*/
package jalview.ws;
+import java.util.Locale;
+
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.Enumeration;
+import java.util.HashMap;
+import java.util.Hashtable;
+import java.util.List;
+import java.util.Map;
+import java.util.StringTokenizer;
+import java.util.Vector;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
+
import jalview.analysis.AlignSeq;
+import jalview.api.FeatureSettingsModelI;
import jalview.bin.Cache;
import jalview.datamodel.AlignmentI;
import jalview.datamodel.DBRefEntry;
import jalview.datamodel.DBRefSource;
import jalview.datamodel.Mapping;
-import jalview.datamodel.SequenceFeature;
import jalview.datamodel.SequenceI;
-import jalview.gui.AlignFrame;
import jalview.gui.CutAndPasteTransfer;
import jalview.gui.Desktop;
+import jalview.gui.FeatureSettings;
import jalview.gui.IProgressIndicator;
import jalview.gui.OOMWarning;
+import jalview.util.DBRefUtils;
import jalview.util.MessageManager;
-import jalview.ws.dbsources.das.api.jalviewSourceI;
import jalview.ws.seqfetcher.DbSourceProxy;
-
-import java.util.ArrayList;
-import java.util.Enumeration;
-import java.util.Hashtable;
-import java.util.List;
-import java.util.StringTokenizer;
-import java.util.Vector;
-
import uk.ac.ebi.picr.model.UPEntry;
+import uk.ac.ebi.www.picr.AccessionMappingService.AccessionMapperServiceLocator;
/**
* Implements a runnable for validating a sequence against external databases
*/
public class DBRefFetcher implements Runnable
{
- SequenceI[] dataset;
+ private static final String NEWLINE = System.lineSeparator();
- IProgressIndicator af;
+ public static final String TRIM_RETRIEVED_SEQUENCES = "TRIM_FETCHED_DATASET_SEQS";
- CutAndPasteTransfer output = new CutAndPasteTransfer();
+ public interface FetchFinishedListenerI
+ {
+ void finished();
+ }
- StringBuffer sbuffer = new StringBuffer();
+ SequenceI[] dataset;
- boolean running = false;
+ IProgressIndicator progressWindow;
+
+ CutAndPasteTransfer output = new CutAndPasteTransfer();
/**
* picr client instance
*/
uk.ac.ebi.www.picr.AccessionMappingService.AccessionMapperInterface picrClient = null;
- // /This will be a collection of Vectors of sequenceI refs.
+ // This will be a collection of Vectors of sequenceI refs.
// The key will be the seq name or accession id of the seq
- Hashtable seqRefs;
+ Hashtable<String, Vector<SequenceI>> seqRefs;
DbSourceProxy[] dbSources;
SequenceFetcher sfetcher;
+ private List<FetchFinishedListenerI> listeners;
+
private SequenceI[] alseqs;
- /**
+ /*
* when true - retrieved sequences will be trimmed to cover longest derived
* alignment sequence
*/
private boolean trimDsSeqs = true;
- public DBRefFetcher()
- {
- }
-
- /**
- * Creates a new SequenceFeatureFetcher object and fetches from the currently
- * selected set of databases.
- *
- * @param seqs
- * fetch references for these sequences
- * @param af
- * the parent alignframe for progress bar monitoring.
- */
- public DBRefFetcher(SequenceI[] seqs, AlignFrame af)
- {
- this(seqs, af, null);
- }
-
/**
- * Creates a new SequenceFeatureFetcher object and fetches from the currently
- * selected set of databases.
+ * Creates a new DBRefFetcher object and fetches from the currently selected
+ * set of databases, if this is null then it fetches based on feature settings
*
* @param seqs
- * fetch references for these sequences
- * @param af
- * the parent alignframe for progress bar monitoring.
+ * fetch references for these SequenceI array
+ * @param progressIndicatorFrame
+ * the frame for progress bar monitoring
* @param sources
- * array of database source strings to query references from
+ * array of DbSourceProxy to query references form
+ * @param featureSettings
+ * FeatureSettings to get alternative DbSourceProxy from
+ * @param isNucleotide
+ * indicates if the array of SequenceI are Nucleotides or not
*/
- public DBRefFetcher(SequenceI[] seqs, AlignFrame af,
- DbSourceProxy[] sources)
+ public DBRefFetcher(SequenceI[] seqs,
+ IProgressIndicator progressIndicatorFrame,
+ DbSourceProxy[] sources, FeatureSettings featureSettings,
+ boolean isNucleotide)
{
- this.af = af;
+ listeners = new ArrayList<>();
+ this.progressWindow = progressIndicatorFrame;
alseqs = new SequenceI[seqs.length];
SequenceI[] ds = new SequenceI[seqs.length];
for (int i = 0; i < seqs.length; i++)
{
alseqs[i] = seqs[i];
if (seqs[i].getDatasetSequence() != null)
+ {
ds[i] = seqs[i].getDatasetSequence();
+ }
else
+ {
ds[i] = seqs[i];
+ }
}
this.dataset = ds;
// TODO Jalview 2.5 lots of this code should be in the gui package!
