();
+ if (refs == null || entry == null)
+ {
+ return rfs;
+ }
+ for (int i = 0, n = refs.size(); i < n; i++)
{
- return dbrefs;
+ DBRefEntry e = refs.get(i);
+ if (comparator.matches(entry, e, SEARCH_MODE_FULL))
+ {
+ rfs.add(e);
+ }
}
- Hashtable srcs = new Hashtable();
- Vector res = new Vector();
+ return rfs;
+ }
+
+ interface DbRefComp
+ {
+ default public boolean matches(DBRefEntry refa, DBRefEntry refb) {
+ return matches(refa, refb, SEARCH_MODE_FULL);
+ };
- for (int i = 0; i < sources.length; i++)
+ public boolean matches(DBRefEntry refa, DBRefEntry refb, int mode);
+}
+
+ /**
+ * match on all non-null fields in refa
+ */
+ // TODO unused - remove?
+ public static DbRefComp matchNonNullonA = new DbRefComp()
+ {
+ @Override
+ public boolean matches(DBRefEntry refa, DBRefEntry refb, int mode)
{
- srcs.put(new String(sources[i]), new Integer(i));
+ if ((mode & DB_SOURCE) != 0 && refa.getSource() == null
+ || DBRefUtils.getCanonicalName(refb.getSource()).equals(
+ DBRefUtils.getCanonicalName(refa.getSource())))
+ {
+ if ((mode & DB_VERSION) != 0 && refa.getVersion() == null
+ || refb.getVersion().equals(refa.getVersion()))
+ {
+ if ((mode & DB_ID) != 0 && refa.getAccessionId() == null
+ || refb.getAccessionId().equals(refa.getAccessionId()))
+ {
+ if ((mode & DB_MAP) != 0 && refa.getMap() == null || (refb.getMap() != null
+ && refb.getMap().equals(refa.getMap())))
+ {
+ return true;
+ }
+ }
+ }
+ }
+ return false;
}
- for (int i = 0, j = dbrefs.length; i < j; i++)
+ };
+
+ /**
+ * either field is null or field matches for all of source, version, accession
+ * id and map.
+ */
+ // TODO unused - remove?
+ public static DbRefComp matchEitherNonNull = new DbRefComp()
+ {
+ @Override
+ public boolean matches(DBRefEntry refa, DBRefEntry refb, int mode)
{
- if (srcs.containsKey(dbrefs[i].getSource()))
+ if (nullOrEqualSource(refa.getSource(), refb.getSource())
+ && nullOrEqual(refa.getVersion(), refb.getVersion())
+ && nullOrEqual(refa.getAccessionId(), refb.getAccessionId())
+ && nullOrEqual(refa.getMap(), refb.getMap()))
{
- res.add(dbrefs[i]);
+ return true;
}
+ return false;
}
- if (res.size() > 0)
+ };
+
+ /**
+ * accession ID and DB must be identical. Version is ignored. Map is either
+ * not defined or is a match (or is compatible?)
+ */
+ // TODO unused - remove?
+ public static DbRefComp matchDbAndIdAndEitherMap = new DbRefComp()
+ {
+ @Override
+ public boolean matches(DBRefEntry refa, DBRefEntry refb, int mode)
{
- DBRefEntry[] reply = new DBRefEntry[res.size()];
- for (int i = 0; i < res.size(); i++)
+ if (refa.getSource() != null && refb.getSource() != null
+ && DBRefUtils.getCanonicalName(refb.getSource()).equals(
+ DBRefUtils.getCanonicalName(refa.getSource())))
{
- reply[i] = (DBRefEntry) res.elementAt(i);
+ // We dont care about version
+ if (refa.getAccessionId() != null && refb.getAccessionId() != null
+ // FIXME should be && not || here?
+ || refb.getAccessionId().equals(refa.getAccessionId()))
+ {
+ if ((refa.getMap() == null || refb.getMap() == null)
+ || (refa.getMap() != null && refb.getMap() != null
+ && refb.getMap().equals(refa.getMap())))
+ {
+ return true;
+ }
+ }
}
- return reply;
+ return false;
}
- res = null;
- // there are probable memory leaks in the hashtable!
- return null;
+ };
+
+ /**
+ * accession ID and DB must be identical. Version is ignored. No map on either
+ * or map but no maplist on either or maplist of map on a is the complement of
+ * maplist of map on b.
+ */
+ // TODO unused - remove?
