFeature())
{
- contactFeatureEnds = new ArrayList<>();
+ return containsContactFeature(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,
- f -> f.getBegin() >= feature.getBegin());
- contactFeatureStarts.add(insertPosition, feature);
+ return containsPositionalFeature(feature);
+ }
- /*
- * insert into list sorted by end (second contact position):
- * binary search the sorted list to find the insertion point
- */
- insertPosition = BinarySearcher.findFirst(contactFeatureEnds,
- f -> f.getEnd() >= feature.getEnd());
- contactFeatureEnds.add(insertPosition, feature);
+ private boolean containsPositionalFeature(SequenceFeature feature)
+ {
+ return features == null || feature.begin > maxStart ? false
+ : features.contains(feature);
+ }
- return true;
+ /**
+ * Answers true if this store already contains a contact feature equal to the
+ * given feature (by {@code SequenceFeature.equals()} test), else false
+ *
+ * @param feature
+ * @return
+ */
+ private boolean containsContactFeature(SequenceFeature feature)
+ {
+ return contactFeatureStarts != null && feature.begin <= maxContactStart
+ && listContains(contactFeatureStarts, feature);
}
/**
- * 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.
+ * Answers true if this store already contains a non-positional feature equal
+ * to the given feature (by {@code SequenceFeature.equals()} test), else false
*
- * @param features
* @param feature
* @return
*/
- protected static boolean listContains(List<SequenceFeature> features,
- SequenceFeature feature)
+ private boolean containsNonPositionalFeature(SequenceFeature feature)
{
- if (features == null || feature == null)
- {
- return false;
- }
+ return nonPositionalFeatures == null ? false
+ : nonPositionalFeatures.contains(feature);
+ }
+
+ /**
+ * Deletes the given feature from the store, returning true if it was found
+ * (and deleted), else false. This method makes no assumption that the feature
+ * is in the 'expected' place in the store, in case it has been modified since
+ * it was added.
+ *
+ * @param sf
+ */
+ public synchronized boolean delete(SequenceFeature sf)
+ {
+ boolean removed = false;
/*
- * locate the first entry in the list which does not precede the feature
+ * try contact positions (and if found, delete
+ * from both lists of contact positions)
*/
- // int pos = binarySearch(features,
- // SearchCriterion.byFeature(feature, RangeComparator.BY_START_POSITION));
- int pos = BinarySearcher.findFirst(features,
- val -> val.getBegin() >= feature.getBegin());
- int len = features.size();
- while (pos < len)
+ if (!removed && contactFeatureStarts != null)
{
- SequenceFeature sf = features.get(pos);
- if (sf.getBegin() > feature.getBegin())
- {
- return false; // no match found
- }
- if (sf.equals(feature))
+ removed = contactFeatureStarts.remove(sf);
+ if (removed)
{
- return true;
+ contactFeatureEnds.remove(sf);
}
- pos++;
}
- return false;
+
+ /*
+ * if not found, try non-positional features
+ */
+ if (!removed && nonPositionalFeatures != null)
+ {
+ removed = nonPositionalFeatures.remove(sf);
+ }
+
+ /*
+ * if not found, try nested features
+ */
+ if (!removed && features != null)
+ {
+ removed = features.remove(sf);
+ }
+
+ if (removed)
+ {
+ rescanAfterDelete();
+ }
+
+ return removed;
+ }
+
+ public List<SequenceFeature> findFeatures(long start, long end)
+ {
+ return findFeatures(start, end, null);
}
/**
* Returns a (possibly empty) list of features whose extent overlaps the given
* range. The returned list is not ordered. Contact features are included if
- * either of the contact points lies within the range.
+ * either of the contact points lies within the range. If the {@code result}
+ * parameter is not null, new entries are added to this list and the (possibly
+ * extended) list returned.
