X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Futil%2FDBRefUtils.java;h=4d5a0250b24c403591ae80f5b6b6e6bddb1924f0;hb=c17981672620e0b780a2338bd0c74e55cf9ddec2;hp=424d40bd45f4f83b975e0d0e2eead6682ee19109;hpb=964f27f06912574ac9d0c65c3f741852f48e558a;p=jalview.git diff --git a/src/jalview/util/DBRefUtils.java b/src/jalview/util/DBRefUtils.java index 424d40b..4d5a025 100755 --- a/src/jalview/util/DBRefUtils.java +++ b/src/jalview/util/DBRefUtils.java @@ -26,11 +26,12 @@ import jalview.datamodel.PDBEntry; import jalview.datamodel.SequenceI; import java.util.ArrayList; +import java.util.Arrays; import java.util.HashMap; import java.util.HashSet; -import java.util.Hashtable; import java.util.List; import java.util.Map; +import java.util.Set; import com.stevesoft.pat.Regex; @@ -42,9 +43,8 @@ public class DBRefUtils /* * lookup from lower-case form of a name to its canonical (standardised) form */ - private static Map canonicalSourceNameLookup = new HashMap(); + private static Map canonicalSourceNameLookup = new HashMap<>(); - private static Map dasCoordinateSystemsLookup = new HashMap(); static { @@ -59,19 +59,30 @@ public class DBRefUtils 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); + + // Make sure we have lowercase entries for all canonical string lookups + Set keys = canonicalSourceNameLookup.keySet(); + for (String k : keys) + { + canonicalSourceNameLookup.put(k.toLowerCase(), + canonicalSourceNameLookup.get(k)); + } - dasCoordinateSystemsLookup.put("pdbresnum", DBRefSource.PDB); - dasCoordinateSystemsLookup.put("uniprot", DBRefSource.UNIPROT); - dasCoordinateSystemsLookup.put("embl", DBRefSource.EMBL); - // dasCoordinateSystemsLookup.put("embl", DBRefSource.EMBLCDS); } /** + * 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 - * array of DBRef objects to search + * DBRefEntry objects to search * @param sources - * String[] array of source DBRef IDs to retrieve + * array of sources to select * @return */ public static DBRefEntry[] selectRefs(DBRefEntry[] dbrefs, @@ -81,17 +92,17 @@ public class DBRefUtils { return dbrefs; } - HashSet srcs = new HashSet(); + HashSet srcs = new HashSet<>(); for (String src : sources) { - srcs.add(src); + srcs.add(src.toUpperCase()); } - List res = new ArrayList(); + List res = new ArrayList<>(); for (DBRefEntry dbr : dbrefs) { String source = getCanonicalName(dbr.getSource()); - if (srcs.contains(source)) + if (srcs.contains(source.toUpperCase())) { res.add(dbr); } @@ -106,29 +117,6 @@ public class DBRefUtils } /** - * isDasCoordinateSystem - * - * @param string - * String - * @param dBRefEntry - * DBRefEntry - * @return boolean true if Source DBRefEntry is compatible with DAS - * CoordinateSystem name - */ - - public static boolean isDasCoordinateSystem(String string, - DBRefEntry dBRefEntry) - { - if (string == null || dBRefEntry == null) - { - return false; - } - String coordsys = dasCoordinateSystemsLookup.get(string.toLowerCase()); - return coordsys == null ? false : coordsys.equals(dBRefEntry - .getSource()); - } - - /** * look up source in an internal list of database reference sources and return * the canonical jalview name for the source, or the original string if it has * no canonical form. @@ -148,8 +136,8 @@ public class DBRefUtils } /** - * 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 *
    *
  • database sources are the same
    • *
  • accession ids are the same
    • @@ -162,34 +150,35 @@ public class DBRefUtils * pattern to match * @return */ - public static DBRefEntry[] searchRefs(DBRefEntry[] ref, DBRefEntry entry) + public static List searchRefs(DBRefEntry[] ref, + DBRefEntry entry) { return searchRefs(ref, entry, matchDbAndIdAndEitherMapOrEquivalentMapList); } /** - * Returns an array of those references that match the given accession id + * Returns a list of those references that match the given accession id *
      *
    • database sources are the same
      • *
    • accession ids are the same
      • *
    • both have no mapping, or the mappings are the same
      • *
    * - * @param ref + * @param refs * Set of references to search - * @param entry - * pattern to match + * @param accId + * accession id to match * @return */ - public static DBRefEntry[] searchRefs(DBRefEntry[] ref, String accId) + public static List searchRefs(DBRefEntry[] refs, String accId) { - return searchRefs(ref, new DBRefEntry("", "", accId), matchId); + return searchRefs(refs, new DBRefEntry("", "", accId), matchId); } /** - * Returns an array of those references that match the given entry, according - * to the given comparator. Returns null if no matches. + * 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 @@ -198,14 +187,14 @@ public class DBRefUtils * @param comparator * @return */ - static DBRefEntry[] searchRefs(DBRefEntry[] refs, DBRefEntry entry, + static List