X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fdatamodel%2Ffeatures%2FFeatureStore.java;h=d832f4de226f8ebeaf2c0a7f3fb23b75a2e098fe;hb=da8e34c5aeee5e83aa844d374eb21e4c2e3cebef;hp=cb0c36e91e6ebd4e64a878f87e444e03a9fae980;hpb=6a2fd9a13d79f316acdd4fcff066b8cfaaefa939;p=jalview.git diff --git a/src/jalview/datamodel/features/FeatureStore.java b/src/jalview/datamodel/features/FeatureStore.java index cb0c36e..d832f4d 100644 --- a/src/jalview/datamodel/features/FeatureStore.java +++ b/src/jalview/datamodel/features/FeatureStore.java @@ -1,80 +1,155 @@ +/* + * 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 . + * The Jalview Authors are detailed in the 'AUTHORS' file. + */ package jalview.datamodel.features; import jalview.datamodel.SequenceFeature; import java.util.ArrayList; +import java.util.Collection; import java.util.Collections; -import java.util.Comparator; import java.util.HashSet; import java.util.List; import java.util.Set; -/** - * 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 abstract class FeatureStore implements FeatureStoreI { + /** - * a class providing criteria for performing a binary search of a list + * track last start for quick insertion of ordered features */ - abstract static class SearchCriterion - { - /** - * Answers true if the entry passes the search criterion test - * - * @param entry - * @return - */ - abstract boolean compare(SequenceFeature entry); + protected int lastStart = -1, lastContactStart = -1; - static SearchCriterion byStart(final long target) + /** + * 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 new SearchCriterion() { - - @Override - boolean compare(SequenceFeature entry) - { - return entry.getBegin() >= target; - } - }; + return 0; } - - static SearchCriterion byEnd(final long target) + if (feature.isContactFeature()) { - return new SearchCriterion() - { + return 1; + } + return 1 + feature.getEnd() - feature.getBegin(); + } - @Override - boolean compare(SequenceFeature entry) - { - return entry.getEnd() >= target; - } - }; + /** + * 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 listContains(List list, + SequenceFeature feature) + { + if (list == null || feature == null) + { + return false; } - static SearchCriterion byFeature(final ContiguousI to, - final Comparator rc) + return (getEquivalentFeatureIndex(list, feature) >= 0); + } + + /** + * Binary search for the index (>= 0) of a feature in a list. + * + * @param list + * @param feature + * @return index if found; -1 if not + */ + protected int getEquivalentFeatureIndex(List list, + SequenceFeature feature) + { + + /* + * locate the first entry in the list which does not precede the feature + */ + int begin = feature.begin; + int pos = findFirstBegin(list, begin); + int len = list.size(); + while (pos < len) { - return new SearchCriterion() + SequenceFeature sf = list.get(pos); + if (sf.begin > begin) { - - @Override - boolean compare(SequenceFeature entry) - { - return rc.compare(entry, to) >= 0; - } - }; + return -1; // no match found + } + if (sf.equals(feature)) + { + return pos; + } + pos++; } + return -1; } - Comparator startOrdering = new RangeComparator(true); + /** + * 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); + } + } - Comparator endOrdering = new RangeComparator(false); + /** + * 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); + } + } /* * Non-positional features have no (zero) start/end position. @@ -83,14 +158,6 @@ public class FeatureStore List 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 nonNestedFeatures; - - /* * contact features ordered by first contact position */ List contactFeatureStarts; @@ -101,13 +168,10 @@ public class FeatureStore List 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 nestedFeatures; + Collection features; /* * Feature groups represented in stored positional features @@ -127,16 +191,60 @@ public class FeatureStore */ int totalExtent; + float positionalMinScore; + + float positionalMaxScore; + + float nonPositionalMinScore; + + float nonPositionalMaxScore; + /** * Constructor */ public FeatureStore() { - nonNestedFeatures = new ArrayList(); - positionalFeatureGroups = new HashSet(); + 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 + } + + /** + * 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 (contactFeatureStarts == null) + { + contactFeatureStarts = new ArrayList<>(); + contactFeatureEnds = new ArrayList<>(); + } + + /* + * insert into list sorted by start (first contact position): + * binary search the sorted list to find the insertion point + */ + contactFeatureStarts.add( + findFirstBegin(contactFeatureStarts, feature.begin), feature); + /* + * insert into list sorted by end (second contact position): + * binary search the sorted list to find the insertion point + */ + contactFeatureEnds.add(findFirstEnd(contactFeatureEnds, feature.end), + feature); - // we