import java.util.List;
import java.util.Set;
+import intervalstore.api.IntervalStoreI;
import intervalstore.impl.BinarySearcher;
+import intervalstore.impl.BinarySearcher.Compare;
-public abstract class FeatureStore implements FeatureStoreI
+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.
* IntervalStore holds remaining features and provides efficient
* query for features overlapping any given interval
*/
- Collection<SequenceFeature> features;
-
- @Override
- public Collection<SequenceFeature> getFeatures()
- {
- return features;
- }
+ IntervalStoreI<SequenceFeature> features;
/*
* Feature groups represented in stored positional features
float nonPositionalMaxScore;
/**
- * Constructor
+ * Answers the 'length' of the feature, counting 0 for non-positional features
+ * and 1 for contact features
+ *
+ * @param feature
+ * @return
*/
- public FeatureStore()
+ protected static int getFeatureLength(SequenceFeature feature)
{
- 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
+ if (feature.isNonPositional())
+ {
+ return 0;
+ }
+ if (feature.isContactFeature())
+ {
+ return 1;
+ }
+ return 1 + feature.getEnd() - feature.getBegin();
}
/**
- * 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.
+ * 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
*/
-
- @Override
- public boolean addFeature(SequenceFeature feature)
+ public static boolean listContains(List<SequenceFeature> list,
+ SequenceFeature feature)
{
- if (contains(feature))
+ if (list == null || feature == null)
{
return false;
}
/*
- * keep a record of feature groups
- */
- if (!feature.isNonPositional())
- {
- positionalFeatureGroups.add(feature.getFeatureGroup());
- }
-
- if (feature.isContactFeature())
- {
- addContactFeature(feature);
- }
- else if (feature.isNonPositional())
- {
- addNonPositionalFeature(feature);
- }
- else
- {
- 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
+ * locate the first entry in the list which does not precede the feature
*/
- float score = feature.getScore();
- if (!Float.isNaN(score))
+ int begin = feature.begin;
+ int pos = BinarySearcher.findFirst(list, true, Compare.GE, begin);
+ int len = list.size();
+ while (pos < len)
{
- if (feature.isNonPositional())
+ SequenceFeature sf = list.get(pos);
+ if (sf.begin > begin)
{
- nonPositionalMinScore = min(nonPositionalMinScore, score);
- nonPositionalMaxScore = max(nonPositionalMaxScore, score);
+ return false; // no match found
}
- else
+ if (sf.equals(feature))
{
- positionalMinScore = min(positionalMinScore, score);
- positionalMaxScore = max(positionalMaxScore, score);
+ return true;
}
+ pos++;
}
-
- return true;
+ return false;
}
/**
- * Answers true if this store contains the given feature (testing by
- * SequenceFeature.equals), 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 feature
- * @return
+ * @param f1
+ * @param f2
*/
- @Override
- public boolean contains(SequenceFeature feature)
+ protected static float max(float f1, float f2)
{
- if (feature.isNonPositional())
+ if (Float.isNaN(f1))
{
- return nonPositionalFeatures == null ? false
- : nonPositionalFeatures.contains(feature);
+ return Float.isNaN(f2) ? f1 : f2;
}
-
- if (feature.isContactFeature())
+ else
{
- return contactFeatureStarts != null
- && listContains(contactFeatureStarts, feature);
+ return Float.isNaN(f2) ? f1 : Math.max(f1, f2);
}
-
- 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 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 feature
- * @return
+ * @param f1
+ * @param f2
*/
- protected static int getFeatureLength(SequenceFeature feature)
+ protected static float min(float f1, float f2)
{
- if (feature.isNonPositional())
+ if (Float.isNaN(f1))
{
- return 0;
+ return Float.isNaN(f2) ? f1 : f2;
}
- if (feature.isContactFeature())
+ else
{
- return 1;
+ return Float.isNaN(f2) ? f1 : Math.min(f1, f2);
}
- return 1 + feature.getEnd() - feature.getBegin();
}
/**
- * 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
+ * Constructor that defaults to using NCList IntervalStore
*/
- protected boolean addNonPositionalFeature(SequenceFeature feature)
+ public FeatureStore()
{
- if (nonPositionalFeatures == null)
- {
- nonPositionalFeatures = new ArrayList<>();
- }
-
- nonPositionalFeatures.add(feature);
+ this(IntervalStoreType.INTERVAL_STORE_NCLIST);
+ }
- nonPositionalFeatureGroups.add(feature.getFeatureGroup());
+ /**