+ public static DbRefComp matchDbAndIdAndComplementaryMapList = new DbRefComp()
+ {
+ @Override
+ public boolean matches(DBRefEntry refa, DBRefEntry refb, int mode)
+ {
+ if (refa.getSource() != null && refb.getSource() != null
+ && DBRefUtils.getCanonicalName(refb.getSource()).equals(
+ DBRefUtils.getCanonicalName(refa.getSource())))
+ {
+ // We dont care about version
+ if (refa.getAccessionId() != null && refb.getAccessionId() != null
+ || refb.getAccessionId().equals(refa.getAccessionId()))
+ {
+ if ((refa.getMap() == null && refb.getMap() == null)
+ || (refa.getMap() != null && refb.getMap() != null))
+ {
+ if ((refb.getMap().getMap() == null
+ && refa.getMap().getMap() == null)
+ || (refb.getMap().getMap() != null
+ && refa.getMap().getMap() != null
+ && refb.getMap().getMap().getInverse()
+ .equals(refa.getMap().getMap())))
+ {
+ return true;
+ }
+ }
+ }
+ }
+ return false;
+ }
+ };
+
+ /**
+ * accession ID and DB must be identical. Version is ignored. No map on both
+ * or or map but no maplist on either or maplist of map on a is equivalent to
+ * the maplist of map on b.
+ */
+ // TODO unused - remove?
+ public static DbRefComp matchDbAndIdAndEquivalentMapList = new DbRefComp()
+ {
+ @Override
+ public boolean matches(DBRefEntry refa, DBRefEntry refb, int mode)
+ {
+ if (refa.getSource() != null && refb.getSource() != null
+ && DBRefUtils.getCanonicalName(refb.getSource()).equals(
+ DBRefUtils.getCanonicalName(refa.getSource())))
+ {
+ // We dont care about version
+ // if ((refa.getVersion()==null || refb.getVersion()==null)
+ // || refb.getVersion().equals(refa.getVersion()))
+ // {
+ if (refa.getAccessionId() != null && refb.getAccessionId() != null
+ || refb.getAccessionId().equals(refa.getAccessionId()))
+ {
+ if (refa.getMap() == null && refb.getMap() == null)
+ {
+ return true;
+ }
+ if (refa.getMap() != null && refb.getMap() != null
+ && ((refb.getMap().getMap() == null
+ && refa.getMap().getMap() == null)
+ || (refb.getMap().getMap() != null
+ && refa.getMap().getMap() != null
+ && refb.getMap().getMap()
+ .equals(refa.getMap().getMap()))))
+ {
+ return true;
+ }
+ }
+ }
+ return false;
+ }
+ };
+
+ /**
+ * accession ID and DB must be identical, or null on a. Version is ignored. No
+ * map on either or map but no maplist on either or maplist of map on a is
+ * equivalent to the maplist of map on b.
+ */
+ public static DbRefComp matchDbAndIdAndEitherMapOrEquivalentMapList = new DbRefComp()
+ {
+ @Override
+ public boolean matches(DBRefEntry refa, DBRefEntry refb, int mode)
+ {
+ if (refa.getSource() != null && refb.getSource() != null
+ && DBRefUtils.getCanonicalName(refb.getSource()).equals(
+ DBRefUtils.getCanonicalName(refa.getSource())))
+ {
+ // We dont care about version
+
+ if (refa.getAccessionId() == null
+ || refa.getAccessionId().equals(refb.getAccessionId()))
+ {
+ if (refa.getMap() == null || refb.getMap() == null)
+ {
+ return true;
+ }
+ if ((refa.getMap() != null && refb.getMap() != null)
+ && (refb.getMap().getMap() == null
+ && refa.getMap().getMap() == null)
+ || (refb.getMap().getMap() != null
+ && refa.getMap().getMap() != null
+ && (refb.getMap().getMap()
+ .equals(refa.getMap().getMap()))))
+ {
+ return true;
+ }
+ }
+ }
+ return false;
+ }
+ };
+
+ /**
+ * accession ID only must be identical.
+ */
+ public static DbRefComp matchId = new DbRefComp()
+ {
+ @Override
+ public boolean matches(DBRefEntry refa, DBRefEntry refb, int mode)
+ {
+ if (refa.getAccessionId() != null && refb.getAccessionId() != null
+ && refb.getAccessionId().equals(refa.getAccessionId()))
+ {
+ return true;
+ }
+ return false;
+ }
+ };
+
+ /**
+ * Parses a DBRefEntry and adds it to the sequence, also a PDBEntry if the
+ * database is PDB.