- sfetcher = jalview.gui.SequenceFetcher.getSequenceFetcherSingleton(af);
+ sfetcher = jalview.gui.SequenceFetcher.getSequenceFetcherSingleton();
// set default behaviour for transferring excess sequence data to the
// dataset
- trimDsSeqs = Cache.getDefault("TRIM_FETCHED_DATASET_SEQS", true);
+ trimDsSeqs = Cache.getDefault(TRIM_RETRIEVED_SEQUENCES, true);
if (sources == null)
{
- // af.featureSettings_actionPerformed(null);
- String[] defdb = null, otherdb = sfetcher
- .getDbInstances(jalview.ws.dbsources.das.datamodel.DasSequenceSource.class);
- List<DbSourceProxy> selsources = new ArrayList<DbSourceProxy>();
- Vector dasselsrc = (af.featureSettings != null) ? af.featureSettings
- .getSelectedSources() : new jalview.gui.DasSourceBrowser()
- .getSelectedSources();
- Enumeration<jalviewSourceI> en = dasselsrc.elements();
- while (en.hasMoreElements())
- {
- jalviewSourceI src = en.nextElement();
- List<DbSourceProxy> sp = src.getSequenceSourceProxies();
- if (sp != null)
- {
- selsources.addAll(sp);
- if (sp.size() > 1)
- {
- Cache.log.debug("Added many Db Sources for :" + src.getTitle());
- }
- }
- }
- // select appropriate databases based on alignFrame context.
- if (af.getViewport().getAlignment().isNucleotide())
- {
- defdb = DBRefSource.DNACODINGDBS;
- }
- else
- {
- defdb = DBRefSource.PROTEINDBS;
- }
- List<DbSourceProxy> srces = new ArrayList<DbSourceProxy>();
- for (String ddb : defdb)
- {
- List<DbSourceProxy> srcesfordb = sfetcher.getSourceProxy(ddb);
- if (srcesfordb != null)
- {
- srces.addAll(srcesfordb);
- }
- }
-
- // append the selected sequence sources to the default dbs
- srces.addAll(selsources);
- dbSources = srces.toArray(new DbSourceProxy[0]);
+ setDatabaseSources(featureSettings, isNucleotide);
}
else
{
}
/**
- * retrieve all the das sequence sources and add them to the list of db
- * sources to retrieve from
+ * Helper method to configure the list of database sources to query
+ *
+ * @param featureSettings
+ * @param forNucleotide
*/
- public void appendAllDasSources()
+ void setDatabaseSources(FeatureSettings featureSettings,
+ boolean forNucleotide)
{
- if (dbSources == null)
+ // af.featureSettings_actionPerformed(null);
+ String[] defdb = null;
+ List<DbSourceProxy> selsources = new ArrayList<>();
+ // select appropriate databases based on alignFrame context.
+ if (forNucleotide)
{
- dbSources = new DbSourceProxy[0];
+ defdb = DBRefSource.DNACODINGDBS;
}
- // append additional sources
- DbSourceProxy[] otherdb = sfetcher
- .getDbSourceProxyInstances(jalview.ws.dbsources.das.datamodel.DasSequenceSource.class);
- if (otherdb != null && otherdb.length > 0)
+ else
{
- DbSourceProxy[] newsrc = new DbSourceProxy[dbSources.length
- + otherdb.length];
- System.arraycopy(dbSources, 0, newsrc, 0, dbSources.length);
- System.arraycopy(otherdb, 0, newsrc, dbSources.length, otherdb.length);
- dbSources = newsrc;
+ defdb = DBRefSource.PROTEINDBS;
}
+ List<DbSourceProxy> srces = new ArrayList<>();
+ for (String ddb : defdb)
+ {
+ List<DbSourceProxy> srcesfordb = sfetcher.getSourceProxy(ddb);
+ if (srcesfordb != null)
+ {
+ for (DbSourceProxy src : srcesfordb)
+ {
+ if (!srces.contains(src))
+ {
+ srces.addAll(srcesfordb);
+ }
+ }
+ }
+ }
+ // append the PDB data source, since it is 'special', catering for both
+ // nucleotide and protein
+ // srces.addAll(sfetcher.getSourceProxy(DBRefSource.PDB));
+
+ srces.addAll(selsources);
+ dbSources = srces.toArray(new DbSourceProxy[srces.size()]);
+ }
+
+ /**
+ * Constructor with only sequences provided
+ *
+ * @param sequences
+ */
+ public DBRefFetcher(SequenceI[] sequences)
+ {
+ this(sequences, null, null, null, false);
+ }
+
+ /**
+ * Add a listener to be notified when sequence fetching is complete
+ *
+ * @param l
+ */
+ public void addListener(FetchFinishedListenerI l)
+ {
+ listeners.add(l);
}
/**
*/
public void fetchDBRefs(boolean waitTillFinished)
{
- Thread thread = new Thread(this);
- thread.start();
- running = true;
-
if (waitTillFinished)
{
- while (running)
- {
- try
- {
- Thread.sleep(500);
- } catch (Exception ex)
- {
- }
- }
+ run();
+ }
+ else
+ {
+ new Thread(this).start();
}
}
*/
void addSeqId(SequenceI seq, String key)
{
- key = key.toUpperCase();
+ key = key.toUpperCase(Locale.ROOT);
- Vector seqs;
+ Vector<SequenceI> seqs;
if (seqRefs.containsKey(key))
{
- seqs = (Vector) seqRefs.get(key);
+ seqs = seqRefs.get(key);
if (seqs != null && !seqs.contains(seq))
{
}
else if (seqs == null)
{
- seqs = new Vector();
+ seqs = new Vector<>();
seqs.addElement(seq);
}
}
else
{
- seqs = new Vector();
+ seqs = new Vector<>();
seqs.addElement(seq);
}
/**
* DOCUMENT ME!
