+ * Routines for approximate Sequence Id resolution by name using string
+ * containment (on word boundaries) rather than equivalence. It also attempts to
+ * resolve ties where no exact match is available by picking the the id closest
+ * to the query.
+ */
+public class SequenceIdMatcher
+{
+ private HashMap<SeqIdName, List<SequenceI>> names;
+
+ private Map<SeqIdName, List<SequenceI>> excludes;
+
+ public SequenceIdMatcher(List<SequenceI> seqs)
+ {
+ names = new HashMap<>();
+ excludes = new HashMap<>();
+ addAll(seqs);
+ }
+
+ /**
+ * Adds sequences to this matcher
+ *
+ * @param seqs
+ */
+ public void addAll(List<SequenceI> seqs)
+ {
+ for (SequenceI seq : seqs)
+ {
+ add(seq);
+ }
+ }
+
+ /**
+ * Adds one sequence to this matcher
+ *
+ * @param seq
+ */
+ public void add(SequenceI seq)
+ {
+ SeqIdName key = new SeqIdName(seq.getDisplayId(true));
+ addMatchCandidate(key, seq);
+ SequenceI dbseq = seq;
+ while (dbseq.getDatasetSequence() != null)
+ {
+ dbseq = dbseq.getDatasetSequence();
+ }
+ // add in any interesting identifiers
+ if (dbseq.getDBRefs() != null)
+ {
+ DBRefEntry dbr[] = dbseq.getDBRefs();
+ for (int r = 0; r < dbr.length; r++)
+ {
+ DBRefEntry dbref = dbr[r];
+ SeqIdName sid = new SeqIdName(dbref.getAccessionId());
+ if (dbref.getMap() != null
+ && dbref.getMap().getMap().isTripletMap())
+ {
+ /*
+ * dbref with 3:1 or 1:3 mapping (e.g. CDS/protein);
+ * mark as not a valid match for this id
+ */
+ List<SequenceI> excluded = excludes.get(sid);
+ if (excluded == null)
+ {
+ excludes.put(sid, excluded = new ArrayList<>());
+ }
+ excluded.add(seq);
+ System.out.println("Excluding " + sid + "->" + seq);
+ continue;
+ }
+ addMatchCandidate(sid, seq);
+ }
+ }
+ }
+
+ void addMatchCandidate(SeqIdName key, SequenceI seq)
+ {
+ List<SequenceI> namesList = names.get(key);
+ if (namesList == null)
+ {
+ names.put(key, namesList = new ArrayList<>());
+ }
+ if (!namesList.contains(seq))
+ {
+ namesList.add(seq);
+ System.out.println("Adding " + key + "->" + seq);
+ }
+ }
+
+ /**
+ * convenience method to make a matcher from concrete array
+ *
+ * @param sequences
+ */
+ public SequenceIdMatcher(SequenceI[] sequences)
+ {
+ this(Arrays.asList(sequences));
+ }
+
+ /**
+ * returns the closest SequenceI in matches to SeqIdName and returns all the
+ * matches to the names hash.
+ *
+ * @param candName
+ * SeqIdName
+ * @param matches
+ * List of SequenceI objects
+ * @return SequenceI closest SequenceI to SeqIdName
+ */
+ private SequenceI pickbestMatch(SeqIdName candName,
+ List<SequenceI> matches)
+ {
+ List<SequenceI> st = pickbestMatches(candName, matches);
+ return st == null || st.size() == 0 ? null : st.get(0);
+ }
+
+ /**
+ * returns the closest SequenceI in matches to SeqIdName and returns all the
+ * matches to the names hash.
+ *
+ * @param candName
+ * SeqIdName
+ * @param matches
+ * Vector of SequenceI objects
+ * @return Object[] { SequenceI closest SequenceI to SeqIdName, SequenceI[]
+ * ties }
+ */
+ private List<SequenceI> pickbestMatches(SeqIdName candName,
+ List<SequenceI> matches)
+ {
+ List<SequenceI> best = new ArrayList<>();
+ if (candName == null || matches == null || matches.size() == 0)
+ {
+ return null;
+ }
+ SequenceI match = matches.remove(0);
+ best.add(match);
+ addMatchCandidate(new SeqIdName(match.getName()), match);
+ int matchlen = match.getName().length();
+ int namlen = candName.id.length();
+ while (matches.size() > 0)
+ {
+ // look through for a better one.