+ *
+ * Used by file parsers to generate DBRefs from annotation within file (eg
+ * Stockholm)
+ *
+ * @param dbname
+ * @param version
+ * @param acn
+ * @param seq
+ * where to annotate with reference
+ * @return parsed version of entry that was added to seq (if any)
+ */
+ public static DBRefEntry parseToDbRef(SequenceI seq, String dbname,
+ String version, String acn)
+ {
+ DBRefEntry ref = null;
+ if (dbname != null)
+ {
+ String locsrc = DBRefUtils.getCanonicalName(dbname);
+ if (locsrc.equals(DBRefSource.PDB))
+ {
+ /*
+ * Check for PFAM style stockhom PDB accession id citation e.g.
+ * "1WRI A; 7-80;"
+ */
+ Regex r = new com.stevesoft.pat.Regex(
+ "([0-9][0-9A-Za-z]{3})\\s*(.?)\\s*;\\s*([0-9]+)-([0-9]+)");
+ if (r.search(acn.trim()))
+ {
+ String pdbid = r.stringMatched(1);
+ String chaincode = r.stringMatched(2);
+ if (chaincode == null)
+ {
+ chaincode = " ";
+ }
+ // String mapstart = r.stringMatched(3);
+ // String mapend = r.stringMatched(4);
+ if (chaincode.equals(" "))
+ {
+ chaincode = "_";
+ }
+ // construct pdb ref.
+ ref = new DBRefEntry(locsrc, version, pdbid + chaincode);
+ PDBEntry pdbr = new PDBEntry();
+ pdbr.setId(pdbid);
+ pdbr.setType(PDBEntry.Type.PDB);
+ pdbr.setChainCode(chaincode);
+ seq.addPDBId(pdbr);
+ }
+ else
+ {
+ System.err.println("Malformed PDB DR line:" + acn);
+ }
+ }
+ else
+ {
+ // default:
+ ref = new DBRefEntry(locsrc, version, acn);
+ }
+ }
+ if (ref != null)
+ {
+ seq.addDBRef(ref);
+ }
+ return ref;
}
/**
- * isDasCoordinateSystem
- *
- * @param string String
- * @param dBRefEntry DBRefEntry
- * @return boolean true if Source DBRefEntry is compatible with DAS CoordinateSystem name
+ * Returns true if either object is null, or they are equal
+ *
+ * @param o1
+ * @param o2
+ * @return
*/
- public static Hashtable DasCoordinateSystemsLookup = null;
- public static boolean isDasCoordinateSystem(String string,
- DBRefEntry dBRefEntry)
+ public static boolean nullOrEqual(Object o1, Object o2)
{
- if (DasCoordinateSystemsLookup == null)
- { // Initialise
- DasCoordinateSystemsLookup = new Hashtable();
- DasCoordinateSystemsLookup.put("pdbresnum",
- jalview.datamodel.DBRefSource.PDB);
- DasCoordinateSystemsLookup.put("uniprot",
- jalview.datamodel.DBRefSource.UNIPROT);
+ if (o1 == null || o2 == null)
+ {
+ return true;
}
+ return o1.equals(o2);
+ }
+
+ /**
+ * canonicalise source string before comparing. null is always wildcard
+ *
+ * @param o1
+ * - null or source string to compare
+ * @param o2
+ * - null or source string to compare
+ * @return true if either o1 or o2 are null, or o1 equals o2 under
+ * DBRefUtils.getCanonicalName
+ * (o1).equals(DBRefUtils.getCanonicalName(o2))
+ */
+ public static boolean nullOrEqualSource(String o1, String o2)
+ {
+ if (o1 == null || o2 == null)
+ {
+ return true;
+ }
+ return DBRefUtils.getCanonicalName(o1)
+ .equals(DBRefUtils.getCanonicalName(o2));
+ }
+
+ /**
+ * Selects just the DNA or protein references from a set of references
+ *
+ * @param selectDna
+ * if true, select references to 'standard' DNA databases, else to
+ * 'standard' peptide databases
+ * @param refs
+ * a set of references to select from
+ * @return
+ */
+ public static List selectDbRefs(boolean selectDna,
+ List refs)
+ {
+ return selectRefs(refs,
+ selectDna ? DBRefSource.DNACODINGDBS : DBRefSource.PROTEINDBS);
+ // could attempt to find other cross
+ // refs here - ie PDB xrefs
+ // (not dna, not protein seq)
+ }
- String coordsys = (String) DasCoordinateSystemsLookup.get(string.
