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.Arrays;
27 import java.util.Comparator;
28 import java.util.List;
31 * An adaption of FeatureStore that is efficient and lightweight, accelerating
32 * processing speed in JavaScript.
35 * @author Bob Hanson 2019.08.03
38 public class FeatureStoreJS extends FeatureStore
40 boolean contactsTainted = true;
45 public FeatureStoreJS()
47 features = new ArrayList<>();
51 * Returns a (possibly empty) list of features whose extent overlaps the given
52 * range. The returned list is not ordered. Contact features are included if
53 * either of the contact points lies within the range.
56 * start position of overlap range (inclusive)
58 * end position of overlap range (inclusive)
63 public List<SequenceFeature> findOverlappingFeatures(long start, long end,
64 List<SequenceFeature> result)
68 result = new ArrayList<>();
70 if (contactFeatureStarts != null)
74 findContactPoints(contactFeatureStarts, start, result, true);
75 findContactPoints(contactFeatureEnds, start, result, false);
79 findContactFeatures(start, end, result);
82 if (features.size() > 0)
84 findOverlaps(start, end, result);
89 // The following methods use a linked list of containment in SequenceFeature
90 // rather than IntervalStore.
92 // There are two parts --- initialization, and overlap searching.
94 // Initialization involves two steps:
96 // (1) sorting of features by start position using a standard Array.sort with
98 // (2) linking of features, effectively nesting them.
100 // Searching involves three steps:
102 // (1) binary search for a starting point within the sorted features array.
103 // (2) traverse the linked lists with an end check to read out the
104 // overlapped features at this position.
105 // (3) For an interval, find the last feature that starts in this interval,
106 // and add all features up through that feature.
108 // All of this is done with very simple standard methods.
113 * contact features ordered by first contact position
115 private SequenceFeature[] orderedFeatureStarts;
117 private void rebuildArrays(int n)
119 if (startComp == null)
121 startComp = new StartComparator();
123 orderedFeatureStarts = new SequenceFeature[n];
124 for (int i = n; --i >= 0;)
126 SequenceFeature sf = ((ArrayList<SequenceFeature>) features).get(i);
127 sf.index = i; // for debugging only
128 orderedFeatureStarts[i] = sf;
130 Arrays.sort(orderedFeatureStarts, startComp);
131 linkFeatures(orderedFeatureStarts);
135 * just a standard Comparator
137 private static StartComparator startComp;
139 class StartComparator implements Comparator<SequenceFeature>
143 public int compare(SequenceFeature o1, SequenceFeature o2)
147 return (p1 < p2 ? -1 : p1 > p2 ? 1 : 0);
153 * Run through the sorted sequence array once, building the containedBy linked
154 * list references. Does a check first to make sure there is actually
155 * something out there that is overlapping. A null for sf.containedBy means
156 * there are no overlaps for this feature.
160 private void linkFeatures(SequenceFeature[] intervals)
162 if (intervals.length < 2)
166 int maxEnd = intervals[0].end;
167 for (int i = 1, n = intervals.length; i < n; i++)
169 SequenceFeature sf = intervals[i];
170 if (sf.begin <= maxEnd)
172 sf.containedBy = getContainedBy(intervals[i - 1], sf);
182 * Since we are traversing the sorted feature array in a forward direction,
183 * all elements prior to the one we are working on have been fully linked. All
184 * we are doing is following those links until we find the first array feature
185 * with a containedBy element that has an end >= our begin point. It is
186 * generally a very short list -- maybe one or two depths. But it might be
193 private SequenceFeature getContainedBy(SequenceFeature sf,
196 int begin = sf0.begin;
201 // System.out.println("\nFS found " + sf0.index + ":" + sf0
202 // + "\nFS in " + sf.index + ":" + sf);
210 // search-stage methods
213 * Binary search for contact start or end at a given (Overview) position.
