forester/java/src/org/forester/archaeopteryx/tools/SequenceDataRetriver.java

   1 // Exp $
   2 // forester -- software libraries and applications
   3 // for genomics and evolutionary biology research.
   4 //
   5 // Copyright (C) 2010 Christian M Zmasek
   6 // Copyright (C) 2010 Sanford-Burnham Medical Research Institute
   7 // All rights reserved
   8 //
   9 // This library is free software; you can redistribute it and/or
  10 // modify it under the terms of the GNU Lesser General Public
  11 // License as published by the Free Software Foundation; either
  12 // version 2.1 of the License, or (at your option) any later version.
  13 //
  14 // This library is distributed in the hope that it will be useful,
  15 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  16 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  17 // Lesser General Public License for more details.
  18 //
  19 // You should have received a copy of the GNU Lesser General Public
  20 // License along with this library; if not, write to the Free Software
  21 // Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
  22 //
  23 // Contact: phylosoft @ gmail . com
  24 // WWW: www.phylosoft.org/forester
  25
  26 package org.forester.archaeopteryx.tools;
  27
  28 import java.io.IOException;
  29 import java.net.UnknownHostException;
  30 import java.util.SortedSet;
  31 import java.util.TreeSet;
  32
  33 import javax.swing.JOptionPane;
  34
  35 import org.forester.archaeopteryx.MainFrameApplication;
  36 import org.forester.archaeopteryx.TreePanel;
  37 import org.forester.phylogeny.Phylogeny;
  38 import org.forester.phylogeny.PhylogenyNode;
  39 import org.forester.phylogeny.data.Accession;
  40 import org.forester.phylogeny.data.Identifier;
  41 import org.forester.phylogeny.data.Sequence;
  42 import org.forester.phylogeny.data.Taxonomy;
  43 import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
  44 import org.forester.util.ForesterUtil;
  45 import org.forester.util.SequenceIdParser;
  46 import org.forester.ws.seqdb.SequenceDatabaseEntry;
  47 import org.forester.ws.seqdb.SequenceDbWsTools;
  48
  49 public final class SequenceDataRetriver extends RunnableProcess {
  50
  51     public final static int DEFAULT_LINES_TO_RETURN = 50;
  52
  53     private final Phylogeny            _phy;
  54     private final MainFrameApplication _mf;
  55     private final TreePanel            _treepanel;
  56     private final static boolean       DEBUG = true;
  57
  58     private enum Db {
  59         UNIPROT, EMBL, NCBI, NONE, REFSEQ;
  60     }
  61
  62     public SequenceDataRetriver( final MainFrameApplication mf, final TreePanel treepanel, final Phylogeny phy ) {
  63         _phy = phy;
  64         _mf = mf;
  65         _treepanel = treepanel;
  66     }
  67
  68     private void execute() {
  69         start( _mf, "sequence data" );
  70         SortedSet<String> not_found = null;
  71         try {
  72             not_found = obtainSeqInformation( _phy );
  73         }
  74         catch ( final UnknownHostException e ) {
  75             final String what = "_"; //TODO FIXME
  76             JOptionPane.showMessageDialog( _mf,
  77                                            "Could not connect to \"" + what + "\"",
  78                                            "Network error during taxonomic information gathering",
  79                                            JOptionPane.ERROR_MESSAGE );
  80             return;
  81         }
  82         catch ( final IOException e ) {
  83             e.printStackTrace();
  84             JOptionPane.showMessageDialog( _mf,
  85                                            e.toString(),
  86                                            "Failed to obtain taxonomic information",
  87                                            JOptionPane.ERROR_MESSAGE );
  88             return;
  89         }
  90         finally {
  91             end( _mf );
  92         }
  93         _treepanel.setTree( _phy );
  94         _mf.showWhole();
  95         _treepanel.setEdited( true );
  96         if ( ( not_found != null ) && ( not_found.size() > 0 ) ) {
  97             int max = not_found.size();
  98             boolean more = false;
  99             if ( max > 20 ) {
 100                 more = true;
 101                 max = 20;
 102             }
 103             final StringBuffer sb = new StringBuffer();
 104             if ( not_found.size() == 1 ) {
 105                 sb.append( "Data for the following sequence identifier was not found:\n" );
 106             }
 107             else {
 108                 sb.append( "Data for the following sequence identifiers was not found (total: " + not_found.size()
 109                         + "):\n" );
 110             }
 111             int i = 0;
 112             for( final String string : not_found ) {
 113                 if ( i > 19 ) {
 114                     break;
 115                 }
 116                 sb.append( string );
 117                 sb.append( "\n" );
 118                 ++i;
 119             }
 120             if ( more ) {
 121                 sb.append( "..." );
 122             }
 123             try {
 124                 JOptionPane.showMessageDialog( _mf,
 125                                                sb.toString(),
 126                                                "Sequence Tool Completed",
 127                                                JOptionPane.WARNING_MESSAGE );
 128             }
 129             catch ( final Exception e ) {
 130                 // Not important if this fails, do nothing.
