import java.util.ArrayList;
import java.util.HashMap;
+import java.util.HashSet;
import java.util.Hashtable;
import java.util.List;
import java.util.Map;
import com.stevesoft.pat.Regex;
+/**
+ * Utilities for handling DBRef objects and their collections.
+ */
public class DBRefUtils
{
+ /*
+ * lookup from lower-case form of a name to its canonical (standardised) form
+ */
private static Map<String, String> canonicalSourceNameLookup = new HashMap<String, String>();
private static Map<String, String> dasCoordinateSystemsLookup = new HashMap<String, String>();
canonicalSourceNameLookup.put("uniprotkb/swiss-prot",
DBRefSource.UNIPROT);
canonicalSourceNameLookup.put("uniprotkb/trembl", DBRefSource.UNIPROT);
+
+ // Ensembl values for dbname in xref REST service:
+ canonicalSourceNameLookup.put("uniprot/sptrembl", DBRefSource.UNIPROT);
+ canonicalSourceNameLookup.put("uniprot/swissprot", DBRefSource.UNIPROT);
+
canonicalSourceNameLookup.put("pdb", DBRefSource.PDB);
+ canonicalSourceNameLookup.put("ensembl", DBRefSource.ENSEMBL);
+ // Ensembl Gn and Tr are for Ensembl genomic and transcript IDs as served
+ // from ENA.
+ canonicalSourceNameLookup.put("ensembl-tr", DBRefSource.ENSEMBL);
+ canonicalSourceNameLookup.put("ensembl-gn", DBRefSource.ENSEMBL);
+
dasCoordinateSystemsLookup.put("pdbresnum", DBRefSource.PDB);
dasCoordinateSystemsLookup.put("uniprot", DBRefSource.UNIPROT);
}
/**
- * Utilities for handling DBRef objects and their collections.
- */
- /**
+ * Returns those DBRefEntry objects whose source identifier (once converted to
+ * Jalview's canonical form) is in the list of sources to search for. Returns
+ * null if no matches found.
*
* @param dbrefs
- * Vector of DBRef objects to search
+ * DBRefEntry objects to search
* @param sources
- * String[] array of source DBRef IDs to retrieve
- * @return Vector
+ * array of sources to select
+ * @return
*/
public static DBRefEntry[] selectRefs(DBRefEntry[] dbrefs,
String[] sources)
{
- if (dbrefs == null)
- {
- return null;
- }
- if (sources == null)
+ if (dbrefs == null || sources == null)
{
return dbrefs;
}
- Map<String, Integer> srcs = new HashMap<String, Integer>();
- ArrayList<DBRefEntry> res = new ArrayList<DBRefEntry>();
-
- for (int i = 0; i < sources.length; i++)
+ HashSet<String> srcs = new HashSet<String>();
+ for (String src : sources)
{
- srcs.put(new String(sources[i]), new Integer(i));
+ srcs.add(src);
}
- for (int i = 0, j = dbrefs.length; i < j; i++)
+
+ List<DBRefEntry> res = new ArrayList<DBRefEntry>();
+ for (DBRefEntry dbr : dbrefs)
{
- if (srcs.containsKey(dbrefs[i].getSource()))
+ String source = getCanonicalName(dbr.getSource());
+ if (srcs.contains(source))
{
- res.add(dbrefs[i]);
+ res.add(dbr);
}
}
DBRefEntry[] reply = new DBRefEntry[res.size()];
return res.toArray(reply);
}
- res = null;
- // there are probable memory leaks in the hashtable!