- toLowerCase());
- if (coordsys != null)
+ /**
+ * Returns the (possibly empty) list of those supplied dbrefs which have the
+ * specified source database, with a case-insensitive match of source name
+ *
+ * @param dbRefs
+ * @param source
+ * @return
+ */
+ public static List searchRefsForSource(List dbRefs,
+ String source)
+ {
+ List matches = new ArrayList();
+ if (dbRefs != null && source != null)
{
- return coordsys.equals(dBRefEntry.getSource());
+ for (DBRefEntry dbref : dbRefs)
+ {
+ if (source.equalsIgnoreCase(dbref.getSource()))
+ {
+ matches.add(dbref);
+ }
+ }
}
- return false;
+ return matches;
}
+
+ /**
+ * promote direct database references to primary for nucleotide or protein
+ * sequences if they have an appropriate primary ref
+ *
+ *
+ * | Seq Type |
+ * Primary DB |
+ * Direct which will be promoted |
+ *
+ *
+ * | peptides |
+ * Ensembl |
+ * Uniprot |
+ *
+ *
+ * | peptides |
+ * Ensembl |
+ * Uniprot |
+ *
+ *
+ * | dna |
+ * Ensembl |
+ * ENA |
+ *
+ *
+ *
+ * @param sequence
+ */
+ public static void ensurePrimaries(SequenceI sequence, List pr)
+ {
+ if (pr.size() == 0)
+ {
+ // nothing to do
+ return;
+ }
+ int sstart = sequence.getStart();
+ int send = sequence.getEnd();
+ boolean isProtein = sequence.isProtein();
+ BitSet bsSelect = new BitSet();
+
+// List selfs = new ArrayList();
+// {
+
+// List selddfs = selectDbRefs(!isprot, sequence.getDBRefs());
+// if (selfs == null || selfs.size() == 0)
+// {
+// // nothing to do
+// return;
+// }
+
+ List dbrefs = sequence.getDBRefs();
+ bsSelect.set(0, dbrefs.size());
+
+ if (!selectRefsBS(dbrefs, isProtein ? DBRefSource.PROTEINDBSKEYS : DBRefSource.DNACODINGDBSKEYS, bsSelect))
+ return;
+
+// selfs.addAll(selfArray);
+// }
+
+ // filter non-primary refs
+ for (int ip = pr.size(); --ip >= 0;)
+ {
+ DBRefEntry p = pr.get(ip);
+ for (int i = bsSelect.nextSetBit(0); i >= 0; i = bsSelect.nextSetBit(i + 1)) {
+ if (dbrefs.get(i) == p)
+ bsSelect.clear(i);
+ }
+// while (selfs.contains(p))
+// {
+// selfs.remove(p);
+// }
+ }
+// List toPromote = new ArrayList();
+
+
+
+ for (int ip = pr.size(), keys = 0; --ip >= 0 && keys != DBRefSource.ALL_MASKS;)
+ {
+ DBRefEntry p = pr.get(ip);
+ if (isProtein)
+ {
+ switch (getCanonicalName(p.getSource()))
+ {
+ case DBRefSource.UNIPROT:
+ keys |= DBRefSource.UNIPROT_MASK;
+ break;
+ case DBRefSource.ENSEMBL:
+ keys |= DBRefSource.ENSEMBL_MASK;
+ break;
+ }
+ }
+ else
+ {
+ // TODO: promote transcript refs ??
+ }
+ if (keys == 0 || !selectRefsBS(dbrefs, DBRefSource.PROMTYPES[keys], bsSelect))
+ return;
+// if (candidates != null)
+ {
+ for (int ic = bsSelect.nextSetBit(0); ic >= 0; ic = bsSelect.nextSetBit(ic + 1))
+// for (int ic = 0, n = candidates.size(); ic < n; ic++)
+ {
+ DBRefEntry cand = dbrefs.get(ic);//candidates.get(ic);
+ if (cand.hasMap())
+ {
+ Mapping map = cand.getMap();
+ SequenceI cto = map.getTo();
+ if (cto != null
+ && cto != sequence)
+ {
+ // can't promote refs with mappings to other sequences
+ continue;
+ }
+ MapList mlist = map.getMap();
+ if (mlist.getFromLowest() != sstart
+ && mlist.getFromHighest() != send)
+ {
+ // can't promote refs with mappings from a region of this sequence
+ // - eg CDS
+ continue;
+ }
+ }
+ // and promote
+ cand.setVersion(p.getVersion() + " (promoted)");
+ bsSelect.clear(ic);
+ //selfs.remove(cand);
+// toPromote.add(cand);
+ if (!cand.isPrimaryCandidate())
+ {
+ System.out.println(
+ "Warning: Couldn't promote dbref " + cand.toString()
+ + " for sequence " + sequence.toString());
+ }
+ }
+ }
+ }
+ }
+
}