+ * 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;
+ positionalMaxScore = Float.NaN;
+ nonPositionalMinScore = Float.NaN;
+ nonPositionalMaxScore = Float.NaN;
- return true;
+ // only construct nonPositionalFeatures or contactFeatures if needed
}
/**
- * Adds one feature to the IntervalStore that can manage nested features
- * (creating the IntervalStore if necessary)
+ * Returns a new instance of IntervalStoreI of implementation as selected by
+ * the type parameter
+ *
+ * @param type
+ * @return
*/
- protected synchronized void addNestedFeature(SequenceFeature feature)
+ private IntervalStoreI<SequenceFeature> getIntervalStore(
+ IntervalStoreType type)
{
- features.add(feature);
+ 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<>();
+ }
}
+
/**
* 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
* 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);
-
+ 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
*/
- insertPosition = BinarySearcher.findFirst(contactFeatureEnds,
- f -> f.getEnd() >= feature.getEnd());
- contactFeatureEnds.add(insertPosition, feature);
+ contactFeatureEnds.add(findFirstEnd(contactFeatureEnds, feature.end),
+ 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.
+ * 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 features
* @param feature
- * @return
*/
- protected static boolean listContains(List<SequenceFeature> features,
- SequenceFeature feature)
+ public boolean addFeature(SequenceFeature feature)
{
- if (features == null || feature == null)
+ if (feature.isContactFeature())
{
- return false;
+ 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
+ {
+ if (!features.add(feature, false))
+ {
+ return false;
+ }
+ positionalFeatureGroups.add(feature.getFeatureGroup());
+ if (feature.begin > maxStart)
+ {
+ maxStart = feature.begin;
+ }
}
/*
- * locate the first entry in the list which does not precede the feature
+ * record the total extent of positional features, to make
+ * getTotalFeatureLength possible; we count the length of a
+ * contact feature as 1
*/
- // 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)
+ totalExtent += getFeatureLength(feature);
+
+ /*
+ * record the minimum and maximum score for positional
+ * and non-positional features
+ */
+ float score = feature.getScore();
+ if (!Float.isNaN(score))
{
- SequenceFeature sf = features.get(pos);
- if (sf.getBegin() > feature.getBegin())
+ if (feature.isNonPositional())
{
- return false; // no match found
+ nonPositionalMinScore = min(nonPositionalMinScore, score);
+ nonPositionalMaxScore = max(nonPositionalMaxScore, score);
}
- if (sf.equals(feature))
+ else
{
- return true;
+ positionalMinScore = min(positionalMinScore, score);
+ positionalMaxScore = max(positionalMaxScore, score);
}
- pos++;
}
- return false;
- }
- abstract protected void findContactFeatures(long from, long to,
- List<SequenceFeature> result);
+ return true;
+ }
/**
- * Answers a list of all positional features stored, in no guaranteed order
+ * A helper method that adds to the result list any features from the
+ * collection provided whose feature group matches the specified group
*
- * @return
+ * @param group
+ * @param sfs
+ * @param result
*/
-
- @Override
- public List<SequenceFeature> getPositionalFeatures(
- List<SequenceFeature> result)
+ private void addFeaturesForGroup(String group,
+ Collection<SequenceFeature> sfs, List<SequenceFeature> result)
{
-
- /*
- * add any contact features - from the list by start position
- */
- if (contactFeatureStarts != null)
+ if (sfs == null)
{
- result.addAll(contactFeatureStarts);
+ return;
}
+ for (SequenceFeature sf : sfs)
+ {
+ String featureGroup = sf.getFeatureGroup();
+ if (group == null && featureGroup == null
+ || group != null && group.equals(featureGroup))
+ {
+ result.add(sf);
+ }
+ }
+ }
- /*
- * add any nested features
- */
- if (features != null)
+ /**
+ * 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
+ */
+ protected boolean addNonPositionalFeature(SequenceFeature feature)
+ {
+ if (nonPositionalFeatures == null)
{
- result.addAll(features);
+ nonPositionalFeatures = new ArrayList<>();
}
- return result;
+ nonPositionalFeatures.add(feature);
+
+ nonPositionalFeatureGroups.add(feature.getFeatureGroup());
+
+ return true;
}
/**
- * Answers a list of all contact features. If there are none, returns an
- * immutable empty list.
