// Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
//
// Contact: phylosoft @ gmail . com
-// WWW: www.phylosoft.org/forester
+// WWW: https://sites.google.com/site/cmzmasek/home/software/forester
package org.forester.archaeopteryx.tools;
-import java.io.FileNotFoundException;
import java.io.IOException;
import java.net.UnknownHostException;
import java.util.SortedSet;
-import java.util.TreeSet;
import javax.swing.JOptionPane;
import org.forester.archaeopteryx.MainFrameApplication;
import org.forester.archaeopteryx.TreePanel;
import org.forester.phylogeny.Phylogeny;
-import org.forester.phylogeny.PhylogenyNode;
-import org.forester.phylogeny.data.Accession;
-import org.forester.phylogeny.data.Identifier;
-import org.forester.phylogeny.data.Sequence;
-import org.forester.phylogeny.data.Taxonomy;
-import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
-import org.forester.util.ForesterUtil;
-import org.forester.ws.uniprot.DatabaseTools;
-import org.forester.ws.uniprot.SequenceDatabaseEntry;
-import org.forester.ws.uniprot.UniProtWsTools;
+import org.forester.ws.seqdb.SequenceDbWsTools;
public final class SequenceDataRetriver extends RunnableProcess {
+ private final static int DEFAULT_LINES_TO_RETURN = 4000;
private final Phylogeny _phy;
private final MainFrameApplication _mf;
private final TreePanel _treepanel;
- private final static boolean DEBUG = false;
-
- private enum Db {
- UNKNOWN, UNIPROT, EMBL;
- }
+ public final static boolean DEBUG = false;
public SequenceDataRetriver( final MainFrameApplication mf, final TreePanel treepanel, final Phylogeny phy ) {
_phy = phy;
_treepanel = treepanel;
}
- private String getBaseUrl() {
- return UniProtWsTools.BASE_URL;
+ @Override
+ public void run() {
+ execute();
}
private void execute() {
start( _mf, "sequence data" );
SortedSet<String> not_found = null;
try {
- not_found = obtainSeqInformation( _phy );
+ not_found = SequenceDbWsTools.obtainSeqInformation( _phy, false, true, DEFAULT_LINES_TO_RETURN );
}
catch ( final UnknownHostException e ) {
JOptionPane.showMessageDialog( _mf,
- "Could not connect to \"" + getBaseUrl() + "\"",
- "Network error during taxonomic information gathering",
+ e.getLocalizedMessage(),
+ "Network error during sequence data gathering",
JOptionPane.ERROR_MESSAGE );
return;
}
e.printStackTrace();
JOptionPane.showMessageDialog( _mf,
e.toString(),
- "Failed to obtain taxonomic information",
+ "Failed to obtain sequence data",
JOptionPane.ERROR_MESSAGE );
return;
}
}
final StringBuffer sb = new StringBuffer();
if ( not_found.size() == 1 ) {
- sb.append( "Data for the following sequence identifier was not found:\n" );
+ sb.append( "For the following node no data was found:\n" );
}
else {
- sb.append( "Data for the following sequence identifiers was not found (total: " + not_found.size()
- + "):\n" );
+ sb.append( "For the following nodes no data was found: (total: " + not_found.size() + "):\n" );
}
int i = 0;
for( final String string : not_found ) {
else {
try {
JOptionPane.showMessageDialog( _mf,
- "UniProt sequence tool successfully completed",
- "UniProt Sequence Tool Completed",
+ "Sequence tool successfully completed",
+ "Sequence Tool Completed",
JOptionPane.INFORMATION_MESSAGE );
}
catch ( final Exception e ) {
}
}
}
-
- public static SortedSet<String> obtainSeqInformation( final Phylogeny phy ) throws IOException {
- final SortedSet<String> not_found = new TreeSet<String>();
- for( final PhylogenyNodeIterator iter = phy.iteratorPostorder(); iter.hasNext(); ) {
