X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Futil%2FDBRefUtils.java;h=e6aa47246536da0f8e6a6e950c08df3ff2264a5e;hb=37de9310bec3501cbc6381e0c3dcb282fcaad812;hp=b8f1dd566d9f0f68df83bc0fc501490b06a8d971;hpb=9623cea766a766683243235557ad48e6f7659e6a;p=jalview.git diff --git a/src/jalview/util/DBRefUtils.java b/src/jalview/util/DBRefUtils.java index b8f1dd5..e6aa472 100755 --- a/src/jalview/util/DBRefUtils.java +++ b/src/jalview/util/DBRefUtils.java @@ -24,18 +24,25 @@ import jalview.datamodel.DBRefEntry; import jalview.datamodel.DBRefSource; import jalview.datamodel.PDBEntry; import jalview.datamodel.SequenceI; -import jalview.ws.seqfetcher.DbSourceProxy; import java.util.ArrayList; +import java.util.Arrays; import java.util.HashMap; -import java.util.Hashtable; +import java.util.HashSet; import java.util.List; import java.util.Map; +import java.util.Set; 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 canonicalSourceNameLookup = new HashMap(); private static Map dasCoordinateSystemsLookup = new HashMap(); @@ -46,7 +53,25 @@ public class DBRefUtils 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); + + // 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); @@ -55,39 +80,36 @@ public class DBRefUtils } /** - * 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 srcs = new HashMap(); - ArrayList res = new ArrayList(); - - for (int i = 0; i < sources.length; i++) + HashSet srcs = new HashSet(); + 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 res = new ArrayList(); + 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); } } @@ -96,8 +118,6 @@ public class DBRefUtils DBRefEntry[] reply = new DBRefEntry[res.size()]; return res.toArray(reply); } - res = null; - // there are probable memory leaks in the hashtable! return null; } @@ -144,8 +164,8 @@ 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
    • @@ -158,15 +178,35 @@ 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 entry, according - * to the given comparator. Returns null if no matches. + * 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 refs + * Set of references to search + * @param accId + * accession id to match + * @return + */ + public static List 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 @@ -175,14 +215,14 @@ 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])) @@ -190,7 +230,7 @@ public class DBRefUtils rfs.add(refs[i]); } } - return rfs.size() == 0 ? null : rfs.toArray(new DBRefEntry[rfs.size()]); + return rfs; } interface DbRefComp @@ -201,12 +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())) @@ -231,27 +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; } @@ -261,18 +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) @@ -292,17 +328,18 @@ 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())) { @@ -329,12 +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) @@ -363,25 +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) { @@ -393,7 +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; } } @@ -403,6 +443,23 @@ public class DBRefUtils }; /** + * 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. *

    @@ -450,9 +507,7 @@ 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 @@ -474,34 +529,207 @@ public class DBRefUtils } /** - * Extract valid accession strings from a query string. Used by the - * SequenceFetcher and DBRefFetcher to create valid accession strings from an - * ID string for database sources with a Regex validation field. + * 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 proxy - * @param q - * @return q if proxy.getAccessionValidator()==null, otherwise the matched - * region or the first subgroup match from the matched region + * @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 String processQueryToAccessionFor(DbSourceProxy proxy, - String q) + public static DBRefEntry[] selectDbRefs(boolean selectDna, + DBRefEntry[] refs) { - if (proxy.getAccessionValidator() != null) + 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) { - Regex vgr = proxy.getAccessionValidator(); - vgr.search(q); - if (vgr.numSubs() > 0) + for (DBRefEntry dbref : dbRefs) { - return (vgr.stringMatched(1)); + if (source.equalsIgnoreCase(dbref.getSource())) + { + matches.add(dbref); + } } - else + } + 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) { - return (vgr.stringMatched()); + // nothing to do + return; } + selfs.addAll(Arrays.asList(selfArray)); } - else + + // filter non-primary refs + for (DBRefEntry p : pr) { - return (q); + 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()); + } + } + } } }