+ * Answers true if this store contains the given feature (testing by
+ * SequenceFeature.equals), else false
*
+ * @param feature
* @return
*/
-
- @Override
- public List<SequenceFeature> getContactFeatures(
- List<SequenceFeature> result)
+ public boolean contains(SequenceFeature feature)
{
- if (contactFeatureStarts != null)
+ if (feature.isNonPositional())
{
- result.addAll(contactFeatureStarts);
+ return containsNonPositionalFeature(feature);
}
- return result;
+
+ if (feature.isContactFeature())
+ {
+ return containsContactFeature(feature);
+ }
+
+ return containsPositionalFeature(feature);
+ }
+
+ private boolean containsPositionalFeature(SequenceFeature feature)
+ {
+ return features == null || feature.begin > maxStart ? false
+ : features.contains(feature);
}
/**
- * Answers a list of all non-positional features. If there are none, returns
- * an immutable empty list.
+ * 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);
+ }
- @Override
- public List<SequenceFeature> getNonPositionalFeatures(
- List<SequenceFeature> result)
+ /**
+ * Answers true if this store already contains a non-positional feature equal
+ * to the given feature (by {@code SequenceFeature.equals()} test), else false
+ *
+ * @param feature
+ * @return
+ */
+ private boolean containsNonPositionalFeature(SequenceFeature feature)
{
- if (nonPositionalFeatures != null)
- {
- result.addAll(nonPositionalFeatures);
- }
- return result;
+ return nonPositionalFeatures == null ? false
+ : nonPositionalFeatures.contains(feature);
}
/**
*
* @param sf
*/
-
- @Override
public synchronized boolean delete(SequenceFeature sf)
{
boolean removed = false;
}
}
- boolean removedNonPositional = false;
-
/*
* if not found, try non-positional features
*/
if (!removed && nonPositionalFeatures != null)
{
- removedNonPositional = nonPositionalFeatures.remove(sf);
- removed = removedNonPositional;
+ removed = nonPositionalFeatures.remove(sf);
}
/*
return removed;
}
+ public List<SequenceFeature> findFeatures(long start, long end)
+ {
+ return findFeatures(start, end, null);
+ }
+
/**
- * 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.