- final PhylogenyNode node = iter.next();
- Sequence seq = null;
- Taxonomy tax = null;
- if ( node.getNodeData().isHasSequence() ) {
- seq = node.getNodeData().getSequence();
- }
- else {
- seq = new Sequence();
- }
- if ( node.getNodeData().isHasTaxonomy() ) {
- tax = node.getNodeData().getTaxonomy();
- }
- else {
- tax = new Taxonomy();
- }
- String query = null;
- Db db = Db.UNKNOWN;
- if ( node.getNodeData().isHasSequence() && ( node.getNodeData().getSequence().getAccession() != null )
- && !ForesterUtil.isEmpty( node.getNodeData().getSequence().getAccession().getSource() )
- && !ForesterUtil.isEmpty( node.getNodeData().getSequence().getAccession().getValue() )
- && node.getNodeData().getSequence().getAccession().getValue().toLowerCase().startsWith( "uniprot" ) ) {
- query = node.getNodeData().getSequence().getAccession().getValue();
- db = Db.UNIPROT;
- }
- else if ( node.getNodeData().isHasSequence()
- && ( node.getNodeData().getSequence().getAccession() != null )
- && !ForesterUtil.isEmpty( node.getNodeData().getSequence().getAccession().getSource() )
- && !ForesterUtil.isEmpty( node.getNodeData().getSequence().getAccession().getValue() )
- && ( node.getNodeData().getSequence().getAccession().getValue().toLowerCase().startsWith( "embl" ) || node
- .getNodeData().getSequence().getAccession().getValue().toLowerCase().startsWith( "ebi" ) ) ) {
- query = node.getNodeData().getSequence().getAccession().getValue();
- db = Db.EMBL;
- }
- else if ( !ForesterUtil.isEmpty( node.getName() ) ) {
- if ( ( query = UniProtWsTools.parseUniProtAccessor( node.getName() ) ) != null ) {
- db = Db.UNIPROT;
- }
- else if ( ( query = DatabaseTools.parseGenbankAccessor( node.getName() ) ) != null ) {
- db = Db.EMBL;
- }
- }
- if ( !ForesterUtil.isEmpty( query ) ) {
- SequenceDatabaseEntry db_entry = null;
- if ( db == Db.UNIPROT ) {
- if ( DEBUG ) {
- System.out.println( "uniprot: " + query );
- }
- try {
- db_entry = UniProtWsTools.obtainUniProtEntry( query, 200 );
- }
- catch ( final FileNotFoundException e ) {
- // Ignore.
- }
- }
- if ( ( db == Db.EMBL ) || ( ( db == Db.UNIPROT ) && ( db_entry == null ) ) ) {
- if ( DEBUG ) {
- System.out.println( "embl: " + query );
- }
- try {
- db_entry = UniProtWsTools.obtainEmblEntry( query, 200 );
- }
- catch ( final FileNotFoundException e ) {
- // Ignore.
- }
- if ( ( db == Db.UNIPROT ) && ( db_entry != null ) ) {
- db = Db.EMBL;
- }
- }
- if ( ( db_entry != null ) && !db_entry.isEmpty() ) {
- if ( !ForesterUtil.isEmpty( db_entry.getAccession() ) ) {
- String type = null;
- if ( db == Db.EMBL ) {
- type = "embl";
- }
- else if ( db == Db.UNIPROT ) {
- type = "uniprot";
- }
- seq.setAccession( new Accession( db_entry.getAccession(), type ) );
- }
- if ( !ForesterUtil.isEmpty( db_entry.getSequenceName() ) ) {
- seq.setName( db_entry.getSequenceName() );
- }
- if ( !ForesterUtil.isEmpty( db_entry.getSequenceSymbol() ) ) {
- seq.setSymbol( db_entry.getSequenceSymbol() );
- }
- if ( !ForesterUtil.isEmpty( db_entry.getTaxonomyScientificName() ) ) {
- tax.setScientificName( db_entry.getTaxonomyScientificName() );
- }
- if ( !ForesterUtil.isEmpty( db_entry.getTaxonomyIdentifier() ) ) {
- tax.setIdentifier( new Identifier( db_entry.getTaxonomyIdentifier(), "uniprot" ) );
- }
- node.getNodeData().setTaxonomy( tax );
- node.getNodeData().setSequence( seq );
- }
- else {
- not_found.add( node.getName() );
- }
- }
- }
- return not_found;
- }
-
- @Override
- public void run() {
- execute();
- }
}
git://source.jalview.org / jalview.git / blobdiff