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.analysis;
23 import jalview.datamodel.DBRefEntry;
24 import jalview.datamodel.SequenceI;
26 import java.util.ArrayList;
27 import java.util.Arrays;
28 import java.util.HashMap;
29 import java.util.HashSet;
30 import java.util.List;
34 * Routines for approximate Sequence Id resolution by name using string
35 * containment (on word boundaries) rather than equivalence. It also attempts to
36 * resolve ties where no exact match is available by picking the the id closest
39 public class SequenceIdMatcher
42 * weak hash for each sequence
44 private HashMap<SeqIdName, Set<SequenceI>> names;
47 // * cache of values removed for each query string.
49 // private HashMap<String, List<SequenceI>> resolved;
52 * do we index sequences on all 'words' in ID string ?
54 private boolean wordBased = false;
57 * Characters that define the end of a unique sequence ID at the beginning of
58 * an arbitrary ID string JBPNote: This is a heuristic that will fail for
59 * arbritrarily extended sequence id's (like portions of an aligned set of
60 * repeats from one sequence)
62 private static String WORD_SEP = "~. |#\\/<>!\"" + ((char) 0x00A4)
66 * @return true if matcher is word-based (ie string key matches one of the
67 * words within the body of one or more sequence IDs)
69 public boolean isWordBased()
75 * Construct a standard (non-word based) matcher. To configure word based
76 * matching, use the fully qualified constructor
80 public SequenceIdMatcher(List<SequenceI> seqs)
86 * construct a new matcher for a set of sequences, configured as required.
87 * Note: enabling word based matching
89 * @param wordBasedMatch
90 * - when true, "myseq" matches "X|myseq" and "myseq"
93 public SequenceIdMatcher(boolean wordBasedMatch,
96 wordBased = wordBasedMatch;
97 names = new HashMap<SeqIdName, Set<SequenceI>>();
102 * add more sequences to this matcher - also used by the constructor
106 public void addAll(List<SequenceI> seqs)
108 for (SequenceI seq : seqs)
114 private void addSeqIdName(SeqIdName idname, SequenceI seq)
116 Set<SequenceI> seqset = names.get(idname);
119 seqset = new HashSet<SequenceI>();
120 names.put(idname, seqset);
125 public void addSeq(SequenceI seq)
127 // TODO: deal with ID collisions - SequenceI should be appended to list
128 // associated with this key.
129 addSeqIdName(new SeqIdName(seq.getDisplayId(true)), seq);
132 for (SeqIdName key : getWordsFor(seq))
134 addSeqIdName(key, seq);
137 SequenceI dbseq = seq;
138 // TODO add test for database xref resolution
139 while (dbseq.getDatasetSequence() != null)
141 dbseq = dbseq.getDatasetSequence();
143 // add in any interesting identifiers
144 if (dbseq.getDBRefs() != null)
146 DBRefEntry dbr[] = dbseq.getDBRefs();
147 SeqIdName sid = null;
148 for (int r = 0; r < dbr.length; r++)
150 sid = new SeqIdName(dbr[r].getAccessionId());
151 if (!names.containsKey(sid))
153 addSeqIdName(sid, seq);
161 * generate word based keys for the given sequence
164 * @return list of split keys
166 public List<SeqIdName> getWordsFor(SequenceI seq)
168 ArrayList<SeqIdName> keys = new ArrayList<SeqIdName>();
169 String name = seq.getName(), limits = "/" + seq.getStart() + "-"
171 int namel = name.length();
172 char[] sep = new char[WORD_SEP.length()];
173 // find only the separators present in the ID.
174 for (int i = 0; i < sep.length; i++)
176 sep[i] = WORD_SEP.charAt(i);
177 if (seq.getName().indexOf("" + sep[i]) == -1)
184 for (int i = 0; i < sep.length; i++)
189 while ((m = name.indexOf(sep[i], p)) > p)
192 if (m > 0 && m - p > 5)
194 // split to end of word m with this delimiter
195 keys.add(new SeqIdName(name.substring(p, m - 1) + limits));
199 // index word after this delimiter m
200 keys.add(new SeqIdName(name.substring(m + 1) + limits));
206 // index word after this delimiter m
207 keys.add(new SeqIdName(name.substring(p) + limits));
215 * convenience method to make a matcher from concrete array Note: in order to
216 * support word based matching, use the fully qualified constructor
220 public SequenceIdMatcher(SequenceI[] sequences)
222 this(Arrays.asList(sequences));
226 * returns the closest SequenceI in matches to SeqIdName and returns all the
227 * matches to the names hash.
232 * List of SequenceI objects
233 * @return SequenceI closest SequenceI to SeqIdName
235 private SequenceI pickbestMatch(SeqIdName candName,
236 List<SequenceI> matches)
238 List<SequenceI> st = pickbestMatches(candName, matches);
239 return st == null || st.size() == 0 ? null : st.get(0);
243 * returns the closest SequenceI in matches to SeqIdName and returns all the
244 * matches to the names hash.
