- * <p>Title: </p>
- * SequenceIdMatcher
- * <p>Description: </p>
- * Routine which does approximate Sequence Id resolution by name using string containment rather than equivalence
- * <p>Copyright: Copyright (c) 2004</p>
- *
- * <p>Company: Dundee University</p>
- *
- * @author not attributable
- * @version 1.0
- */\r
-public class SequenceIdMatcher {\r
- private Hashtable names;\r
-\r
- public SequenceIdMatcher(SequenceI[] seqs) {\r
- names = new Hashtable();\r
-\r
- for (int i = 0; i < seqs.length; i++) {\r
- names.put(new SeqIdName(seqs[i].getName()), seqs[i]);\r
- }\r
- }\r
-\r
- SequenceI findIdMatch(SequenceI seq) {\r
- SeqIdName nam = new SeqIdName(seq.getName());\r
-\r
- if (names.containsKey(nam)) {\r
- return (SequenceI) names.get(nam);\r
- }\r
-\r
- return null;\r
- }\r
-\r
- SequenceI findIdMatch(String seqnam) {\r
- SeqIdName nam = new SeqIdName(seqnam);\r
-\r
- if (names.containsKey(nam)) {\r
- return (SequenceI) names.get(nam);\r
- }\r
-\r
- return null;\r
- }\r
-\r
+ * 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 Hashtable names;
+
+ public SequenceIdMatcher(SequenceI[] seqs)
+ {
+ names = new Hashtable();
+ for (int i = 0; i < seqs.length; i++)
+ {
+ // TODO: deal with ID collisions - SequenceI should be appended to list
+ // associated with this key.
+ names.put(new SeqIdName(seqs[i].getDisplayId(true)), seqs[i]);
+ // add in any interesting identifiers
+ if (seqs[i].getDBRef() != null)
+ {
+ DBRefEntry dbr[] = seqs[i].getDBRef();
+ SeqIdName sid = null;
+ for (int r = 0; r < dbr.length; r++)
+ {
+ sid = new SeqIdName(dbr[r].getAccessionId());
+ if (!names.contains(sid))
+ {
+ names.put(sid, seqs[i]);
+ }
+ }
+ }
+ }
+ }
+
+ /**
+ * 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 SequenceI closest SequenceI to SeqIdName
+ */
+ private SequenceI pickbestMatch(SeqIdName candName, Vector matches)
+ {
+ SequenceI[] st = pickbestMatches(candName, matches);
+ return st == null || st.length == 0 ? null : st[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 SequenceI[] pickbestMatches(SeqIdName candName, Vector matches)
+ {
+ ArrayList best = new ArrayList();
+ SequenceI match = null;
+ if (candName == null || matches == null || matches.size() == 0)
+ {
+ return null;
+ }
+ match = (SequenceI) matches.elementAt(0);
+ matches.removeElementAt(0);
+ best.add(match);
+ names.put(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 = (SequenceI) matches.elementAt(0);
+ matches.remove(0);
+ names.put(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 (SequenceI[]) best.toArray(new SequenceI[0]);
+ }
+
+ /**
+ * 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.
+ */
+ public SequenceI[] findAllIdMatches(String seqnam)
+ {
+
+ SeqIdName nam = new SeqIdName(seqnam);
+ return findAllIdMatches(nam);
+ }
+
+ /**
+ * 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(
+ jalview.analysis.SequenceIdMatcher.SeqIdName nam)
+ {
+ Vector matches = new Vector();
+ while (names.containsKey(nam))
+ {
+ matches.addElement(names.remove(nam));
+ }
+ return pickbestMatch(nam, matches);
+ }
+
+ /**
+ * core findIdMatch search method for finding all equivalent matches
+ *
+ * @param nam
+ * SeqIdName
+ * @return SequenceI[]
+ */
+ private SequenceI[] findAllIdMatches(
+ jalview.analysis.SequenceIdMatcher.SeqIdName nam)
+ {
+ Vector matches = new Vector();
+ while (names.containsKey(nam))
+ {
+ matches.addElement(names.remove(nam));
+ }
+ SequenceI[] r = pickbestMatches(nam, matches);
+ return r;
+ }
+
+ private class SeqIdName
+ {
+ String id;
+
+ SeqIdName(String s)
+ {
+ if (s != null)
+ {
+ id = new String(s);
+ }
+ else
+ {
+ id = "";
+ }
+ }
+
+ public int hashCode()
+ {
+ return ((id.length() >= 4) ? id.substring(0, 4).hashCode() : id
+ .hashCode());
+ }
+
+ public boolean equals(Object s)
+ {
+ if (s instanceof SeqIdName)
+ {
+ return this.equals((SeqIdName) s);
+ }
+ else
+ {
+ if (s instanceof String)
+ {
+ return this.equals((String) s);
+ }
+ }
+
+ return false;
+ }
+