return null;
}
{
return false;
}
- String coordsys = dasCoordinateSystemsLookup.get(string
- .toLowerCase());
+ String coordsys = dasCoordinateSystemsLookup.get(string.toLowerCase());
return coordsys == null ? false : coordsys.equals(dBRefEntry
.getSource());
}
{
return null;
}
- String canonical = canonicalSourceNameLookup.get(source
- .toLowerCase());
+ String canonical = canonicalSourceNameLookup.get(source.toLowerCase());
return canonical == null ? source : canonical;
}
/**
- * Returns an array of those references that match the given entry, or null if
- * no matches. Currently uses a comparator which matches if
+ * Returns a (possibly empty) list of those references that match the given
+ * entry. Currently uses a comparator which matches if
* <ul>
* <li>database sources are the same</li>
* <li>accession ids are the same</li>
* pattern to match
* @return
*/
- public static DBRefEntry[] searchRefs(DBRefEntry[] ref, DBRefEntry entry)
+ public static List<DBRefEntry> searchRefs(DBRefEntry[] ref,
+ DBRefEntry entry)
{
return searchRefs(ref, entry,
matchDbAndIdAndEitherMapOrEquivalentMapList);
}
/**
- * Returns an array of those references that match the given entry, according
- * to the given comparator. Returns null if no matches.
+ * Returns a list of those references that match the given accession id
+ * <ul>
+ * <li>database sources are the same</li>
+ * <li>accession ids are the same</li>
+ * <li>both have no mapping, or the mappings are the same</li>
+ * </ul>
+ *
+ * @param refs
+ * Set of references to search
+ * @param accId
+ * accession id to match
+ * @return
+ */
+ public static List<DBRefEntry> searchRefs(DBRefEntry[] refs, String accId)
+ {
+ return searchRefs(refs, new DBRefEntry("", "", accId), matchId);
+ }
+
+ /**
+ * Returns a (possibly empty) list of those references that match the given
+ * entry, according to the given comparator.
*
* @param refs
* an array of database references to search
* @param comparator
* @return
*/
- static DBRefEntry[] searchRefs(DBRefEntry[] refs, DBRefEntry entry,
+ static List<DBRefEntry> searchRefs(DBRefEntry[] refs, DBRefEntry entry,
DbRefComp comparator)
{
+ List<DBRefEntry> rfs = new ArrayList<DBRefEntry>();
if (refs == null || entry == null)
{
- return null;
+ return rfs;
}
- List<DBRefEntry> rfs = new ArrayList<DBRefEntry>();
for (int i = 0; i < refs.length; i++)
{
if (comparator.matches(entry, refs[i]))
rfs.add(refs[i]);
}
}
- return rfs.size() == 0 ? null : rfs.toArray(new DBRefEntry[rfs.size()]);
+ return rfs;
}
interface DbRefComp
/**
* 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
* 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 (nullOrEqual(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;
}
* 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()))
{
// 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)
* 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()))
{
// 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()))
{
* 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
};
/**
- * 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()))
{
// 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)
{
|| (refb.getMap().getMap() != null
&& refa.getMap().getMap() != null && (refb
.getMap().getMap().equals(refa.getMap().getMap()))))
- { // getMap().getMap().containsEither(false,refa.getMap().getMap())
+ {
return true;
}
}
};
/**
+ * 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.
* <p>
{
String pdbid = r.stringMatched(1);
String chaincode = r.stringMatched(2);
- if (chaincode==null)
+ if (chaincode == null)
{
chaincode = " ";
}
pdbr.setChainCode(chaincode);
// pdbr.getProperty().put("CHAIN", chaincode);
seq.addPDBId(pdbr);
- } else {
- System.err.println("Malformed PDB DR line:"+acn);
+ }
+ else
+ {
+ System.err.println("Malformed PDB DR line:" + acn);
}
}
else
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 == null ? o2.equals(o1) : o1.equals(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<DBRefEntry> searchRefsForSource(DBRefEntry[] dbRefs,
+ String source)
+ {
+ List<DBRefEntry> matches = new ArrayList<DBRefEntry>();
+ if (dbRefs != null && source != null)
+ {
+ for (DBRefEntry dbref : dbRefs)
+ {
+ if (source.equalsIgnoreCase(dbref.getSource()))
+ {
+ matches.add(dbref);
+ }
+ }
+ }
+ return matches;
+ }
+
}
git://source.jalview.org / jalview.git / blobdiff