only construct nonPositionalFeatures, contactFeatures - // or the NCList if we need to + return true; } /** @@ -147,8 +255,15 @@ public class FeatureStore * * @param feature */ + + @Override public boolean addFeature(SequenceFeature feature) { + if (contains(feature)) + { + return false; + } + /* * keep a record of feature groups */ @@ -157,549 +272,414 @@ public class FeatureStore positionalFeatureGroups.add(feature.getFeatureGroup()); } - boolean added = false; - if (feature.isContactFeature()) { - added = addContactFeature(feature); + if (feature.begin > lastContactStart) + { + lastContactStart = feature.begin; + } + addContactFeature(feature); } else if (feature.isNonPositional()) { - added = addNonPositionalFeature(feature); + addNonPositionalFeature(feature); } else { - if (!nonNestedFeatures.contains(feature)) + addPositionalFeature(feature); + if (feature.begin > lastStart) { - added = addNonNestedFeature(feature); - if (!added) - { - /* - * detected a nested feature - put it in the NCList structure - */ - added = addNestedFeature(feature); - } + lastStart = feature.begin; } } /* * record the total extent of positional features, to make - * getAverageFeatureLength possible; we count the length of a + * getTotalFeatureLength possible; we count the length of a * contact feature as 1 */ - if (added && !feature.isNonPositional()) + totalExtent += getFeatureLength(feature); + + /* + * record the minimum and maximum score for positional + * and non-positional features + */ + float score = feature.getScore(); + if (!Float.isNaN(score)) { - int featureLength = feature.isContactFeature() ? 1 : 1 - + feature.getEnd() - feature.getBegin(); - totalExtent += featureLength; + if (feature.isNonPositional()) + { + 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. - * - * @param feature - */ - protected boolean addNonPositionalFeature(SequenceFeature feature) + private void addFeaturesForGroup(String group, + Collection sfs, List result) { - if (nonPositionalFeatures == null) + if (sfs == null) { - nonPositionalFeatures = new ArrayList(); - nonPositionalFeatureGroups = new HashSet(); + return; } - if (nonPositionalFeatures.contains(feature)) + for (SequenceFeature sf : sfs) { - return false; + String featureGroup = sf.getFeatureGroup(); + if (group == null && featureGroup == null + || group != null && group.equals(featureGroup)) + { + result.add(sf); + } } - - nonPositionalFeatures.add(feature); - - nonPositionalFeatureGroups.add(feature.getFeatureGroup()); - - return true; } /** - * 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. + * Adds one feature to the IntervalStore that can manage nested features + * (creating the IntervalStore if necessary) */ - protected synchronized boolean addNestedFeature(SequenceFeature feature) - { - if (nestedFeatures == null) - { - nestedFeatures = new NCList(feature); - return true; - } - return nestedFeatures.add(feature, false); - } + abstract protected void addPositionalFeature(SequenceFeature feature); /** - * 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) - { - /* - * find the first stored feature which doesn't precede the new one - */ - int insertPosition = binarySearch(nonNestedFeatures, - SearchCriterion.byFeature(feature, startOrdering)); - - /* - * 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()) - { - if (encloses(feature, nonNestedFeatures.get(insertPosition))) - { - return false; - } - } + if (nonPositionalFeatures == null) + { + nonPositionalFeatures = new ArrayList<>(); + } - /* - * checks passed - add the feature - */ - nonNestedFeatures.add(insertPosition, feature); + nonPositionalFeatures.add(feature); - return true; - } + nonPositionalFeatureGroups.add(feature.getFeatureGroup()); + + return true; } /** - * Answers true if range1 properly encloses range2, else false + * Answers true if this store contains the given feature (testing by + * SequenceFeature.equals), else false * - * @param range1 - * @param range2 + * @param feature * @return */ - protected boolean encloses(ContiguousI range1, ContiguousI range2) + @Override + public boolean contains(SequenceFeature feature) { - int begin1 = range1.getBegin(); - int begin2 = range2.getBegin(); - int end1 = range1.getEnd(); - int end2 = range2.getEnd(); - if (begin1 == begin2 && end1 > end2) + if (feature.isNonPositional()) { - return true; + return nonPositionalFeatures == null ? false + : nonPositionalFeatures.contains(feature); } - if (begin1 < begin2 && end1 >= end2) + + if (feature.isContactFeature()) { - return true; + return contactFeatureStarts != null + && feature.begin <= lastContactStart + && listContains(contactFeatureStarts, feature); } - return false; + + return features == null || feature.begin > lastStart ? false + : containsFeature(feature); } + + abstract protected boolean containsFeature(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. If