*/
+ @Override
public void run()
{
if (dbSources == null)
{
- throw new Error(
- MessageManager
- .getString("error.implementation_error_must_init_dbsources"));
+ throw new Error(MessageManager
+ .getString("error.implementation_error_must_init_dbsources"));
}
- running = true;
long startTime = System.currentTimeMillis();
- af.setProgressBar(MessageManager.getString("status.fetching_db_refs"),
- startTime);
+ if (progressWindow != null)
+ {
+ progressWindow.setProgressBar(
+ MessageManager.getString("status.fetching_db_refs"),
+ startTime);
+ }
try
{
if (Cache.getDefault("DBREFFETCH_USEPICR", false))
{
- picrClient = new uk.ac.ebi.www.picr.AccessionMappingService.AccessionMapperServiceLocator()
+ picrClient = new AccessionMapperServiceLocator()
.getAccessionMapperPort();
}
} catch (Exception e)
System.err.println("Couldn't locate PICR service instance.\n");
e.printStackTrace();
}
+
+ Vector<SequenceI> sdataset = new Vector<>(
+ Arrays.asList(dataset));
+ List<String> warningMessages = new ArrayList<>();
+
+ // clear any old feature display settings recorded from past sessions
+ featureDisplaySettings = null;
+
int db = 0;
- Vector sdataset = new Vector();
- for (int s = 0; s < dataset.length; s++)
- {
- sdataset.addElement(dataset[s]);
- }
while (sdataset.size() > 0 && db < dbSources.length)
{
- int maxqlen = 1; // default number of queries made to at one time
- System.err.println("Verifying against " + dbSources[db].getDbName());
- boolean dn = false;
+ int maxqlen = 1; // default number of queries made at one time
+ System.out.println("Verifying against " + dbSources[db].getDbName());
// iterate through db for each remaining un-verified sequence
SequenceI[] currSeqs = new SequenceI[sdataset.size()];
sdataset.copyInto(currSeqs);// seqs that are to be validated against
// dbSources[db]
- Vector queries = new Vector(); // generated queries curSeq
- seqRefs = new Hashtable();
+ Vector<String> queries = new Vector<>(); // generated queries curSeq
+ seqRefs = new Hashtable<>();
int seqIndex = 0;
- jalview.ws.seqfetcher.DbSourceProxy dbsource = dbSources[db];
+ DbSourceProxy dbsource = dbSources[db];
+ // for moment, we dumbly iterate over all retrieval sources for a
+ // particular database
+ // TODO: introduce multithread multisource queries and logic to remove a
+ // query from other sources if any source for a database returns a
+ // record
+ maxqlen = dbsource.getMaximumQueryCount();
+
+ while (queries.size() > 0 || seqIndex < currSeqs.length)
{
- // for moment, we dumbly iterate over all retrieval sources for a
- // particular database
- // TODO: introduce multithread multisource queries and logic to remove a
- // query from other sources if any source for a database returns a
- // record
- if (dbsource.getDbSourceProperties().containsKey(
- DBRefSource.MULTIACC))
- {
- maxqlen = ((Integer) dbsource.getDbSourceProperties().get(
- DBRefSource.MULTIACC)).intValue();
- }
- else
- {
- maxqlen = 1;
- }
- while (queries.size() > 0 || seqIndex < currSeqs.length)
+ if (queries.size() > 0)
{
- if (queries.size() > 0)
- {
- // Still queries to make for current seqIndex
- StringBuffer queryString = new StringBuffer("");
- int numq = 0, nqSize = (maxqlen > queries.size()) ? queries
- .size() : maxqlen;
+ // Still queries to make for current seqIndex
+ StringBuffer queryString = new StringBuffer("");
+ int numq = 0;
+ int nqSize = (maxqlen > queries.size()) ? queries.size()
+ : maxqlen;
- while (queries.size() > 0 && numq < nqSize)
- {
- String query = (String) queries.elementAt(0);
- if (dbsource.isValidReference(query))
- {
- queryString.append((numq == 0) ? "" : dbsource
- .getAccessionSeparator());
- queryString.append(query);
- numq++;
- }
- // remove the extracted query string
- queries.removeElementAt(0);
- }
- // make the queries and process the response
- AlignmentI retrieved = null;
- try
- {
- if (jalview.bin.Cache.log.isDebugEnabled())
- {