220 * @author Bob Hanson 2019.07.30
222 private static void findContactPoints(List<SequenceFeature> l, long pos,
223 List<SequenceFeature> result, boolean isStart)
226 int high = l.size() - 1;
229 int mid = (low + high) >>> 1;
230 SequenceFeature f = l.get(mid);
231 switch (Long.signum((isStart ? f.begin : f.end) - pos))
242 // could be "5" in 12345556788 ?
243 while (++mid <= high && (f = l.get(mid)) != null
244 && (isStart ? f.begin : f.end) == pos)
248 while (--m >= low && (f = l.get(m)) != null
249 && (isStart ? f.begin : f.end) == pos)
259 * Find all overlaps; special case when there is only one feature. The
260 * required array of start-sorted SequenceFeature is created lazily.
266 private void findOverlaps(long start, long end,
267 List<SequenceFeature> result)
269 int n = features.size();
275 checkOne(((ArrayList<SequenceFeature>) features).get(0), start, end,
279 if (orderedFeatureStarts == null)
286 // (1) Find the closest feature to this position.
288 int index = findClosestFeature(orderedFeatureStarts, start);
289 SequenceFeature sf = (index < 0 ? null : orderedFeatureStarts[index]);
291 // (2) Traverse the containedBy field, checking for overlap.
302 // (3) For an interval, find the last feature that starts in this interval,
303 // and add all features up through that feature.
307 // fill in with all features that start within this interval, fully
309 int index2 = findClosestFeature(orderedFeatureStarts, end);
310 while (++index <= index2)
312 result.add(orderedFeatureStarts[index]);
319 * Quick check when we only have one feature.
326 private void checkOne(SequenceFeature sf, long start, long end,
327 List<SequenceFeature> result)
329 if (sf.begin <= end && sf.end >= start)
337 * A binary search identical to the one used for contact start/end, but here
338 * we return the feature itself. Unlike Collection.BinarySearch, all we have
339 * to be is close, not exact, and we make sure if there is a string of
340 * identical starts, then we slide to the end so that we can check all of
347 private int findClosestFeature(SequenceFeature[] l, long pos)
350 int high = l.length - 1;
353 int mid = (low + high) >>> 1;
354 SequenceFeature f = l[mid];
355 switch (Long.signum(f.begin - pos))
365 while (++mid <= high && l[mid].begin == pos)
372 return (high < 0 ? -1 : high);
376 * Adds contact features to the result list where either the second or the
377 * first contact position lies within the target range
384 protected void findContactFeatures(long from, long to,
385 List<SequenceFeature> result)
387 findContactStartOverlaps(from, to, result);
388 findContactEndOverlaps(from, to, result);
392 * Adds to the result list any contact features whose end (second contact
393 * point), but not start (first contact point), lies in the query from-to
401 private void findContactEndOverlaps(long from, long to,
402 List<SequenceFeature> result)
404 // find the first contact feature (if any)
405 // with end point not before the target range
407 for (int i = findFirstEnd(contactFeatureEnds,
408 from), n = contactFeatureEnds.size(); i < n; i++)
410 SequenceFeature sf = contactFeatureEnds.get(i);
411 if (sf.begin >= from && sf.begin <= to)
413 // this feature's first contact position lies in the search range
414 // so we don't include it in results a second time
420 // this feature (and all following) has end point after the target range
424 // feature has end >= from and end <= to
425 // i.e. contact end point lies within overlap search range
431 * Adds contact features whose start position lies in the from-to range to the
439 private void findContactStartOverlaps(long from, long to,
440 List<SequenceFeature> result)
442 for (int i = findFirstBegin(contactFeatureStarts,
443 from), n = contactFeatureStarts.size(); i < n; i++)
445 SequenceFeature sf = contactFeatureStarts.get(i);
454 public static int findFirstBegin(List<SequenceFeature> list, long pos)
457 int end = list.size() - 1;
458 int matched = list.size();
462 int mid = (start + end) / 2;
463 if (list.get(mid).begin >= pos)
476 public static int findFirstEnd(List<SequenceFeature> list, long pos)
479 int end = list.size() - 1;
480 int matched = list.size();
484 int mid = (start + end) / 2;
485 if (list.get(mid).end >= pos)