249 * Vector of SequenceI objects
250 * @return Object[] { SequenceI closest SequenceI to SeqIdName, SequenceI[]
253 private List<SequenceI> pickbestMatches(SeqIdName candName,
254 List<SequenceI> matches)
256 ArrayList<SequenceI> best = new ArrayList<SequenceI>();
257 if (candName == null || matches == null || matches.size() == 0)
261 SequenceI match = matches.remove(0);
264 int matchlen = match.getName().length();
265 int namlen = candName.id.length();
266 while (matches.size() > 0)
268 // look through for a better one.
269 SequenceI cand = matches.remove(0);
271 int q, w, candlen = cand.getName().length();
272 // keep the one with an id 'closer' to the given seqnam string
273 boolean is_closer = ((q = Math.abs(matchlen - namlen)) > (w = Math
274 .abs(candlen - namlen)) && candlen > matchlen);
275 // if not closer, then check if current best is actually identical in case
278 if (is_closer || (!candName.equalsCase(best.get(0).getName())))
285 if (q == w && candlen == matchlen)
287 // equivalently good, and matches with case as well. so
292 if (best.size() == 0)
301 * get SequenceI with closest SequenceI.getName() to seq.getName()
307 public SequenceI findIdMatch(SequenceI seq)
309 SeqIdName nam = new SeqIdName(seq.getName());
310 return findIdMatch(nam);
313 public SequenceI findIdMatch(String seqnam)
315 SeqIdName nam = new SeqIdName(seqnam);
316 return findIdMatch(nam);
320 * Find all matches for a given sequence name.
323 * string to query Matcher with.
324 * @return a new array or (possibly) null
326 public SequenceI[] findAllIdMatches(String seqnam)
329 SeqIdName nam = new SeqIdName(seqnam);
330 List<SequenceI> m = findAllIdMatches(nam);
333 return m.toArray(new SequenceI[m.size()]);
341 * Return pointers to sequences (or sequence object containers) which have
342 * same Id as a given set of different sequence objects
346 * @return SequenceI[]
348 public SequenceI[] findIdMatch(SequenceI[] seqs)
350 SequenceI[] namedseqs = null;
356 namedseqs = new SequenceI[seqs.length];
359 nam = new SeqIdName(seqs[i].getName());
361 if (names.containsKey(nam))
363 namedseqs[i] = findIdMatch(nam);
369 } while (++i < seqs.length);
376 * core findIdMatch search method
382 private SequenceI findIdMatch(
383 jalview.analysis.SequenceIdMatcher.SeqIdName nam)
385 ArrayList<SequenceI> matches = new ArrayList<SequenceI>();
386 while (names.containsKey(nam))
388 matches.addAll(names.remove(nam));
390 return pickbestMatch(nam, matches);
394 * core findIdMatch search method for finding all equivalent matches
398 * @return SequenceI[]
400 private List<SequenceI> findAllIdMatches(
401 jalview.analysis.SequenceIdMatcher.SeqIdName nam)
403 ArrayList<SequenceI> matches = new ArrayList<SequenceI>();
404 while (names.containsKey(nam))
406 matches.addAll(names.remove(nam));
408 List<SequenceI> r = pickbestMatches(nam, matches);
412 private class SeqIdName
420 id = new String(s).toLowerCase();
421 origid = new String(s);
430 public int hashCode()
432 return ((id.length() >= 4) ? id.substring(0, 4).hashCode() : id
437 public boolean equals(Object s)
443 if (s instanceof SeqIdName)
445 return this.equals((SeqIdName) s);
449 if (s instanceof String)
451 return this.equals((String) s);
459 * matches if one ID properly contains another at a whitespace boundary.
460 * TODO: (JBPNote) These are not efficient. should use char[] for speed
461 * todo: (JBPNote) Set separator characters appropriately
467 public boolean equals(SeqIdName s)
469 // TODO: JAL-732 patch for cases when name includes a list of IDs, and the
470 // match contains one ID flanked
471 if (id.length() > s.id.length())
473 return check_wordmatch(id, s.id);
477 return check_wordmatch(s.id, id);
481 private boolean check_wordmatch(String longer, String shorter)
483 boolean elen = longer.length() == shorter.length();
484 int sp = longer.indexOf(shorter);
492 // end of match is word boundary
493 return elen ? true : (WORD_SEP.indexOf(longer.charAt(shorter
494 .length() + sp)) > -1);
496 if (WORD_SEP.indexOf(longer.charAt(sp - 1)) > -1)
498 if (sp + shorter.length() == longer.length())
504 // end of match is word boundary
506 : sp + shorter.length() == longer.length() ? true
507 : (WORD_SEP.indexOf(longer.charAt(shorter
508 .length() + sp)) > -1);
513 // prefix of match is not a word boundary
518 public boolean equals(String s)
520 s = s.toLowerCase(); // TODO: employ faster to lower case operation
521 if (id.length() > s.length())
523 return check_wordmatch(id, s);
527 return check_wordmatch(s, id);
532 public String toString()
537 public boolean equalsCase(String s)
539 if (origid.length() > s.length())
541 return check_wordmatch(origid, s);
545 return check_wordmatch(s, origid);
549 public boolean equalsCase(SeqIdName sid)
551 return equalsCase(sid.origid);