- jalview.bin.Cache.log.debug("Querying "
- + dbsource.getDbName() + " with : '"
- + queryString.toString() + "'");
- }
- retrieved = dbsource.getSequenceRecords(queryString
- .toString());
- } catch (Exception ex)
- {
- ex.printStackTrace();
- } catch (OutOfMemoryError err)
+ while (queries.size() > 0 && numq < nqSize)
+ {
+ String query = queries.elementAt(0);
+ if (dbsource.isValidReference(query))
{
- new OOMWarning("retrieving database references ("
- + queryString.toString() + ")", err);
+ queryString.append(
+ (numq == 0) ? "" : dbsource.getAccessionSeparator());
+ queryString.append(query);
+ numq++;
}
- if (retrieved != null)
+ // remove the extracted query string
+ queries.removeElementAt(0);
+ }
+ // make the queries and process the response
+ AlignmentI retrieved = null;
+ try
+ {
+ if (Cache.isDebugEnabled())
{
- transferReferences(sdataset, dbsource.getDbSource(),
- retrieved, trimDsSeqs);
+ Cache.debug("Querying " + dbsource.getDbName()
+ + " with : '" + queryString.toString() + "'");
}
+ retrieved = dbsource.getSequenceRecords(queryString.toString());
+ } catch (Exception ex)
+ {
+ ex.printStackTrace();
+ } catch (OutOfMemoryError err)
+ {
+ new OOMWarning("retrieving database references ("
+ + queryString.toString() + ")", err);
+ }
+ if (retrieved != null)
+ {
+ transferReferences(sdataset, dbsource, retrieved,
+ trimDsSeqs, warningMessages);
}
- else
+ }
+ else
+ {
+ // make some more strings for use as queries
+ for (int i = 0; (seqIndex < dataset.length)
+ && (i < 50); seqIndex++, i++)
{
- // make some more strings for use as queries
- for (int i = 0; (seqIndex < dataset.length) && (i < 50); seqIndex++, i++)
+ SequenceI sequence = dataset[seqIndex];
+ List<DBRefEntry> uprefs = DBRefUtils
+ .selectRefs(sequence.getDBRefs(), new String[]
+ { dbsource.getDbSource() }); // jalview.datamodel.DBRefSource.UNIPROT
+ // });
+ // check for existing dbrefs to use
+ if (uprefs != null && uprefs.size() > 0)
{
- SequenceI sequence = dataset[seqIndex];
- DBRefEntry[] uprefs = jalview.util.DBRefUtils.selectRefs(
- sequence.getDBRef(),
- new String[] { dbsource.getDbSource() }); // jalview.datamodel.DBRefSource.UNIPROT
- // });
- // check for existing dbrefs to use
- if (uprefs != null && uprefs.length > 0)
+ for (int j = 0, n = uprefs.size(); j < n; j++)
{
- for (int j = 0; j < uprefs.length; j++)
- {
- addSeqId(sequence, uprefs[j].getAccessionId());
- queries.addElement(uprefs[j].getAccessionId()
- .toUpperCase());
- }
+ DBRefEntry upref = uprefs.get(j);
+ addSeqId(sequence, upref.getAccessionId());
+ queries.addElement(
+ upref.getAccessionId().toUpperCase(Locale.ROOT));
}
- else
+ }
+ else
+ {
+ Pattern possibleIds = Pattern.compile("[A-Za-z0-9_]+");
+ // generate queries from sequence ID string
+ Matcher tokens = possibleIds.matcher(sequence.getName());
+ int p=0;
+ while (tokens.find(p))
{
- // generate queries from sequence ID string
- StringTokenizer st = new StringTokenizer(
- sequence.getName(), "|");
- while (st.hasMoreTokens())
+ String token = tokens.group();
+ p = tokens.end();
+ UPEntry[] presp = null;
+ if (picrClient != null)
{
- String token = st.nextToken();
- UPEntry[] presp = null;
- if (picrClient != null)
+ // resolve the string against PICR to recover valid IDs
+ try
{
- // resolve the string against PICR to recover valid IDs
- try
- {
- presp = picrClient.getUPIForAccession(token, null,
- picrClient.getMappedDatabaseNames(), null,
- true);
- } catch (Exception e)
- {
- System.err.println("Exception with Picr for '"
- + token + "'\n");
- e.printStackTrace();
- }
- }
- if (presp != null && presp.length > 0)
+ presp = picrClient.getUPIForAccession(token, null,
+ picrClient.getMappedDatabaseNames(), null,
+ true);
+ } catch (Exception e)
{
- for (int id = 0; id < presp.length; id++)
- {
- // construct sequences from response if sequences are
- // present, and do a transferReferences
- // otherwise transfer non sequence x-references directly.