*
* @param start
* start position of overlap range (inclusive)
* @param end
* end position of overlap range (inclusive)
+ * @param result
* @return
*/
-
- public List<SequenceFeature> findOverlappingFeatures(long start, long end)
+ public List<SequenceFeature> findFeatures(long start, long end,
+ List<SequenceFeature> result)
{
- List<SequenceFeature> result = new ArrayList<>();
-
- findContactFeatures(start, end, result);
-
- if (features != null)
+ if (result == null)
{
- result.addAll(features.findOverlaps(start, end));
+ result = new ArrayList<>();
}
+ findContactFeatures(start, end, result);
+ features.findOverlaps(start, end, result);
+
return result;
}
/**
- * Adds contact features to the result list where either the second or the
- * first contact position lies within the target range
+ * Returns a (possibly empty) list of stored contact features
*
- * @param from
- * @param to
- * @param result
+ * @return
*/
- protected void findContactFeatures(long from, long to,
- List<SequenceFeature> result)
+ public List<SequenceFeature> getContactFeatures()
{
- if (contactFeatureStarts != null)
- {
- findContactStartOverlaps(from, to, result);
- }
- if (contactFeatureEnds != null)
- {
- findContactEndOverlaps(from, to, result);
- }
+ List<SequenceFeature> result = new ArrayList<>();
+ getContactFeatures(result);
+ return result;
}
/**
- * Adds to the result list any contact features whose end (second contact
- * point), but not start (first contact point), lies in the query from-to
- * range
+ * Adds any stored contact features to the result list
*
- * @param from
- * @param to
- * @param result
+ * @return
*/
- protected void findContactEndOverlaps(long from, long to,
- List<SequenceFeature> result)
+ public void getContactFeatures(List<SequenceFeature> result)
{
- /*
- * find the first contact feature (if any)
- * whose end point is not before the target range
- */
- int index = BinarySearcher.findFirst(contactFeatureEnds,
- f -> f.getEnd() >= from);
-
- while (index < contactFeatureEnds.size())
+ if (contactFeatureStarts != null)
{
- SequenceFeature sf = contactFeatureEnds.get(index);
- if (!sf.isContactFeature())
- {
- System.err.println("Error! non-contact feature type " + sf.getType()
- + " in contact features list");
- index++;
- continue;
- }
-
- int begin = sf.getBegin();
- if (begin >= from && begin <= to)
- {
- /*
- * this feature's first contact position lies in the search range
- * so we don't include it in results a second time
- */
- index++;
- continue;
- }
-
- if (sf.getEnd() > to)
- {
- /*
- * this feature (and all following) has end point after the target range
- */
- break;
- }
-
- /*
- * feature has end >= from and end <= to
- * i.e. contact end point lies within overlap search range
- */
- result.add(sf);
- index++;
+ result.addAll(contactFeatureStarts);
}
}
/**
- * Adds contact features whose start position lies in the from-to range to the
- * result list
+ * Answers the number of positional (or non-positional) features stored.
+ * Contact features count as 1.
*
- * @param from
- * @param to
- * @param result
+ * @param positional
+ * @return
*/
- protected void findContactStartOverlaps(long from, long to,
- List<SequenceFeature> result)
+ public int getFeatureCount(boolean positional)
{
- int index = BinarySearcher.findFirst(contactFeatureStarts,
- f -> f.getBegin() >= from);
-
- while (index < contactFeatureStarts.size())
+ if (!positional)
{
- SequenceFeature sf = contactFeatureStarts.get(index);
- if (!sf.isContactFeature())
- {
- System.err.println("Error! non-contact feature " + sf.toString()
- + " in contact features list");
- index++;
- continue;
- }
- if (sf.getBegin() > to)
- {
- /*
- * 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++;
+ return nonPositionalFeatures == null ? 0
+ : nonPositionalFeatures.size();
}
- }
-
- /**
- * Answers a list of all positional features stored, in no guaranteed order
- *
- * @return
- */
- public List<SequenceFeature> getPositionalFeatures()
- {
- List<SequenceFeature> result = new ArrayList<>();
+ int size = 0;
- /*
- * add any contact features - from the list by start position
- */
if (contactFeatureStarts != null)
{
- result.addAll(contactFeatureStarts);
+ // note a contact feature (start/end) counts as one
+ size += contactFeatureStarts.size();
}
- /*
- * add any nested features
- */
if (features != null)
{
- result.addAll(features);
+ size += features.size();
}
-
- return result;
+ return size;
}
/**
- * Answers a list of all contact features. If there are none, returns an
- * immutable empty list.