the contact feature lists already contain the - * given feature (by test for equality), does not add it and returns false. + * 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 feature - * @return + * @param sf */ - protected synchronized boolean addContactFeature(SequenceFeature feature) + + @Override + public synchronized boolean delete(SequenceFeature sf) { - if (contactFeatureStarts == null) + boolean removed = false; + + /* + * try contact positions (and if found, delete + * from both lists of contact positions) + */ + if (!removed && contactFeatureStarts != null) { - contactFeatureStarts = new ArrayList(); + removed = contactFeatureStarts.remove(sf); + if (removed) + { + contactFeatureEnds.remove(sf); + } } - if (contactFeatureEnds == null) + + /* + * if not found, try non-positional features + */ + if (!removed && nonPositionalFeatures != null) { - contactFeatureEnds = new ArrayList(); + removed = nonPositionalFeatures.remove(sf); } - // TODO binary search for insertion points! - if (contactFeatureStarts.contains(feature)) + /* + * if not found, try nested features + */ + if (!removed && features != null) { - return false; + removed = findAndRemoveNonContactFeature(sf); } - contactFeatureStarts.add(feature); - Collections.sort(contactFeatureStarts, startOrdering); - contactFeatureEnds.add(feature); - Collections.sort(contactFeatureEnds, endOrdering); + if (removed) + { + rescanAfterDelete(); + } - return true; + return removed; } - /** - * 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. - * - * @param start - * start position of overlap range (inclusive) - * @param end - * end position of overlap range (inclusive) - * @return - */ - public List findOverlappingFeatures(long start, long end) - { - List result = new ArrayList(); + abstract protected boolean findAndRemoveNonContactFeature(SequenceFeature sf); - findNonNestedFeatures(start, end, result); + abstract protected void findContactFeatures(long from, long to, + List result); - findContactFeatures(start, end, result); + abstract protected int findFirstBegin(List list, + long pos); - if (nestedFeatures != null) - { - result.addAll(nestedFeatures.findOverlaps(start, end)); - } + abstract protected int findFirstEnd(List list, long pos); - return result; + @Override + public List findOverlappingFeatures(long start, long end) + { + return findOverlappingFeatures(start, end, null); + } + + @Override + public List getContactFeatures() + { + return getContactFeatures(new ArrayList<>()); } /** - * Adds contact features to the result list where either the second or the - * first contact position lies within the target range + * Answers a list of all contact features. If there are none, returns an + * immutable empty list. * - * @param from - * @param to - * @param result + * @return */ - protected void findContactFeatures(long from, long to, + + @Override + public List getContactFeatures( List result) { if (contactFeatureStarts != null) { - findContactStartFeatures(from, to, result); - } - if (contactFeatureEnds != null) - { - findContactEndFeatures(from, to, result); + result.addAll(contactFeatureStarts); } + 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 + * 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 findContactEndFeatures(long from, long to, - List result) + + @Override + public int getFeatureCount(boolean positional) { - /* - * find the first contact feature (if any) that does not lie - * entirely before the target range - */ - int startPosition = binarySearch(contactFeatureEnds, - SearchCriterion.byEnd(from)); - for (; startPosition < contactFeatureEnds.size(); startPosition++) + if (!positional) { - SequenceFeature sf = contactFeatureEnds.get(startPosition); - if (!sf.isContactFeature()) - { - System.err.println("Error! non-contact feature type " - + sf.getType() + " in contact features list"); - 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 - */ - continue; - } - - int end = sf.getEnd(); - if (end >= from && end <= to) - { - result.add(sf); - } - if (end > to) - { - break; - } + return nonPositionalFeatures == null ? 0 + : nonPositionalFeatures.size(); } - } - /** - * 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 result) - { - int startIndex = binarySearch(nonNestedFeatures, - SearchCriterion.byEnd(from)); + return (contactFeatureStarts == null ? 