X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;ds=inline;f=src%2Fjalview%2Fanalysis%2FSequenceIdMatcher.java;h=7c974e26873a4dc72bb95abd3f71100be68f9e6d;hb=d587f1aa61946dc14f6f089cf1dc2a3116cfb773;hp=1efe4982810ecb2a79b57a85834178374108926a;hpb=efc31b4a8d5cee63555586804a2b79c06bdb5a14;p=jalview.git diff --git a/src/jalview/analysis/SequenceIdMatcher.java b/src/jalview/analysis/SequenceIdMatcher.java index 1efe498..7c974e2 100755 --- a/src/jalview/analysis/SequenceIdMatcher.java +++ b/src/jalview/analysis/SequenceIdMatcher.java @@ -1,6 +1,6 @@ /* * Jalview - A Sequence Alignment Editor and Viewer - * Copyright (C) 2005 AM Waterhouse, J Procter, G Barton, M Clamp, S Searle + * Copyright (C) 2006 AM Waterhouse, J Procter, G Barton, M Clamp, S Searle * * This program is free software; you can redistribute it and/or * modify it under the terms of the GNU General Public License @@ -26,7 +26,10 @@ import jalview.datamodel.*; *
Title:
* SequenceIdMatcher *Description:
- * Routine which does approximate Sequence Id resolution by name using string containment rather than equivalence + * Routine which does 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. *Copyright: Copyright (c) 2004
* *Company: Dundee University
@@ -41,37 +44,57 @@ public class SequenceIdMatcher public SequenceIdMatcher(SequenceI[] seqs) { names = new Hashtable(); - for (int i = 0; i < seqs.length; i++) { names.put(new SeqIdName(seqs[i].getName()), 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 match=null; + if (candName==null || matches==null || matches.size()==0) + return null; + match=(SequenceI) matches.elementAt(0); + matches.removeElementAt(0); + 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); + names.put(new SeqIdName(cand.getName()), cand); + int candlen = cand.getName().length(); + // keep the one with an id 'closer' to the given seqnam string + if (Math.abs(matchlen-namlen)>Math.abs(candlen-namlen) && candlen>matchlen) { + match = cand; + matchlen = candlen; + } + } + return match; + } + /** + * get SequenceI with closest SequenceI.getName() to seq.getName() + * @param seq SequenceI + * @return SequenceI + */ SequenceI findIdMatch(SequenceI seq) { SeqIdName nam = new SeqIdName(seq.getName()); - - if (names.containsKey(nam)) - { - return (SequenceI) names.get(nam); - } - - return null; + return findIdMatch(nam); } SequenceI findIdMatch(String seqnam) - { - SeqIdName nam = new SeqIdName(seqnam); - - if (names.containsKey(nam)) { - return (SequenceI) names.get(nam); - } - - return null; + SeqIdName nam = new SeqIdName(seqnam); + return findIdMatch(nam); } - /** * findIdMatch * @@ -83,44 +106,63 @@ public class SequenceIdMatcher */ SequenceI[] findIdMatch(SequenceI[] seqs) { - SequenceI[] namedseqs = new SequenceI[seqs.length]; - + 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] = (SequenceI) names.get(nam); + namedseqs[i] = findIdMatch(nam); } else { namedseqs[i] = null; } } - while (i++ < seqs.length); + 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); + } + private class SeqIdName { String id; SeqIdName(String s) { - id = new String(s); + if (s!=null) + id = new String(s); + else + id = ""; } public int hashCode() { - return (id.substring(0, 4).hashCode()); + return ((id.length()>=4) ? id.substring(0, 4).hashCode() : id.hashCode()); } public boolean equals(Object s) @@ -140,24 +182,45 @@ public class SequenceIdMatcher return false; } + /** + * Characters that define the end of a unique sequence ID at + * the beginning of an arbitrary ID string + * JBPNote: This is a heuristic that will fail for arbritrarily extended sequence id's + * (like portions of an aligned set of repeats from one sequence) + */ + private String WORD_SEP="~. |#\\/<>!\"£$%^*)}[@',?_"; + + /** + * matches if one ID properly contains another at a whitespace boundary. + * TODO: (JBPNote) These are not efficient. should use char[] for speed + * todo: (JBPNote) Set separator characters appropriately + * @param s SeqIdName + * @return boolean + */ public boolean equals(SeqIdName s) { - if (id.startsWith(s.id) || s.id.startsWith(id)) - { - return true; - } - - return false; + if (id.length()>s.id.length()) { + return id.startsWith(s.id) ? + (WORD_SEP.indexOf(id.charAt(s.id.length()))>-1) + : false; + } else + return s.id.startsWith(id) ? + (s.id.equals(id) ? true : + (WORD_SEP.indexOf(s.id.charAt(id.length()))>-1)) + : false; } public boolean equals(String s) { - if (id.startsWith(s) || s.startsWith(id)) - { - return true; - } - - return false; + if (id.length()>s.length()) { + return id.startsWith(s) ? + (WORD_SEP.indexOf(id.charAt(s.length()))>-1) + : false; + } else + return s.startsWith(id) ? + (s.equals(id) ? true : + (WORD_SEP.indexOf(s.charAt(id.length()))>-1)) + : false; } } }