+ * Answers the set of distinct feature groups stored, possibly including null,
+ * as an unmodifiable view of the set. The parameter determines whether the
+ * groups for positional or for non-positional features are returned.
*
+ * @param positionalFeatures
* @return
*/
-
- public List<SequenceFeature> getContactFeatures()
+ public Set<String> getFeatureGroups(boolean positionalFeatures)
{
- if (contactFeatureStarts == null)
+ if (positionalFeatures)
{
- return Collections.emptyList();
+ return Collections.unmodifiableSet(positionalFeatureGroups);
}
- return new ArrayList<>(contactFeatureStarts);
- }
-
- /**
- * Answers a list of all non-positional features. If there are none, returns
- * an immutable empty list.
- *
- * @return
- */
-
- public List<SequenceFeature> getNonPositionalFeatures()
- {
- if (nonPositionalFeatures == null)
+ else
{
- return Collections.emptyList();
+ return nonPositionalFeatureGroups == null
+ ? Collections.<String> emptySet()
+ : Collections.unmodifiableSet(nonPositionalFeatureGroups);
}
- return new ArrayList<>(nonPositionalFeatures);
}
/**
- * Deletes the given feature from the store, returning true if it was found
- * (and deleted), else false. This method makes no assumption that the feature
- * is in the 'expected' place in the store, in case it has been modified since
- * it was added.
+ * Answers a list of all either positional or non-positional features whose
+ * feature group matches the given group (which may be null)
*
- * @param sf
+ * @param positional
+ * @param group
+ * @return
*/
-
- public synchronized boolean delete(SequenceFeature sf)
+ public List<SequenceFeature> getFeaturesForGroup(boolean positional,
+ String group)
{
- boolean removed = false;
-
- /*
- * try contact positions (and if found, delete
- * from both lists of contact positions)
- */
- if (!removed && contactFeatureStarts != null)
- {
- removed = contactFeatureStarts.remove(sf);
- if (removed)
- {
- contactFeatureEnds.remove(sf);
- }
- }
-
- boolean removedNonPositional = false;
-
- /*
- * if not found, try non-positional features
- */
- if (!removed && nonPositionalFeatures != null)
- {
- removedNonPositional = nonPositionalFeatures.remove(sf);
- removed = removedNonPositional;
- }
+ List<SequenceFeature> result = new ArrayList<>();
/*
- * if not found, try nested features
+ * if we know features don't include the target group, no need
+ * to inspect them for matches
*/
- if (!removed && features != null)
- {
- removed = features.remove(sf);
- featuresList.remove(sf);
- }
-
- if (removed)
+ if (positional && !positionalFeatureGroups.contains(group)
+ || !positional && !nonPositionalFeatureGroups.contains(group))
{
- rescanAfterDelete();
+ return result;
}
- return removed;
- }
-
- /**
- * 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())
+ if (positional)
{
- nonPositionalFeatureGroups.add(sf.getFeatureGroup());
- float score = sf.getScore();
- nonPositionalMinScore = min(nonPositionalMinScore, score);
- nonPositionalMaxScore = max(nonPositionalMaxScore, score);
+ addFeaturesForGroup(group, contactFeatureStarts, result);
+ addFeaturesForGroup(group, features, result);
}
-
- /*
- * scan positional features for groups, scores and extents
- */
- for (SequenceFeature sf : getPositionalFeatures())
+ else
{
- positionalFeatureGroups.add(sf.getFeatureGroup());
- float score = sf.getScore();
- positionalMinScore = min(positionalMinScore, score);
- positionalMaxScore = max(positionalMaxScore, score);
- totalExtent += getFeatureLength(sf);
+ addFeaturesForGroup(group, nonPositionalFeatures, result);