+ * 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. 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
*/
- protected synchronized void rescanAfterDelete()
+ public List<SequenceFeature> findFeatures(long start, long end,
+ List<SequenceFeature> result)
{
- 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)
+ if (result == null)
{
- for (SequenceFeature sf : nonPositionalFeatures)
- {
- nonPositionalFeatureGroups.add(sf.getFeatureGroup());
- float score = sf.getScore();
- nonPositionalMinScore = min(nonPositionalMinScore, score);
- nonPositionalMaxScore = max(nonPositionalMaxScore, score);
- }
+ result = new ArrayList<>();
}
- /*
- * scan positional features for groups, scores and extents
- */
+ findContactFeatures(start, end, result);
+ features.findOverlaps(start, end, result);
- rescanPositional(contactFeatureStarts);
- rescanPositional(features);
- }
-
- private void rescanPositional(Collection<SequenceFeature> sfs)
- {
- if (sfs == null)
- {
- return;
- }
- for (SequenceFeature sf : sfs)
- {
- positionalFeatureGroups.add(sf.getFeatureGroup());
- float score = sf.getScore();
- positionalMinScore = min(positionalMinScore, score);
- positionalMaxScore = max(positionalMaxScore, score);
- totalExtent += getFeatureLength(sf);
- }
+ 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)
+ * Returns a (possibly empty) list of stored contact features
*
- * @param f1
- * @param f2
+ * @return
*/
- protected static float min(float f1, float f2)
+ public List<SequenceFeature> getContactFeatures()
{
- if (Float.isNaN(f1))
- {
- return Float.isNaN(f2) ? f1 : f2;
- }
- else
- {
- return Float.isNaN(f2) ? f1 : Math.min(f1, f2);
- }
+ List<SequenceFeature> result = new ArrayList<>();
+ getContactFeatures(result);
+ return result;
}
/**
- * 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)
+ * Adds any stored contact features to the result list
*
- * @param f1
- * @param f2
+ * @return
*/
- protected static float max(float f1, float f2)
+ public void getContactFeatures(List<SequenceFeature> result)
{
- if (Float.isNaN(f1))
- {
- return Float.isNaN(f2) ? f1 : f2;
- }
- else
+ if (contactFeatureStarts != null)
{
- return Float.isNaN(f2) ? f1 : Math.max(f1, f2);
+ result.addAll(contactFeatureStarts);
}
}
-
/**
- * Answers true if this store has no features, else false
+ * Answers the number of positional (or non-positional) features stored.
+ * Contact features count as 1.
*
+ * @param positional
* @return
*/
-
- @Override
- public boolean isEmpty()
+ public int getFeatureCount(boolean positional)
{
- boolean hasFeatures = (contactFeatureStarts != null
- && !contactFeatureStarts.isEmpty())
- || (nonPositionalFeatures != null
- && !nonPositionalFeatures.isEmpty())
- || features.size() > 0;
+ if (!positional)
+ {
+ return nonPositionalFeatures == null ? 0
+ : nonPositionalFeatures.size();
+ }
- return !hasFeatures;
+ int size = 0;
+
+ if (contactFeatureStarts != null)
+ {
+ // note a contact feature (start/end) counts as one
+ size += contactFeatureStarts.size();
+ }
+
+ if (features != null)
+ {
+ size += features.size();
+ }
+ return size;
}
/**
* @param positionalFeatures
* @return
*/
-
- @Override
public Set<String> getFeatureGroups(boolean positionalFeatures)
{
if (positionalFeatures)
* @param group
* @return
*/
-
- @Override
public List<SequenceFeature> getFeaturesForGroup(boolean positional,
String group)
{
return result;
}
- private void addFeaturesForGroup(String group,
- Collection<SequenceFeature> sfs, List<SequenceFeature> result)
+ /**
+ * 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)
{
- if (sfs == null)
- {
- return;
- }
- for (SequenceFeature sf : sfs)
- {
- String featureGroup = sf.getFeatureGroup();
- if (group == null && featureGroup == null
- || group != null && group.equals(featureGroup))
- {
- result.add(sf);
- }
- }
+ return positional ? positionalMaxScore : nonPositionalMaxScore;
}
/**
- * Answers the number of positional (or non-positional) features stored.
- * Contact features count as 1.
