2 * Jalview - A Sequence Alignment Editor and Viewer ($$Version-Rel$$)
3 * Copyright (C) $$Year-Rel$$ The Jalview Authors
5 * This file is part of Jalview.
7 * Jalview is free software: you can redistribute it and/or
8 * modify it under the terms of the GNU General Public License
9 * as published by the Free Software Foundation, either version 3
10 * of the License, or (at your option) any later version.
12 * Jalview is distributed in the hope that it will be useful, but
13 * WITHOUT ANY WARRANTY; without even the implied warranty
14 * of MERCHANTABILITY or FITNESS FOR A PARTICULAR
15 * PURPOSE. See the GNU General Public License for more details.
17 * You should have received a copy of the GNU General Public License
18 * along with Jalview. If not, see <http://www.gnu.org/licenses/>.
19 * The Jalview Authors are detailed in the 'AUTHORS' file.
21 package jalview.datamodel.features;
23 import jalview.datamodel.SequenceFeature;
25 import java.util.ArrayList;
26 import java.util.List;
29 * An adaption of FeatureStore that is efficient and lightweight, accelerating
30 * processing speed in JavaScript.
33 * @author Bob Hanson 2019.08.03
36 public class FeatureStoreJS extends FeatureStore
38 boolean contactsTainted = true;
41 * internal reference to features as an ArrayList
43 private ArrayList<SequenceFeature> featureList;
48 public FeatureStoreJS()
50 features = featureList = new ArrayList<>();
54 * Add a contact feature to the lists that hold them ordered by start (first
55 * contact) and by end (second contact) position, ensuring the lists remain
56 * ordered, and returns true. This method allows duplicate features to be
57 * added, so test before calling to avoid this.
63 protected synchronized boolean addContactFeature(SequenceFeature feature)
65 if (contactFeatureStarts == null)
67 contactFeatureStarts = new ArrayList<>();
68 contactFeatureEnds = new ArrayList<>();
70 contactFeatureStarts.add(
71 findFirstBegin(contactFeatureStarts, feature.begin), feature);
72 contactFeatureEnds.add(findFirstEnd(contactFeatureEnds, feature.end),
78 * Adds one feature to the IntervalStore that can manage nested features
79 * (creating the IntervalStore if necessary)
82 protected synchronized void addNestedFeature(SequenceFeature feature)
84 featureList.add(findFirstBegin(featureList, feature.begin), feature);
88 * Returns a (possibly empty) list of features whose extent overlaps the given
89 * range. The returned list is not ordered. Contact features are included if
90 * either of the contact points lies within the range.
93 * start position of overlap range (inclusive)
95 * end position of overlap range (inclusive)
100 public List<SequenceFeature> findOverlappingFeatures(long start, long end,
101 List<SequenceFeature> result)
105 result = new ArrayList<>();
107 if (contactFeatureStarts != null)
111 findContactPoints(contactFeatureStarts, start, result, true);
112 findContactPoints(contactFeatureEnds, start, result, false);
116 findContactFeatures(start, end, result);
119 if (featureList.size() > 0)
121 findOverlaps(start, end, result);
126 // The following methods use a linked list of containment in SequenceFeature
127 // rather than IntervalStore.
129 // There are two parts --- initialization, and overlap searching.
131 // Initialization involves two steps:
133 // (1) sorting of features by start position using a standard Array.sort with
135 // (2) linking of features, effectively nesting them.
137 // Searching involves three steps:
139 // (1) binary search for a starting point within the sorted features array.
140 // (2) traverse the linked lists with an end check to read out the
141 // overlapped features at this position.
142 // (3) For an interval, find the last feature that starts in this interval,
143 // and add all features up through that feature.
145 // All of this is done with very simple standard methods.
150 * contact features ordered by first contact position
152 private SequenceFeature[] orderedFeatureStarts;
154 private void rebuildArrays(int n)
156 // Arrays.sort(orderedFeatureStarts, startComp);
157 orderedFeatureStarts = featureList
158 .toArray(new SequenceFeature[featureList.size()]);
159 linkFeatures(orderedFeatureStarts);
163 // * just a standard Comparator
165 // private static StartComparator startComp;
167 // class StartComparator implements Comparator<SequenceFeature>
171 // public int compare(SequenceFeature o1, SequenceFeature o2)
173 // int p1 = o1.begin;
174 // int p2 = o2.begin;
175 // return (p1 < p2 ? -1 : p1 > p2 ? 1 : 0);
181 * Run through the sorted sequence array once, building the containedBy linked
182 * list references. Does a check first to make sure there is actually
183 * something out there that is overlapping. A null for sf.containedBy means
184 * there are no overlaps for this feature.