- }
- System.out
- .println("Validated ID against PICR... (for what its worth):"
- + token);
- addSeqId(sequence, token);
- queries.addElement(token.toUpperCase());
+ System.err.println(
+ "Exception with Picr for '" + token + "'\n");
+ e.printStackTrace();
}
- else
+ }
+ if (presp != null && presp.length > 0)
+ {
+ for (int id = 0; id < presp.length; id++)
{
- // if ()
- // System.out.println("Not querying source with token="+token+"\n");
- addSeqId(sequence, token);
- queries.addElement(token.toUpperCase());
+ // construct sequences from response if sequences are
+ // present, and do a transferReferences
+ // otherwise transfer non sequence x-references directly.
}
+ System.out.println(
+ "Validated ID against PICR... (for what its worth):"
+ + token);
+ addSeqId(sequence, token);
+ queries.addElement(token.toUpperCase(Locale.ROOT));
+ }
+ else
+ {
+ // if ()
+ // System.out.println("Not querying source with
+ // token="+token+"\n");
+ addSeqId(sequence, token);
+ queries.addElement(token.toUpperCase(Locale.ROOT));
}
}
}
}
// advance to next database
db++;
- } // all databases have been queries.
- if (sbuffer.length() > 0)
+ } // all databases have been queried
+ if (!warningMessages.isEmpty())
{
- output.setText(MessageManager
- .getString("label.your_sequences_have_been_verified")
- + sbuffer.toString());
+ StringBuilder sb = new StringBuilder(warningMessages.size() * 30);
+ sb.append(MessageManager
+ .getString("label.your_sequences_have_been_verified"));
+ for (String msg : warningMessages)
+ {
+ sb.append(msg).append(NEWLINE);
+ }
+ output.setText(sb.toString());
+
Desktop.addInternalFrame(output,
- MessageManager.getString("label.sequence_names_updated"),
- 600, 300);
+ MessageManager.getString("label.sequences_updated"), 600,
+ 300);
// The above is the dataset, we must now find out the index
// of the viewed sequence
}
+ if (progressWindow != null)
+ {
+ progressWindow.setProgressBar(
+ MessageManager.getString("label.dbref_search_completed"),
+ startTime);
+ }
- af.setProgressBar(
- MessageManager.getString("label.dbref_search_completed"),
- startTime);
- // promptBeforeBlast();
-
- running = false;
-
+ for (FetchFinishedListenerI listener : listeners)
+ {
+ listener.finished();
+ }
}
/**
* Verify local sequences in seqRefs against the retrieved sequence database
- * records.
+ * records. Returns true if any sequence was modified as a result (start/end
+ * changed and/or sequence enlarged), else false.
*
+ * @param sdataset
+ * dataset sequences we are retrieving for
+ * @param dbSource
+ * database source we are retrieving from
+ * @param retrievedAl
+ * retrieved sequences as alignment
* @param trimDatasetSeqs
- *
+ * if true, sequences will not be enlarged to match longer retrieved
+ * sequences, only their start/end adjusted
+ * @param warningMessages
+ * a list of messages to add to
*/
- void transferReferences(Vector sdataset, String dbSource,
- AlignmentI retrievedAl, boolean trimDatasetSeqs) // File
- // file)
+ boolean transferReferences(Vector<SequenceI> sdataset,
+ DbSourceProxy dbSourceProxy,
+ AlignmentI retrievedAl, boolean trimDatasetSeqs,
+ List<String> warningMessages)
{
- System.out.println("trimming ? " + trimDatasetSeqs);
+ // System.out.println("trimming ? " + trimDatasetSeqs);
if (retrievedAl == null || retrievedAl.getHeight() == 0)
{
- return;
+ return false;
}
- SequenceI[] retrieved = recoverDbSequences(retrievedAl
- .getSequencesArray());
- SequenceI sequence = null;
- boolean transferred = false;
- StringBuffer messages = new StringBuffer();
- // Vector entries = new Uniprot().getUniprotEntries(file);
+ String dbSource = dbSourceProxy.getDbName();
+ boolean modified = false;
+ SequenceI[] retrieved = recoverDbSequences(
+ retrievedAl.getSequencesArray());
+ SequenceI sequence = null;
- int i, iSize = retrieved.length; // entries == null ? 0 : entries.size();
- // UniprotEntry entry;
- for (i = 0; i < iSize; i++)
+ for (SequenceI retrievedSeq : retrieved)
{
- SequenceI entry = retrieved[i]; // (UniprotEntry) entries.elementAt(i);
-
// Work out which sequences this sequence matches,
