();
+ if (refs == null || entry == null)
{
- if (comparator.matches(entry, ref[i]))
- {
- rfs.addElement(ref[i]);
- }
+ return rfs;
}
- // TODO Auto-generated method stub
- if (rfs.size() > 0)
+ for (int i = 0; i < refs.length; i++)
{
- DBRefEntry[] rf = new DBRefEntry[rfs.size()];
- rfs.copyInto(rf);
- return rf;
+ if (comparator.matches(entry, refs[i]))
+ {
+ rfs.add(refs[i]);
+ }
}
- return null;
+ return rfs;
}
- public interface DbRefComp
+ interface DbRefComp
{
public boolean matches(DBRefEntry refa, DBRefEntry refb);
}
@@ -194,12 +241,15 @@ public class DBRefUtils
/**
* 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)
{
if (refa.getSource() == null
- || refb.getSource().equals(refa.getSource()))
+ || DBRefUtils.getCanonicalName(refb.getSource()).equals(
+ DBRefUtils.getCanonicalName(refa.getSource())))
{
if (refa.getVersion() == null
|| refb.getVersion().equals(refa.getVersion()))
@@ -224,27 +274,18 @@ public class DBRefUtils
* 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)
{
- if ((refa.getSource() == null || refb.getSource() == null)
- || refb.getSource().equals(refa.getSource()))
+ if (nullOrEqualSource(refa.getSource(), refb.getSource())
+ && nullOrEqual(refa.getVersion(), refb.getVersion())
+ && nullOrEqual(refa.getAccessionId(), refb.getAccessionId())
+ && nullOrEqual(refa.getMap(), refb.getMap()))
{
- 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)
- || (refb.getMap() != null && refb.getMap().equals(
- refa.getMap())))
- {
- return true;
- }
- }
- }
+ return true;
}
return false;
}
@@ -254,18 +295,20 @@ public class DBRefUtils
* 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)
{
- if (refa.getSource() != null && refb.getSource() != null
- && refb.getSource().equals(refa.getSource()))
+ 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
+ // FIXME should be && not || here?
|| refb.getAccessionId().equals(refa.getAccessionId()))
{
if ((refa.getMap() == null || refb.getMap() == null)
@@ -285,22 +328,24 @@ public class DBRefUtils
* 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)
{
- if (refa.getSource() != null && refb.getSource() != null
- && refb.getSource().equals(refa.getSource()))
+ 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)
|| (refa.getMap() != null && refb.getMap() != null))
+ {
if ((refb.getMap().getMap() == null && refa.getMap().getMap() == null)
|| (refb.getMap().getMap() != null
&& refa.getMap().getMap() != null && refb
@@ -309,6 +354,7 @@ public class DBRefUtils
{
return true;
}
+ }
}
}
return false;
@@ -320,12 +366,16 @@ public class DBRefUtils
* 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)
{
- if (refa.getSource() != null && refb.getSource() != null
- && refb.getSource().equals(refa.getSource()))
+ 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)
@@ -354,25 +404,24 @@ public class DBRefUtils
};
/**
- * 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 equivalent to the
- * maplist of map on b.
+ * 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)
{
- // System.err.println("Comparing A: "+refa.getSrcAccString()+(refa.hasMap()?" has map.":"."));
- // System.err.println("Comparing B: "+refb.getSrcAccString()+(refb.hasMap()?" has map.":"."));
- if (refa.getSource() != null && refb.getSource() != null
- && refb.getSource().equals(refa.getSource()))
+ 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.getAccessionId() == null
+ || refa.getAccessionId().equals(refb.getAccessionId()))
{
if (refa.getMap() == null || refb.getMap() == null)
{
@@ -384,7 +433,7 @@ public class DBRefUtils
|| (refb.getMap().getMap() != null
&& refa.getMap().getMap() != null && (refb
.getMap().getMap().equals(refa.getMap().getMap()))))
- { // getMap().getMap().containsEither(false,refa.getMap().getMap())
+ {
return true;
}
}
@@ -394,14 +443,34 @@ public class DBRefUtils
};
/**
- * used by file parsers to generate DBRefs from annotation within file (eg
- * stockholm)
+ * accession ID only must be identical.