188 private void linkFeatures(SequenceFeature[] intervals)
190 if (intervals.length < 2)
194 int maxEnd = intervals[0].end;
195 for (int i = 1, n = intervals.length; i < n; i++)
197 SequenceFeature sf = intervals[i];
198 if (sf.begin <= maxEnd)
200 sf.containedBy = getContainedBy(intervals[i - 1], sf);
210 * Since we are traversing the sorted feature array in a forward direction,
211 * all elements prior to the one we are working on have been fully linked. All
212 * we are doing is following those links until we find the first array feature
213 * with a containedBy element that has an end >= our begin point. It is
214 * generally a very short list -- maybe one or two depths. But it might be
221 private SequenceFeature getContainedBy(SequenceFeature sf,
224 int begin = sf0.begin;
229 // System.out.println("\nFS found " + sf0.index + ":" + sf0
230 // + "\nFS in " + sf.index + ":" + sf);
238 // search-stage methods
241 * Binary search for contact start or end at a given (Overview) position.
248 * @author Bob Hanson 2019.07.30
250 private static void findContactPoints(List<SequenceFeature> l, long pos,
251 List<SequenceFeature> result, boolean isStart)
254 int high = l.size() - 1;
257 int mid = (low + high) >>> 1;
258 SequenceFeature f = l.get(mid);
259 switch (Long.signum((isStart ? f.begin : f.end) - pos))
270 // could be "5" in 12345556788 ?
271 while (++mid <= high && (f = l.get(mid)) != null
272 && (isStart ? f.begin : f.end) == pos)
276 while (--m >= low && (f = l.get(m)) != null
277 && (isStart ? f.begin : f.end) == pos)
287 * Find all overlaps; special case when there is only one feature. The
288 * required array of start-sorted SequenceFeature is created lazily.
294 private void findOverlaps(long start, long end,
295 List<SequenceFeature> result)
297 int n = featureList.size();
303 checkOne(featureList.get(0), start, end,
307 if (orderedFeatureStarts == null)
314 // (1) Find the closest feature to this position.
316 int index = findClosestFeature(orderedFeatureStarts, start);
317 SequenceFeature sf = (index < 0 ? null : orderedFeatureStarts[index]);
319 // (2) Traverse the containedBy field, checking for overlap.
330 // (3) For an interval, find the last feature that starts in this interval,
331 // and add all features up through that feature.
335 // fill in with all features that start within this interval, fully
337 int index2 = findClosestFeature(orderedFeatureStarts, end);
338 while (++index <= index2)
340 result.add(orderedFeatureStarts[index]);
347 * Quick check when we only have one feature.
354 private void checkOne(SequenceFeature sf, long start, long end,
355 List<SequenceFeature> result)
357 if (sf.begin <= end && sf.end >= start)
365 protected boolean containsFeature(SequenceFeature feature)
368 int pos = findFirstBegin(featureList,
370 int len = featureList.size();
373 SequenceFeature sf = featureList.get(pos);
374 if (sf.begin > feature.begin)
376 return false; // no match found
378 if (sf.equals(feature))
389 * A binary search identical to the one used for contact start/end, but here
390 * we return the feature itself. Unlike Collection.BinarySearch, all we have
391 * to be is close, not exact, and we make sure if there is a string of
392 * identical starts, then we slide to the end so that we can check all of
399 private int findClosestFeature(SequenceFeature[] l, long pos)
402 int high = l.length - 1;
405 int mid = (low + high) >>> 1;
406 SequenceFeature f = l[mid];
407 switch (Long.signum(f.begin - pos))
417 while (++mid <= high && l[mid].begin == pos)
424 return (high < 0 ? -1 : high);
428 * Adds contact features to the result list where either the second or the
429 * first contact position lies within the target range
436 protected void findContactFeatures(long from, long to,
437 List<SequenceFeature> result)
439 findContactStartOverlaps(from, to, result);
440 findContactEndOverlaps(from, to, result);
444 * Adds to the result list any contact features whose end (second contact
445 * point), but not start (first contact point), lies in the query from-to
453 private void findContactEndOverlaps(long from, long to,
454 List<SequenceFeature> result)
456 // find the first contact feature (if any)
457 // with end point not before the target range
459 for (int i = findFirstEnd(contactFeatureEnds,
460 from), n = contactFeatureEnds.size(); i < n; i++)
462 SequenceFeature sf = contactFeatureEnds.get(i);
463 if (sf.begin >= from && sf.begin <= to)
465 // this feature's first contact position lies in the search range
466 // so we don't include it in results a second time
472 // this feature (and all following) has end point after the target range
476 // feature has end >= from and end <= to
477 // i.e. contact end point lies within overlap search range
483 * Adds contact features whose start position lies in the from-to range to the
491 private void findContactStartOverlaps(long from, long to,
492 List<SequenceFeature> result)
494 for (int i = findFirstBegin(contactFeatureStarts,
495 from), n = contactFeatureStarts.size(); i < n; i++)
497 SequenceFeature sf = contactFeatureStarts.get(i);
507 protected int findFirstBegin(List<SequenceFeature> list, long pos)
510 int end = list.size() - 1;
511 int matched = list.size();
515 int mid = (start + end) / 2;
516 if (list.get(mid).begin >= pos)
530 protected int findFirstEnd(List<SequenceFeature> list, long pos)
533 int end = list.size() - 1;
534 int matched = list.size();
538 int mid = (start + end) / 2;
539 if (list.get(mid).end >= pos)