}
+ return result;
}
/**
- * 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)
+ * 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 f1
- * @param f2
+ * @param positional
+ * @return
*/
- protected static float min(float f1, float f2)
+ public float getMaximumScore(boolean positional)
{
- if (Float.isNaN(f1))
- {
- return Float.isNaN(f2) ? f1 : f2;
- }
- else
- {
- return Float.isNaN(f2) ? f1 : Math.min(f1, f2);
- }
+ return positional ? positionalMaxScore : nonPositionalMaxScore;
}
/**
- * 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)
+ * 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 f1
- * @param f2
+ * @param positional
+ * @return
*/
- protected static float max(float f1, float f2)
+ public float getMinimumScore(boolean positional)
{
- if (Float.isNaN(f1))
- {
- return Float.isNaN(f2) ? f1 : f2;
- }
- else
- {
- return Float.isNaN(f2) ? f1 : Math.max(f1, f2);
- }
+ return positional ? positionalMinScore : nonPositionalMinScore;
}
/**
- * Answers true if this store has no features, else false
+ * Answers a (possibly empty) list of all non-positional features
*
* @return
*/
-
- public boolean isEmpty()
+ public List<SequenceFeature> getNonPositionalFeatures()
{
- boolean hasFeatures = (contactFeatureStarts != null
- && !contactFeatureStarts.isEmpty())
- || (nonPositionalFeatures != null
- && !nonPositionalFeatures.isEmpty())
- || (features != null && features.size() > 0);
-
- return !hasFeatures;
+ List<SequenceFeature> result = new ArrayList<>();
+ getNonPositionalFeatures(result);
+ return result;
}
/**
- * Answers the set of distinct feature groups stored, possibly including null,
- * as an unmodifiable view of the set. The parameter determines whether the
- * groups for positional or for non-positional features are returned.
+ * Adds any stored non-positional features to the result list
*
- * @param positionalFeatures
* @return
*/
-
- public Set<String> getFeatureGroups(boolean positionalFeatures)
+ public void getNonPositionalFeatures(List<SequenceFeature> result)
{
- if (positionalFeatures)
- {
- return Collections.unmodifiableSet(positionalFeatureGroups);
- }
- else
+ if (nonPositionalFeatures != null)
{
- return nonPositionalFeatureGroups == null
- ? Collections.<String> emptySet()
- : Collections.unmodifiableSet(nonPositionalFeatureGroups);
+ result.addAll(nonPositionalFeatures);
}
}
/**
- * Answers the number of positional (or non-positional) features stored.
- * Contact features count as 1.
+ * Returns a (possibly empty) list of all positional features stored
*
- * @param positional
* @return
*/
-
- public int getFeatureCount(boolean positional)
+ public List<SequenceFeature> getPositionalFeatures()
{
- if (!positional)
- {
- return nonPositionalFeatures == null ? 0
- : nonPositionalFeatures.size();
- }
+ List<SequenceFeature> result = new ArrayList<>();
+ getPositionalFeatures(result);
- int size = 0;
+ return result;
+ }
+ /**
+ * Adds all positional features stored to the result list, in no guaranteed
+ * order, and with no check for duplicates
+ */
+ public void getPositionalFeatures(List<SequenceFeature> result)
+ {
+ /*
+ * add any contact features - from the list by start position
+ */
if (contactFeatureStarts != null)
{
- // note a contact feature (start/end) counts as one
- size += contactFeatureStarts.size();
+ result.addAll(contactFeatureStarts);
}
+ /*
+ * add any nested features
+ */
if (features != null)
{
- size += features.size();
+ result.addAll(features);
}
-
- return size;
}
/**
*
* @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.