// taking into account all accessionIds and names in the file
- Vector sequenceMatches = new Vector();
+ Vector<SequenceI> sequenceMatches = new Vector<>();
// look for corresponding accession ids
- DBRefEntry[] entryRefs = jalview.util.DBRefUtils.selectRefs(
- entry.getDBRef(), new String[] { dbSource });
+ List<DBRefEntry> entryRefs = DBRefUtils
+ .selectRefs(retrievedSeq.getDBRefs(), new String[]
+ { dbSource });
if (entryRefs == null)
{
System.err
.println("Dud dbSource string ? no entryrefs selected for "
- + dbSource + " on " + entry.getName());
+ + dbSource + " on " + retrievedSeq.getName());
continue;
}
- for (int j = 0; j < entryRefs.length; j++)
+ for (int j = 0, n = entryRefs.size(); j < n; j++)
{
- String accessionId = entryRefs[j].getAccessionId(); // .getAccession().elementAt(j).toString();
+ DBRefEntry ref = entryRefs.get(j);
+ String accessionId = ref.getAccessionId();
// match up on accessionId
- if (seqRefs.containsKey(accessionId.toUpperCase()))
+ if (seqRefs.containsKey(accessionId.toUpperCase(Locale.ROOT)))
{
- Vector seqs = (Vector) seqRefs.get(accessionId);
+ Vector<SequenceI> seqs = seqRefs.get(accessionId);
for (int jj = 0; jj < seqs.size(); jj++)
{
- sequence = (SequenceI) seqs.elementAt(jj);
+ sequence = seqs.elementAt(jj);
if (!sequenceMatches.contains(sequence))
{
sequenceMatches.addElement(sequence);
}
}
}
- if (sequenceMatches.size() == 0)
+ if (sequenceMatches.isEmpty())
{
// failed to match directly on accessionId==query so just compare all
// sequences to entry
- Enumeration e = seqRefs.keys();
+ Enumeration<String> e = seqRefs.keys();
while (e.hasMoreElements())
{
- Vector sqs = (Vector) seqRefs.get(e.nextElement());
+ Vector<SequenceI> sqs = seqRefs.get(e.nextElement());
if (sqs != null && sqs.size() > 0)
{
- Enumeration sqe = sqs.elements();
+ Enumeration<SequenceI> sqe = sqs.elements();
while (sqe.hasMoreElements())
{
sequenceMatches.addElement(sqe.nextElement());
// could be useful to extend this so we try to find any 'significant'
// information in common between two sequence objects.
/*
- * DBRefEntry[] entryRefs =
+ * List<DBRefEntry> entryRefs =
* jalview.util.DBRefUtils.selectRefs(entry.getDBRef(), new String[] {
* dbSource }); for (int j = 0; j < entry.getName().size(); j++) { String
* name = entry.getName().elementAt(j).toString(); if
* seqs.elementAt(jj); if (!sequenceMatches.contains(sequence)) {
* sequenceMatches.addElement(sequence); } } } }
*/
+ if (sequenceMatches.size() > 0)
+ {
+ addFeatureSettings(dbSourceProxy);
+ }
// sequenceMatches now contains the set of all sequences associated with
// the returned db record
- String entrySeq = entry.getSequenceAsString().toUpperCase();
+ final String retrievedSeqString = retrievedSeq.getSequenceAsString();
+ String entrySeq = retrievedSeqString.toUpperCase(Locale.ROOT);
for (int m = 0; m < sequenceMatches.size(); m++)
{
- sequence = (SequenceI) sequenceMatches.elementAt(m);
+ sequence = sequenceMatches.elementAt(m);
// only update start and end positions and shift features if there are
// no existing references
// TODO: test for legacy where uniprot or EMBL refs exist but no
// mappings are made (but content matches retrieved set)
- boolean updateRefFrame = sequence.getDBRef() == null
- || sequence.getDBRef().length == 0;
+ boolean updateRefFrame = sequence.getDBRefs() == null
+ || sequence.getDBRefs().size() == 0;
// TODO:
// verify sequence against the entry sequence
- String nonGapped = AlignSeq.extractGaps("-. ",
- sequence.getSequenceAsString()).toUpperCase();
+ Mapping mp;
+ final int sequenceStart = sequence.getStart();
+ boolean remoteEnclosesLocal = false;
+ String nonGapped = AlignSeq
+ .extractGaps("-. ", sequence.getSequenceAsString())
+ .toUpperCase(Locale.ROOT);
int absStart = entrySeq.indexOf(nonGapped);
- int mapStart = entry.getStart();
- jalview.datamodel.Mapping mp;
-
if (absStart == -1)
{
- // Is local sequence contained in dataset sequence?
+ // couldn't find local sequence in sequence from database, so check if
+ // the database sequence is a subsequence of local sequence
absStart = nonGapped.indexOf(entrySeq);
if (absStart == -1)
- { // verification failed.