searchRefs(DBRefEntry[] refs, DBRefEntry entry, DbRefComp comparator) { + List rfs = new ArrayList<>(); if (refs == null || entry == null) { - return null; + return rfs; } - List rfs = new ArrayList(); for (int i = 0; i < refs.length; i++) { if (comparator.matches(entry, refs[i])) @@ -213,7 +202,7 @@ public class DBRefUtils rfs.add(refs[i]); } } - return rfs.size() == 0 ? null : rfs.toArray(new DBRefEntry[rfs.size()]); + return rfs; } interface DbRefComp @@ -231,7 +220,8 @@ public class DBRefUtils 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())) @@ -239,9 +229,8 @@ public class DBRefUtils if (refa.getAccessionId() == null || refb.getAccessionId().equals(refa.getAccessionId())) { - if (refa.getMap() == null - || (refb.getMap() != null && refb.getMap().equals( - refa.getMap()))) + if (refa.getMap() == null || (refb.getMap() != null + && refb.getMap().equals(refa.getMap()))) { return true; } @@ -262,7 +251,7 @@ public class DBRefUtils @Override public boolean matches(DBRefEntry refa, DBRefEntry refb) { - if (nullOrEqual(refa.getSource(), refb.getSource()) + if (nullOrEqualSource(refa.getSource(), refb.getSource()) && nullOrEqual(refa.getVersion(), refb.getVersion()) && nullOrEqual(refa.getAccessionId(), refb.getAccessionId()) && nullOrEqual(refa.getMap(), refb.getMap())) @@ -284,16 +273,17 @@ public class DBRefUtils public boolean matches(DBRefEntry refa, DBRefEntry refb) { if (refa.getSource() != null && refb.getSource() != null - && refb.getSource().equals(refa.getSource())) + && DBRefUtils.getCanonicalName(refb.getSource()).equals( + DBRefUtils.getCanonicalName(refa.getSource()))) { // We dont care about version if (refa.getAccessionId() != null && refb.getAccessionId() != null - // FIXME should be && not || here? + // 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()))) + || (refa.getMap() != null && refb.getMap() != null + && refb.getMap().equals(refa.getMap()))) { return true; } @@ -315,7 +305,8 @@ public class DBRefUtils public boolean matches(DBRefEntry refa, DBRefEntry refb) { if (refa.getSource() != null && refb.getSource() != null - && refb.getSource().equals(refa.getSource())) + && DBRefUtils.getCanonicalName(refb.getSource()).equals( + DBRefUtils.getCanonicalName(refa.getSource()))) { // We dont care about version if (refa.getAccessionId() != null && refb.getAccessionId() != null @@ -324,11 +315,12 @@ public class DBRefUtils if ((refa.getMap() == null && refb.getMap() == null) || (refa.getMap() != null && refb.getMap() != null)) { - if ((refb.getMap().getMap() == null && refa.getMap().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()))) + && refa.getMap().getMap() != null + && refb.getMap().getMap().getInverse() + .equals(refa.getMap().getMap()))) { return true; } @@ -351,7 +343,8 @@ public class DBRefUtils public boolean matches(DBRefEntry refa, DBRefEntry refb) { if (refa.getSource() != null && refb.getSource() != null - && refb.getSource().equals(refa.getSource())) + && DBRefUtils.getCanonicalName(refb.getSource()).equals( + DBRefUtils.getCanonicalName(refa.getSource()))) { // We dont care about version // if ((refa.getVersion()==null || refb.getVersion()==null) @@ -364,12 +357,13 @@ public class DBRefUtils { 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())))) + 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; } @@ -380,9 +374,9 @@ 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() { @@ -390,23 +384,26 @@ public class DBRefUtils public boolean matches(DBRefEntry refa, DBRefEntry refb) { if (refa.getSource() != null && refb.getSource() != null - && refb.getSource().equals(refa.getSource())) + && 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.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() != null - && refa.getMap().getMap() != null && (refb - .getMap().getMap().equals(refa.getMap().getMap())))) - { // getMap().getMap().containsEither(false,refa.getMap().getMap()) + && refa.getMap().getMap() != null + && (refb.getMap().getMap() + .equals(refa.getMap().getMap())))) + { return true; } } @@ -480,9 +477,7 @@ public class DBRefUtils PDBEntry pdbr = new PDBEntry(); pdbr.setId(pdbid); pdbr.setType(PDBEntry.Type.PDB); - pdbr.setProperty(new Hashtable()); pdbr.setChainCode(chaincode); - // pdbr.getProperty().put("CHAIN", chaincode); seq.addPDBId(pdbr); } else @@ -493,7 +488,7 @@ public class DBRefUtils else { // default: - ref = new DBRefEntry(locsrc, version, acn); + ref = new DBRefEntry(locsrc, version, acn.trim()); } } if (ref != null) @@ -516,7 +511,195 @@ public class DBRefUtils { return true; } - return (o1 == null ? o2.equals(o1) : o1.equals(o2)); + 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 TypePrimary DBDirect which will be promoted
    peptidesEnsemblUniprot
    peptidesEnsemblUniprot
    dnaEnsemblENA
    + * + * @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()); + } + } + } + } } }