+ * Answers true if this store has no features, else false
*
- * @param positional
* @return
*/
-
- public float getMinimumScore(boolean positional)
+ public boolean isEmpty()
{
- return positional ? positionalMinScore : nonPositionalMinScore;
+ boolean hasFeatures = (contactFeatureStarts != null
+ && !contactFeatureStarts.isEmpty())
+ || (nonPositionalFeatures != null
+ && !nonPositionalFeatures.isEmpty())
+ || (features != null && features.size() > 0);
+
+ return !hasFeatures;
}
/**
- * 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
+ * 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.
*/
-
- public float getMaximumScore(boolean positional)
+ protected synchronized void rescanAfterDelete()
{
- return positional ? positionalMaxScore : nonPositionalMaxScore;
+ 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
+ */
+ if (nonPositionalFeatures != null)
+ {
+ for (int i = 0, n = nonPositionalFeatures.size(); i < n; i++)
+ {
+ SequenceFeature sf = nonPositionalFeatures.get(i);
+ nonPositionalFeatureGroups.add(sf.getFeatureGroup());
+ float score = sf.getScore();
+ nonPositionalMinScore = min(nonPositionalMinScore, score);
+ nonPositionalMaxScore = max(nonPositionalMaxScore, score);
+ }
+ }
+
+ rescanPositional(contactFeatureStarts);
+ rescanPositional(features);
}
/**
- * Answers a list of all either positional or non-positional features whose
- * feature group matches the given group (which may be null)
+ * Scans the given features and updates cached feature groups, minimum and
+ * maximum feature score, and total feature extent (length) for positional
+ * features
*
- * @param positional
- * @param group
- * @return
+ * @param sfs
*/
-
- public List<SequenceFeature> getFeaturesForGroup(boolean positional,
- String group)
+ private void rescanPositional(Collection<SequenceFeature> sfs)
{
- 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))
+ if (sfs == null)
{
- return result;
+ return;
}
-
- List<SequenceFeature> sfs = positional ? getPositionalFeatures()
- : getNonPositionalFeatures();
for (SequenceFeature sf : sfs)
{
- String featureGroup = sf.getFeatureGroup();
- if (group == null && featureGroup == null
- || group != null && group.equals(featureGroup))
- {
- result.add(sf);
- }
+ positionalFeatureGroups.add(sf.getFeatureGroup());
+ float score = sf.getScore();
+ positionalMinScore = min(positionalMinScore, score);
+ positionalMaxScore = max(positionalMaxScore, score);
+ totalExtent += getFeatureLength(sf);
}
- return result;
}
/**
* @param shiftBy
* @return
*/
-
public synchronized boolean shiftFeatures(int fromPosition, int shiftBy)
{
/*
* (Although a simple shift of all values would preserve data integrity!)
*/
boolean modified = false;
- for (SequenceFeature sf : getPositionalFeatures())
+ List<SequenceFeature> list = getPositionalFeatures();
+ for (int i = 0, n = list.size(); i < n; i++)
{
+ SequenceFeature sf = list.get(i);
if (sf.getBegin() >= fromPosition)
{
modified = true;
}
/**
- * Find all features containing this position.
+ * Answers the position (0, 1...) in the list of the first entry whose end
+ * position is not less than {@ pos}. If no such entry is found, answers the
+ * length of the list.
*
+ * @param list
* @param pos
- * @return list of SequenceFeatures
- * @author Bob Hanson 2019.07.30
+ * @return
*/
-
- public List<SequenceFeature> findOverlappingFeatures(int pos,
- List<SequenceFeature> result)
+ protected int findFirstEnd(List<SequenceFeature> list, long pos)
{
- if (result == null)
- {
- result = new ArrayList<>();
- }
-
- if (contactFeatureStarts != null)
- {
- findContacts(contactFeatureStarts, pos, result, true);
- findContacts(contactFeatureEnds, pos, result, false);
- }
- if (featuresList != null)
- {
- findOverlaps(featuresList, pos, result);
- }
- return result;
+ return BinarySearcher.findFirst(list, false, Compare.GE, (int) pos);
}
/**
- * Binary search for contact start or end at a given (Overview) position.