- messages.append(sequence.getName()
- + " SEQUENCE NOT %100 MATCH \n");
+ {
+ // verification failed. couldn't find any relationship between
+ // entrySeq and local sequence
+ // messages suppressed as many-to-many matches are confusing
+ // String msg = sequence.getName()
+ // + " Sequence not 100% match with "
+ // + retrievedSeq.getName();
+ // addWarningMessage(warningMessages, msg);
continue;
}
- transferred = true;
- sbuffer.append(sequence.getName() + " HAS " + absStart
- + " PREFIXED RESIDUES COMPARED TO " + dbSource + "\n");
- //
- // + " - ANY SEQUENCE FEATURES"
- // + " HAVE BEEN ADJUSTED ACCORDINGLY \n");
- // absStart = 0;
- // create valid mapping between matching region of local sequence and
- // the mapped sequence
- mp = new Mapping(null, new int[] {
- sequence.getStart() + absStart,
- sequence.getStart() + absStart + entrySeq.length() - 1 },
- new int[] { entry.getStart(),
- entry.getStart() + entrySeq.length() - 1 }, 1, 1);
- updateRefFrame = false; // mapping is based on current start/end so
- // don't modify start and end
+ /*
+ * retrieved sequence is a proper subsequence of local sequence
+ */
+ String msg = sequence.getName() + " has " + absStart
+ + " prefixed residues compared to "
+ + retrievedSeq.getName();
+ addWarningMessage(warningMessages, msg);
+
+ /*
+ * So create a mapping to the external entry from the matching region of
+ * the local sequence, and leave local start/end untouched.
+ */
+ mp = new Mapping(null,
+ new int[]
+ { sequenceStart + absStart,
+ sequenceStart + absStart + entrySeq.length() - 1 },
+ new int[]
+ { retrievedSeq.getStart(),
+ retrievedSeq.getStart() + entrySeq.length() - 1 },
+ 1, 1);
+ updateRefFrame = false;
}
else
{
- transferred = true;
- // update start and end of local sequence to place it in entry's
- // reference frame.
- // apply identity map map from whole of local sequence to matching
- // region of database
- // sequence
- mp = null; // Mapping.getIdentityMap();
- // new Mapping(null,
- // new int[] { absStart+sequence.getStart(),
- // absStart+sequence.getStart()+entrySeq.length()-1},
- // new int[] { entry.getStart(), entry.getEnd() }, 1, 1);
- // relocate local features for updated start
+ /*
+ * local sequence is a subsequence of (or matches) retrieved sequence
+ */
+ remoteEnclosesLocal = true;
+ mp = null;
+
if (updateRefFrame)
{
- if (sequence.getSequenceFeatures() != null)
+ /*
+ * relocate existing sequence features by offset
+ */
+ int startShift = absStart - sequenceStart + 1;
+ if (startShift != 0)
{
- SequenceFeature[] sf = sequence.getSequenceFeatures();
- int start = sequence.getStart();
- int end = sequence.getEnd();
- int startShift = 1 - absStart - start; // how much the features
- // are
- // to be shifted by
- for (int sfi = 0; sfi < sf.length; sfi++)
- {
- if (sf[sfi].getBegin() >= start && sf[sfi].getEnd() <= end)
- {
- // shift feature along by absstart
- sf[sfi].setBegin(sf[sfi].getBegin() + startShift);
- sf[sfi].setEnd(sf[sfi].getEnd() + startShift);
- }
- }
+ modified |= sequence.getFeatures().shiftFeatures(1,
+ startShift);
}
}
}
System.out.println("Adding dbrefs to " + sequence.getName()
- + " from " + dbSource + " sequence : " + entry.getName());
- sequence.transferAnnotation(entry, mp);
- // unknownSequences.remove(sequence);
- int absEnd = absStart + nonGapped.length();
- absStart += 1;
+ + " from " + dbSource + " sequence : "
+ + retrievedSeq.getName());
+ sequence.transferAnnotation(retrievedSeq, mp);
+
+ absStart += retrievedSeq.getStart();
+ int absEnd = absStart + nonGapped.length() - 1;
if (!trimDatasetSeqs)
{
- // insert full length sequence from record
- sequence.setSequence(entry.getSequenceAsString());
- sequence.setStart(entry.getStart());
+ /*
+ * update start position and/or expand to longer retrieved sequence
+ */
+ if (!retrievedSeqString.equals(sequence.getSequenceAsString())
+ && remoteEnclosesLocal)
+ {
+ sequence.setSequence(retrievedSeqString);
+ modified = true;
+ addWarningMessage(warningMessages,
+ "Sequence for " + sequence.getName() + " expanded from "
+ + retrievedSeq.getName());
+ }
+ if (sequence.getStart() != retrievedSeq.getStart())
+ {