0 : contactFeatureStarts.size()) + + features.size(); - 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). + * 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 startIndex - * @param from - * @param to - * @param result + * @param positionalFeatures * @return */ - protected int findNonNestedFeatures(final int startIndex, long from, - long to, List result) + + @Override + public Set getFeatureGroups(boolean positionalFeatures) { - int i = startIndex; - while (i < nonNestedFeatures.size()) + if (positionalFeatures) { - 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++; + return Collections.unmodifiableSet(positionalFeatureGroups); + } + else + { + return nonPositionalFeatureGroups == null + ? Collections. emptySet() + : Collections.unmodifiableSet(nonPositionalFeatureGroups); } - return i; } - /** - * Adds contact features whose start position lies in the from-to range to the - * result list - * - * @param from - * @param to - * @param result - */ - protected void findContactStartFeatures(long from, long to, - List result) + @Override + public Collection getFeatures() { - int startPosition = binarySearch(contactFeatureStarts, - SearchCriterion.byStart(from)); - - for (; startPosition < contactFeatureStarts.size(); startPosition++) - { - SequenceFeature sf = contactFeatureStarts.get(startPosition); - if (!sf.isContactFeature()) - { - System.err.println("Error! non-contact feature type " - + sf.getType() + " in contact features list"); - continue; - } - int begin = sf.getBegin(); - if (begin >= from && begin <= to) - { - result.add(sf); - } - } + return features; } /** - * Answers a list of all positional features stored, in no guaranteed order + * 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 getPositionalFeatures() + + @Override + public List getFeaturesForGroup(boolean positional, + String group) { - /* - * add non-nested features (may be all features for many cases) - */ - List result = new ArrayList(); - result.addAll(nonNestedFeatures); + List result = new ArrayList<>(); /* - * add any contact features - from the list by start position + * if we know features don't include the target group, no need + * to inspect them for matches */ - if (contactFeatureStarts != null) + if (positional && !positionalFeatureGroups.contains(group) + || !positional && !nonPositionalFeatureGroups.contains(group)) { - result.addAll(contactFeatureStarts); + return result; } - /* - * add any nested features - */ - if (nestedFeatures != null) + if (positional) { - result.addAll(nestedFeatures.getEntries()); + addFeaturesForGroup(group, contactFeatureStarts, result); + addFeaturesForGroup(group, features, result); + } + else + { + addFeaturesForGroup(group, nonPositionalFeatures, result); } - return result; } /** - * Answers a list of all contact features. If there are none, returns an - * immutable empty list. + * 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 List getContactFeatures() + + @Override + public float getMaximumScore(boolean positional) { - if (contactFeatureStarts == null) - { - return Collections.emptyList(); - } - return new ArrayList(contactFeatureStarts); + return positional ? positionalMaxScore : nonPositionalMaxScore; } /** - * Answers a list of all non-positional features. If there are none, returns - * an immutable empty list. + * 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 */ + + @Override + public float getMinimumScore(boolean positional) + { + return positional ? positionalMinScore : nonPositionalMinScore; + } + + @Override public List getNonPositionalFeatures() { - if (nonPositionalFeatures == null) - { - return Collections.emptyList(); - } - return new ArrayList(nonPositionalFeatures); + return getNonPositionalFeatures(new ArrayList<>()); } /** - * 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 non-positional features. If there are none, returns + * an immutable empty list. * - * @param sf + * @return */ - public synchronized boolean delete(SequenceFeature sf) - { - /* - * try the non-nested positional features first - */ - boolean removed = nonNestedFeatures.remove(sf); - /* - * if not found, try contact positions (and if found, delete - * from both lists of contact positions) - */ - if (!removed && contactFeatureStarts != null) + @Override + public List getNonPositionalFeatures( + List result) + { + if (nonPositionalFeatures != null) { - removed = contactFeatureStarts.remove(sf); - if (removed) - { - contactFeatureEnds.remove(sf); - } + result.addAll(nonPositionalFeatures); } + return result; + } - boolean removedNonPositional = false; + @Override + public List getPositionalFeatures() + { + return getPositionalFeatures(new ArrayList<>()); + } + + /** + * Answers a list of all positional features stored, in no guaranteed order + * + * @return + */ + + @Override + public List getPositionalFeatures( + List result) + { /* - * if not found, try non-positional features + * add any contact features - from the list by start position */ - if (!removed && nonPositionalFeatures != null) + if (contactFeatureStarts != null) { - removedNonPositional = nonPositionalFeatures.remove(sf); - removed = removedNonPositional; + result.addAll(contactFeatureStarts); } /* - * if not found, try nested features + * add any nested features */ - if (!removed && nestedFeatures != null) - { - removed = nestedFeatures.delete(sf); - } - - if (removed) + if (features != null) { - rebuildFeatureGroups(sf.getFeatureGroup(), removedNonPositional); - // TODO and recalculate totalExtent (feature may have changed length!) + result.addAll(features); } - 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 - * - * @param featureGroup - * @param nonPositional - */ - protected void rebuildFeatureGroups(String featureGroup, - boolean nonPositional) - { - if (nonPositional && nonPositionalFeatures != null) - { - 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); - } - } - else if (!findFeatureGroup(featureGroup)) - { - positionalFeatureGroups.remove(featureGroup); - } + return result; } /** - * Scans all positional features to check whether the given feature group is - * found, and returns true if found, else false + * Answers the total length of positional features (or zero if there are + * none). Contact features contribute a value of 1 to the total. * - * @param featureGroup * @return */ - protected boolean findFeatureGroup(String featureGroup) + + @Override + public int getTotalFeatureLength() { - for (SequenceFeature sf : getPositionalFeatures()) - { - String group = sf.getFeatureGroup(); - if (group == featureGroup - || (group != null && group.equals(featureGroup))) - { - return true; - } - } - return false; + return totalExtent; } /** @@ -707,114 +687,116 @@ public class FeatureStore * * @return */ + + @Override public boolean isEmpty() { - boolean hasFeatures = !nonNestedFeatures.isEmpty() - || (contactFeatureStarts != null && !contactFeatureStarts - .isEmpty()) - || (nonPositionalFeatures != null && !nonPositionalFeatures - .isEmpty()) - || (nestedFeatures != null && nestedFeatures.size() > 0); + boolean hasFeatures = (contactFeatureStarts != null + && !contactFeatureStarts.isEmpty()) + || (nonPositionalFeatures != null + && !nonPositionalFeatures.isEmpty()) + || features.size() > 0; return !hasFeatures; } /** - * 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 Set getFeatureGroups(boolean positionalFeatures) - { - if (positionalFeatures) - { - return Collections.unmodifiableSet(positionalFeatureGroups); - } - else - { - return nonPositionalFeatureGroups == null ? Collections - . emptySet() : Collections - .unmodifiableSet(nonPositionalFeatureGroups); - } - } - - /** - * 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. - * - * @param features - * @param sc - * @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. */ - protected int binarySearch(List features, - SearchCriterion sc) + protected synchronized void rescanAfterDelete() { - int start = 0; - int end = features.size() - 1; - int matched = features.size(); - - while (start <= end) + 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) { - int mid = (start + end) / 2; - SequenceFeature entry = features.get(mid); - boolean compare = sc.compare(entry); - if (compare) + List list = nonPositionalFeatures; + for (int i = 0, n = list.size(); i < n; i++) { - matched = mid; - end = mid - 1; - } - else - { - start = mid + 1; + SequenceFeature sf = list.get(i); + nonPositionalFeatureGroups.add(sf.getFeatureGroup()); + float score = sf.getScore(); + nonPositionalMinScore = min(nonPositionalMinScore, score); + nonPositionalMaxScore = max(nonPositionalMaxScore, score); } } - return matched; + /* + * scan positional features for groups, scores and extents + */ + + rescanPositional(contactFeatureStarts); + rescanPositional(features); } - /** - * Answers the number of positional (or non-positional) features stored. - * Contact features count as 1. - * - * @param positional - * @return - */ - public int getFeatureCount(boolean positional) + private void rescanPositional(Collection sfs) { - if (!positional) - { - return nonPositionalFeatures == null ? 0 : nonPositionalFeatures - .size(); - } - - int size = nonNestedFeatures.size(); - - if (contactFeatureStarts != null) + if (sfs == null) { - // note a contact feature (start/end) counts as one - size += contactFeatureStarts.size(); + return; } - - if (nestedFeatures != null) + for (SequenceFeature sf : sfs) { - size += nestedFeatures.size(); + positionalFeatureGroups.add(sf.getFeatureGroup()); + float score = sf.getScore(); + positionalMinScore = min(positionalMinScore, score); + positionalMaxScore = max(positionalMaxScore, score); + totalExtent += getFeatureLength(sf); } - - 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. + * 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 int getTotalFeatureLength() + + @Override + public synchronized boolean shiftFeatures(int fromPosition, int shiftBy) { - return totalExtent; + /* + * 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; + List list = getPositionalFeatures(); + for (int i = 0, n = list.size(); i < n; i++) + { + SequenceFeature sf = list.get(i); + if (sf.getBegin() >= fromPosition) + { + 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 modified; } + }