+ * Adds contact features to the result list where either the second or the
+ * first contact position lies within the target range
*
- * @param l
- * @param pos
+ * @param from
+ * @param to
* @param result
- * @param isStart
- *
- * @author Bob Hanson 2019.07.30
*/
- private static void findContacts(List<SequenceFeature> l, int pos,
- List<SequenceFeature> result, boolean isStart)
+ protected void findContactFeatures(long from, long to,
+ List<SequenceFeature> result)
{
- int low = 0;
- int high = l.size() - 1;
- while (low <= high)
+ if (contactFeatureStarts != null)
{
- int mid = (low + high) >>> 1;
- SequenceFeature f = l.get(mid);
- switch (Long.signum((isStart ? f.begin : f.end) - pos))
- {
- case -1:
- low = mid + 1;
- continue;
- case 1:
- high = mid - 1;
- continue;
- case 0:
- int m = mid;
- result.add(f);
- // could be "5" in 12345556788 ?
- while (++mid <= high && (f = l.get(mid)) != null
- && (isStart ? f.begin : f.end) == pos)
- {
- result.add(f);
- }
- while (--m >= low && (f = l.get(m)) != null
- && (isStart ? f.begin : f.end) == pos)
- {
- result.add(f);
- }
- return;
- }
+ findContactStartOverlaps(from, to, result);
+ findContactEndOverlaps(from, to, result);
}
}
- BitSet bs = new BitSet();
-
/**
- * Double binary sort with bitset correlation
- *
+ * Adds to the result list any contact features whose end (second contact
+ * point), but not start (first contact point), lies in the query from-to
+ * range
*
- * @param features
- * @param pos
+ * @param from
+ * @param to
* @param result
*/
- private void findOverlaps(List<SequenceFeature> features, int pos,
+ private void findContactEndOverlaps(long from, long to,
List<SequenceFeature> result)
{
- int n = featuresList.size();
- if (n == 1)
- {
- checkOne(featuresList.get(0), pos, result);
- return;
- }
- if (orderedFeatureStarts == null)
- {
- rebuildArrays(n);
- }
- SequenceFeature sf = findClosestFeature(orderedFeatureStarts, pos);
- while (sf != null) {
- if (sf.end >= pos)
- {
- result.add(sf);
- }
- sf = sf.containedBy;
- }
- }
+ /*
+ * find the first contact feature (if any)
+ * whose end point is not before the target range
+ */
+ int index = findFirstEnd(contactFeatureEnds, from);
- private void linkFeatures(SequenceFeature[] intervals)
- {
- if (intervals.length < 2)
- {
- return;
- }
- int maxEnd = intervals[0].end;
- for (int i = 1, n = intervals.length; i < n; i++)
+ int n = contactFeatureEnds.size();
+ while (index < n)
{
- SequenceFeature ithis = intervals[i];
- if (ithis.begin <= maxEnd)
- {
- ithis.containedBy = getContainedBy(intervals[i - 1], ithis);
- }
- if (ithis.end > maxEnd)
+ SequenceFeature sf = contactFeatureEnds.get(index);
+ if (!sf.isContactFeature())
{
- maxEnd = ithis.end;
+ System.err.println("Error! non-contact feature type " + sf.getType()
+ + " in contact features list");
+ index++;
+ continue;
}
- }
- }
- private SequenceFeature getContainedBy(SequenceFeature sf,
- SequenceFeature sf0)
- {
- int begin = sf0.begin;
- while (sf != null)
- {
- if (begin <= sf.end)
+ int begin = sf.getBegin();
+ if (begin >= from && begin <= to)
{
- System.out.println("\nFS found " + sf0.index + ":" + sf0
- + "\nFS in " + sf.index + ":" + sf);
- return sf;
+ /*
+ * this feature's first contact position lies in the search range
+ * so we don't include it in results a second time
+ */
+ index++;
+ continue;
}
- sf = sf.containedBy;
- }
- return null;
- }
- private SequenceFeature findClosestFeature(SequenceFeature[] l, int pos)
- {
- int low = 0;
- int high = l.length - 1;
- while (low <= high)
- {
- int mid = (low + high) >>> 1;
- SequenceFeature f = l[mid];
- switch (Long.signum(f.begin - pos))
+ if (sf.getEnd() > to)
{
- case -1:
- low = mid + 1;
- continue;
- case 1:
- high = mid - 1;
- continue;
- case 0:
-
- while (++mid <= high && l[mid].begin == pos)
- {
- ;
- }
- mid--;
- return l[mid];
+ /*
+ * this feature (and all following) has end point after the target range
+ */
+ break;
}
- }
- // -1 here?