+ */
+ public static DbRefComp matchId = new DbRefComp()
+ {
+ @Override
+ public boolean matches(DBRefEntry refa, DBRefEntry refb)
+ {
+ 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 anotate with reference
+ * 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,
@@ -410,19 +479,25 @@ public class DBRefUtils
DBRefEntry ref = null;
if (dbname != null)
{
- String locsrc = jalview.util.DBRefUtils.getCanonicalName(dbname);
- if (locsrc.equals(jalview.datamodel.DBRefSource.PDB))
+ String locsrc = DBRefUtils.getCanonicalName(dbname);
+ if (locsrc.equals(DBRefSource.PDB))
{
- // check for chaincode and mapping
- // PFAM style stockhom PDB citation
- com.stevesoft.pat.Regex r = new com.stevesoft.pat.Regex(
- "([0-9][0-9A-Za-z]{3})\\s*(.?)\\s*;([0-9]+)-([0-9]+)");
+ /*
+ * 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);
- String mapstart = r.stringMatched(3);
- String mapend = r.stringMatched(4);
+ if (chaincode == null)
+ {
+ chaincode = " ";
+ }
+ // String mapstart = r.stringMatched(3);
+ // String mapend = r.stringMatched(4);
if (chaincode.equals(" "))
{
chaincode = "_";
@@ -431,8 +506,14 @@ public class DBRefUtils
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
{
@@ -447,4 +528,209 @@ public class DBRefUtils
return ref;
}
+ /**
+ * Returns true if either object is null, or they are equal
+ *
+ * @param o1
+ * @param o2
+ * @return
+ */
+ public static boolean nullOrEqual(Object o1, Object o2)
+ {
+ 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 DBRefEntry[] selectDbRefs(boolean selectDna,
+ DBRefEntry[] 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)
+ }
+
+ /**
+ * 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(DBRefEntry[] dbRefs,
+ String source)
+ {
+ List matches = new ArrayList();
+ if (dbRefs != null && source != null)
+ {
+ for (DBRefEntry dbref : dbRefs)
+ {
+ if (source.equalsIgnoreCase(dbref.getSource()))
+ {
+ matches.add(dbref);
+ }
+ }
+ }
+ 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 = sequence.getPrimaryDBRefs();
+ if (pr.size() == 0)
+ {
+ // nothing to do
+ return;
+ }
+ List selfs = new ArrayList();
+ {
+ DBRefEntry[] selfArray = selectDbRefs(!sequence.isProtein(),
+ sequence.getDBRefs());
+ if (selfArray == null || selfArray.length == 0)
+ {
+ // nothing to do
+ return;
+ }
+ selfs.addAll(Arrays.asList(selfArray));
+ }
+
+ // filter non-primary refs
+ for (DBRefEntry p : pr)
+ {
+ while (selfs.contains(p))
+ {
+ selfs.remove(p);
+ }
+ }
+ List toPromote = new ArrayList();
+
+ for (DBRefEntry p : pr)
+ {
+ List promType = new ArrayList();
+ if (sequence.isProtein())
+ {
+ switch (getCanonicalName(p.getSource()))
+ {
+ case DBRefSource.UNIPROT:
+ // case DBRefSource.UNIPROTKB:
+ // case DBRefSource.UP_NAME:
+ // search for and promote ensembl
+ promType.add(DBRefSource.ENSEMBL);
+ break;
+ case DBRefSource.ENSEMBL:
+ // search for and promote Uniprot
+ promType.add(DBRefSource.UNIPROT);
+ break;
+ }
+ }
+ else
+ {
+ // TODO: promote transcript refs
+ }
+
+ // collate candidates and promote them
+ DBRefEntry[] candidates = selectRefs(
+ selfs.toArray(new DBRefEntry[0]),
+ promType.toArray(new String[0]));
+ if (candidates != null)
+ {
+ for (DBRefEntry cand : candidates)
+ {
+ if (cand.hasMap())
+ {
+ if (cand.getMap().getTo() != null
+ && cand.getMap().getTo() != sequence)
+ {
+ // can't promote refs with mappings to other sequences
+ continue;
+ }
+ if (cand.getMap().getMap().getFromLowest() != sequence
+ .getStart()
+ && cand.getMap().getMap().getFromHighest() != sequence
+ .getEnd())
+ {
+ // can't promote refs with mappings from a region of this sequence
+ // - eg CDS
+ continue;
+ }
+ }
+ // and promote
+ cand.setVersion(p.getVersion() + " (promoted)");
+ selfs.remove(cand);
+ toPromote.add(cand);
+ if (!cand.isPrimaryCandidate())
+ {
+ System.out.println("Warning: Couldn't promote dbref "
+ + cand.toString() + " for sequence "
+ + sequence.toString());
+ }
+ }
+ }
+ }
+ }
+
}