+ SequenceI cand = matches.remove(0);
+ addMatchCandidate(new SeqIdName(cand.getName()), cand);
+ int q, w, candlen = cand.getName().length();
+ // keep the one with an id 'closer' to the given seqnam string
+ if ((q = Math.abs(matchlen - namlen)) > (w = Math
+ .abs(candlen - namlen)) && candlen > matchlen)
+ {
+ best.clear();
+ match = cand;
+ matchlen = candlen;
+ best.add(match);
+ }
+ if (q == w && candlen == matchlen)
+ {
+ // record any ties
+ best.add(cand);
+ }
+ }
+ if (best.size() == 0)
+ {
+ return null;
+ }
+ ;
+ return best;
+ }
+
+ /**
+ * get SequenceI with closest SequenceI.getName() to seq.getName()
+ *
+ * @param seq
+ * SequenceI
+ * @return SequenceI
+ */
+ public SequenceI findIdMatch(SequenceI seq)
+ {
+ SeqIdName nam = new SeqIdName(seq.getName());
+ return findIdMatch(nam);
+ }
+
+ public SequenceI findIdMatch(String seqnam)
+ {
+ SeqIdName nam = new SeqIdName(seqnam);
+ return findIdMatch(nam);
+ }
+
+ /**
+ * Find all matches for a given sequence name.
+ *
+ * @param seqnam
+ * string to query Matcher with.
+ * @return a new array or (possibly) null
+ */
+ public SequenceI[] findAllIdMatches(String seqnam)
+ {
+
+ SeqIdName nam = new SeqIdName(seqnam);
+ List<SequenceI> m = findAllIdMatches(nam);
+ if (m != null)
+ {
+ return m.toArray(new SequenceI[m.size()]);
+ }
+ return null;
+ }
+
+ /**
+ * findIdMatch
+ *
+ * Return pointers to sequences (or sequence object containers) which have
+ * same Id as a given set of different sequence objects
+ *
+ * @param seqs
+ * SequenceI[]
+ * @return SequenceI[]
+ */
+ public SequenceI[] findIdMatch(SequenceI[] seqs)
+ {
+ SequenceI[] namedseqs = null;
+ int i = 0;
+ SeqIdName nam;
+
+ if (seqs.length > 0)
+ {
+ namedseqs = new SequenceI[seqs.length];
+ do
+ {
+ nam = new SeqIdName(seqs[i].getName());
+
+ if (names.containsKey(nam))
+ {
+ namedseqs[i] = findIdMatch(nam);
+ }
+ else
+ {
+ namedseqs[i] = null;
+ }
+ } while (++i < seqs.length);
+ }
+
+ return namedseqs;
+ }
+
+ /**
+ * core findIdMatch search method
+ *
+ * @param nam
+ * SeqIdName
+ * @return SequenceI
+ */
+ private SequenceI findIdMatch(SeqIdName nam)
+ {
+ List<SequenceI> matches = new ArrayList<>();
+ while (names.containsKey(nam))
+ {
+ List<SequenceI> candidates = names.remove(nam);
+ List<SequenceI> except = excludes.get(nam);
+ int j = candidates.size();
+ for (int i = 0; i < j; i++)
+ {
+ SequenceI candidate = candidates.get(i);
+ if (!except.contains(candidate))
+ {
+ matches.add(candidate);
+ }
+ }
+ }
+ return pickbestMatch(nam, matches);
+ }
+
+ /**
+ * core findIdMatch search method for finding all equivalent matches
+ *
+ * @param nam
+ * SeqIdName
+ * @return SequenceI[]
+ */
+ private List<SequenceI> findAllIdMatches(
+ jalview.analysis.SequenceIdMatcher.SeqIdName nam)
+ {
+ List<SequenceI> matches = new ArrayList<>();
+ while (names.containsKey(nam))
+ {
+ matches.addAll(names.remove(nam));
+ }
+ List<SequenceI> r = pickbestMatches(nam, matches);
+ return r;
+ }
+
+ class SeqIdName
+ {
+ String id;
+
+ SeqIdName(String s)
+ {
+ if (s != null)
+ {
+ id = s.toLowerCase();
+ }
+ else
+ {
+ id = "";
+ }
+ }
+
+ @Override
+ public int hashCode()
+ {
+ return ((id.length() >= 4) ? id.substring(0, 4).hashCode()
+ : id.hashCode());
+ }
+
+ @Override
+ public boolean equals(Object s)
+ {
+ if (s == null)
+ {
+ return false;
+ }
+ if (s instanceof SeqIdName)
+ {
+ return this.stringequals(((SeqIdName) s).id);
+ }
+ else
+ {
+ if (s instanceof String)
+ {
+ return this.stringequals(((String) s).toLowerCase());
+ }
+ }
+
+ return false;
+ }
+