- return (high < 0 || low >= l.length ? null : l[high]);
- }
- private void checkOne(SequenceFeature sf, int pos,
- List<SequenceFeature> result)
- {
- if (sf.begin <= pos && sf.end >= pos)
- {
+ /*
+ * feature has end >= from and end <= to
+ * i.e. contact end point lies within overlap search range
+ */
result.add(sf);
+ index++;
}
- return;
}
- /*
- * contact features ordered by first contact position
+ /**
+ * Adds contact features whose start position lies in the from-to range to the
+ * result list
+ *
+ * @param from
+ * @param to
+ * @param result
*/
- private SequenceFeature[] orderedFeatureStarts;
-
- private void rebuildArrays(int n)
+ private void findContactStartOverlaps(long from, long to,
+ List<SequenceFeature> result)
{
- if (startComp == null)
- {
- startComp = new StartComparator();
- }
- orderedFeatureStarts = new SequenceFeature[n];
+ int index = BinarySearcher.findFirst(contactFeatureStarts, true,
+ Compare.GE, (int) from);
- for (int i = n; --i >= 0;)
+ while (index < contactFeatureStarts.size())
{
- SequenceFeature sf = featuresList.get(i);
- sf.index = i;
- orderedFeatureStarts[i] = sf;
- }
- Arrays.sort(orderedFeatureStarts, startComp);
- linkFeatures(orderedFeatureStarts);
- }
-
- class StartComparator implements Comparator<SequenceFeature>
- {
-
- int pos;
+ SequenceFeature sf = contactFeatureStarts.get(index);
+ if (!sf.isContactFeature())
+ {
+ System.err.println("Error! non-contact feature " + sf.toString()
+ + " in contact features list");
+ index++;
+ continue;
+ }
+ if (sf.getBegin() > to)
+ {
+ /*
+ * this feature's start (and all following) follows the target range
+ */
+ break;
+ }
- @Override
- public int compare(SequenceFeature o1, SequenceFeature o2)
- {
- int p1 = o1.begin;
- int p2 = o2.begin;
- return (p1 < p2 ? -1 : p1 > p2 ? 1 : 0);
+ /*
+ * feature has begin >= from and begin <= to
+ * i.e. contact start point lies within overlap search range
+ */
+ result.add(sf);
+ index++;
}
-
}
- static StartComparator startComp;
-
- // class EndComparator implements Comparator<SequenceFeature>
- // {
- //
- // int pos;
- //
- // @Override
- // public int compare(SequenceFeature o1, SequenceFeature o2)
- // {
- // int p1 = o1.end;
- // int p2 = o2.end;
- // int val = (p1 < p2 ? 1 : p1 > p2 ? -1 : 0);
- // return val;
- // }
- //
- // }
- //
- // static EndComparator endComp;
-
}