+ * 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;
+ }
- @Override
- public int getFeatureCount(boolean positional)
+ /**
+ * Answers a (possibly empty) list of all non-positional features
+ *
+ * @return
+ */
+ public List<SequenceFeature> getNonPositionalFeatures()
{
- if (!positional)
+ List<SequenceFeature> result = new ArrayList<>();
+ getNonPositionalFeatures(result);
+ return result;
+ }
+
+ /**
+ * Adds any stored non-positional features to the result list
+ *
+ * @return
+ */
+ public void getNonPositionalFeatures(List<SequenceFeature> result)
+ {
+ if (nonPositionalFeatures != null)
{
- return nonPositionalFeatures == null ? 0
- : nonPositionalFeatures.size();
+ result.addAll(nonPositionalFeatures);
}
+ }
- return (contactFeatureStarts == null ? 0 : contactFeatureStarts.size())
- + features.size();
+ /**
+ * Returns a (possibly empty) list of all positional features stored
+ *
+ * @return
+ */
+ public List<SequenceFeature> getPositionalFeatures()
+ {
+ List<SequenceFeature> result = new ArrayList<>();
+ getPositionalFeatures(result);
+ 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)
+ {
+ result.addAll(contactFeatureStarts);
+ }
+
+ /*
+ * add any nested features
+ */
+ if (features != null)
+ {
+ result.addAll(features);
+ }
+ }
/**
* Answers the total length of positional features (or zero if there are
*
* @return
*/
-
- @Override
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 boolean isEmpty()
+ {
+ boolean hasFeatures = (contactFeatureStarts != null
+ && !contactFeatureStarts.isEmpty())
+ || (nonPositionalFeatures != null
+ && !nonPositionalFeatures.isEmpty())
+ || (features != null && features.size() > 0);
- @Override
- public float getMinimumScore(boolean positional)
+ return !hasFeatures;
+ }
+
+ /**
+ * 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()
{
- return positional ? positionalMinScore : nonPositionalMinScore;
+ 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 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.
+ * Scans the given features and updates cached feature groups, minimum and
+ * maximum feature score, and total feature extent (length) for positional
+ * features
*
- * @param positional
- * @return
+ * @param sfs
*/
-
- @Override
- public float getMaximumScore(boolean positional)
+ private void rescanPositional(Collection<SequenceFeature> sfs)
{
- return positional ? positionalMaxScore : nonPositionalMaxScore;
+ if (sfs == null)
+ {
+ return;
+ }
+ for (SequenceFeature sf : sfs)
+ {
+ positionalFeatureGroups.add(sf.getFeatureGroup());
+ float score = sf.getScore();
+ positionalMinScore = min(positionalMinScore, score);
+ positionalMaxScore = max(positionalMaxScore, score);
+ totalExtent += getFeatureLength(sf);
+ }
}
-
/**
* 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
* @param shiftBy
* @return
*/
-
- @Override
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;
return modified;
}
-
- @Override
- public List<SequenceFeature> findOverlappingFeatures(long start, long end)
+ /**
+ * 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
+ */
+ protected int findFirstEnd(List<SequenceFeature> list, long pos)
{
- return findOverlappingFeatures(start, end, null);
+ return BinarySearcher.findFirst(list, false, Compare.GE, (int) pos);
}
- @Override
- public List<SequenceFeature> getPositionalFeatures()
+ /**
+ * Adds contact features to the result list where either the second or the
+ * first contact position lies within the target range
+ *
+ * @param from
+ * @param to
+ * @param result
+ */
+ protected void findContactFeatures(long from, long to,
+ List<SequenceFeature> result)
{
- return getPositionalFeatures(new ArrayList<>());
+ if (contactFeatureStarts != null)
+ {
+ findContactStartOverlaps(from, to, result);
+ findContactEndOverlaps(from, to, result);
+ }
}
- @Override
- public List<SequenceFeature> getContactFeatures()
+ /**
+ * 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 from
+ * @param to
+ * @param result
+ */
+ private void findContactEndOverlaps(long from, long to,
+ List<SequenceFeature> result)
{
- return getContactFeatures(new ArrayList<>());
+ /*
+ * find the first contact feature (if any)
+ * whose end point is not before the target range
+ */
+ int index = findFirstEnd(contactFeatureEnds, from);
+
+ int n = contactFeatureEnds.size();
+ while (index < n)
+ {
+ 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++;
+ }
}
- @Override
- public List<SequenceFeature> getNonPositionalFeatures()
+ /**
+ * Adds contact features whose start position lies in the from-to range to the
+ * result list
+ *
+ * @param from
+ * @param to
+ * @param result
+ */
+ private void findContactStartOverlaps(long from, long to,
+ List<SequenceFeature> result)
{
- return getNonPositionalFeatures(new ArrayList<>());
+ int index = BinarySearcher.findFirst(contactFeatureStarts, true,
+ Compare.GE, (int) from);
+
+ while (index < contactFeatureStarts.size())
+ {
+ 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++;
+ }
}
}