+ sequence.setStart(retrievedSeq.getStart());
+ modified = true;
+ if (absStart != sequenceStart)
+ {
+ addWarningMessage(warningMessages,
+ "Start/end position for " + sequence.getName()
+ + " updated from " + retrievedSeq.getName());
+ }
+ }
}
if (updateRefFrame)
{
if (trimDatasetSeqs)
{
// just fix start/end
- sequence.setStart(absStart);
- sequence.setEnd(absEnd);
+ if (sequence.getStart() != absStart
+ || sequence.getEnd() != absEnd)
+ {
+ sequence.setStart(absStart);
+ sequence.setEnd(absEnd);
+ modified = true;
+ addWarningMessage(warningMessages,
+ "Start/end for " + sequence.getName()
+ + " updated from " + retrievedSeq.getName());
+ }
}
// search for alignment sequences to update coordinate frame for
for (int alsq = 0; alsq < alseqs.length; alsq++)
{
if (alseqs[alsq].getDatasetSequence() == sequence)
{
- String ngAlsq = AlignSeq.extractGaps("-. ",
- alseqs[alsq].getSequenceAsString()).toUpperCase();
+ String ngAlsq = AlignSeq
+ .extractGaps("-. ",
+ alseqs[alsq].getSequenceAsString())
+ .toUpperCase(Locale.ROOT);
int oldstrt = alseqs[alsq].getStart();
alseqs[alsq].setStart(sequence.getSequenceAsString()
- .toUpperCase().indexOf(ngAlsq)
- + sequence.getStart());
+ .toUpperCase(Locale.ROOT).indexOf(ngAlsq) + sequence.getStart());
if (oldstrt != alseqs[alsq].getStart())
{
- alseqs[alsq].setEnd(ngAlsq.length()
- + alseqs[alsq].getStart() - 1);
+ alseqs[alsq].setEnd(
+ ngAlsq.length() + alseqs[alsq].getStart() - 1);
+ modified = true;
}
}
}
// and remove it from the rest
// TODO: decide if we should remove annotated sequence from set
sdataset.remove(sequence);
- // TODO: should we make a note of sequences that have received new DB
- // ids, so we can query all enabled DAS servers for them ?
}
}
- if (!transferred)
+ return modified;
+ }
+
+ Map<String, FeatureSettingsModelI> featureDisplaySettings = null;
+
+ private void addFeatureSettings(DbSourceProxy dbSourceProxy)
+ {
+ FeatureSettingsModelI fsettings = dbSourceProxy
+ .getFeatureColourScheme();
+ if (fsettings != null)
+ {
+ if (featureDisplaySettings == null)
+ {
+ featureDisplaySettings = new HashMap<>();
+ }
+ featureDisplaySettings.put(dbSourceProxy.getDbName(), fsettings);
+ }
+ }
+
+ /**
+ *
+ * @return any feature settings associated with sources that have provided sequences
+ */
+ public List<FeatureSettingsModelI>getFeatureSettingsModels()
+ {
+ return featureDisplaySettings == null
+ ? Arrays.asList(new FeatureSettingsModelI[0])
+ : Arrays.asList(featureDisplaySettings.values()
+ .toArray(new FeatureSettingsModelI[1]));
+ }
+ /**
+ * Adds the message to the list unless it already contains it
+ *
+ * @param messageList
+ * @param msg
+ */
+ void addWarningMessage(List<String> messageList, String msg)
+ {
+ if (!messageList.contains(msg))
{
- // report the ID/sequence mismatches
- sbuffer.append(messages);
+ messageList.add(msg);
}
}
*/
private SequenceI[] recoverDbSequences(SequenceI[] sequencesArray)
{
- Vector nseq = new Vector();
- for (int i = 0; sequencesArray != null && i < sequencesArray.length; i++)
+ int n;
+ if (sequencesArray == null || (n = sequencesArray.length) == 0)
+ return sequencesArray;
+ ArrayList<SequenceI> nseq = new ArrayList<>();
+ for (int i = 0;i < n; i++)
{
- nseq.addElement(sequencesArray[i]);
- DBRefEntry dbr[] = sequencesArray[i].getDBRef();
- jalview.datamodel.Mapping map = null;
- for (int r = 0; (dbr != null) && r < dbr.length; r++)
+ nseq.add(sequencesArray[i]);
+ List<DBRefEntry> dbr = sequencesArray[i].getDBRefs();
+ Mapping map = null;
+ if (dbr != null)
{
- if ((map = dbr[r].getMap()) != null)
+ for (int r = 0, rn = dbr.size(); r < rn; r++)
{
- if (map.getTo() != null && !nseq.contains(map.getTo()))
+ if ((map = dbr.get(r).getMap()) != null)
{
- nseq.addElement(map.getTo());
- }
+ if (map.getTo() != null && !nseq.contains(map.getTo()))
+ {
+ nseq.add(map.getTo());
+ }
+ }
}
}
}
+ // BH 2019.01.25 question here if this is the right logic. Return the original if nothing found?
if (nseq.size() > 0)
{
- sequencesArray = new SequenceI[nseq.size()];
- nseq.toArray(sequencesArray);
+ return nseq.toArray(new SequenceI[nseq.size()]);
}
return sequencesArray;
}
git://source.jalview.org / jalview.git / blobdiff