 131             }
 132         }
 133         else {
 134             try {
 135                 JOptionPane.showMessageDialog( _mf,
 136                                                "Sequence tool successfully completed",
 137                                                "Sequence Tool Completed",
 138                                                JOptionPane.INFORMATION_MESSAGE );
 139             }
 140             catch ( final Exception e ) {
 141                 // Not important if this fails, do nothing.
 142             }
 143         }
 144     }
 145
 146     public static SortedSet<String> obtainSeqInformation( final Phylogeny phy ) throws IOException {
 147         final SortedSet<String> not_found = new TreeSet<String>();
 148         for( final PhylogenyNodeIterator iter = phy.iteratorPostorder(); iter.hasNext(); ) {
 149             final PhylogenyNode node = iter.next();
 150             final Sequence seq = node.getNodeData().isHasSequence() ?  node.getNodeData().getSequence() : new Sequence() ;
 151             final Taxonomy tax = node.getNodeData().isHasTaxonomy() ? node.getNodeData().getTaxonomy() : new Taxonomy() ;
 152
 153
 154             String query = null;
 155             Identifier id = null;
 156             Db db = Db.NONE;
 157             if ( node.getNodeData().isHasSequence() && ( node.getNodeData().getSequence().getAccession() != null )
 158                     && !ForesterUtil.isEmpty( node.getNodeData().getSequence().getAccession().getSource() )
 159                     && !ForesterUtil.isEmpty( node.getNodeData().getSequence().getAccession().getValue() )
 160                     && node.getNodeData().getSequence().getAccession().getValue().toLowerCase().startsWith( "uniprot" ) ) {
 161                 query = node.getNodeData().getSequence().getAccession().getValue();
 162                 db = Db.UNIPROT;
 163             }
 164             else if ( node.getNodeData().isHasSequence()
 165                     && ( node.getNodeData().getSequence().getAccession() != null )
 166                     && !ForesterUtil.isEmpty( node.getNodeData().getSequence().getAccession().getSource() )
 167                     && !ForesterUtil.isEmpty( node.getNodeData().getSequence().getAccession().getValue() )
 168                     && ( node.getNodeData().getSequence().getAccession().getValue().toLowerCase().startsWith( "embl" ) || node
 169                             .getNodeData().getSequence().getAccession().getValue().toLowerCase().startsWith( "ebi" ) ) ) {
 170                 query = node.getNodeData().getSequence().getAccession().getValue();
 171                 db = Db.EMBL;
 172             }
 173             else if ( !ForesterUtil.isEmpty( node.getName() ) ) {
 174                 if ( ( query = SequenceDbWsTools.parseUniProtAccessor( node.getName() ) ) != null ) {
 175                     db = Db.UNIPROT;
 176                 }
 177                 else if ( ( id = SequenceIdParser.parse( node.getName() ) ) != null ) {
 178
 179                     if ( id.getProvider().equalsIgnoreCase( Identifier.NCBI ) ) {
 180                         db = Db.NCBI;
 181                     }
 182                     else if ( id.getProvider().equalsIgnoreCase( Identifier.REFSEQ ) ) {
 183                         db = Db.REFSEQ;
 184                     }
 185
 186                 }
 187             }
 188             SequenceDatabaseEntry db_entry = null;
 189             if ( !ForesterUtil.isEmpty( query ) ) {
 190                 if ( db == Db.UNIPROT ) {
 191                     if ( DEBUG ) {
 192                         System.out.println( "uniprot: " + query );
 193                     }
 194                     db_entry = SequenceDbWsTools.obtainUniProtEntry( query, DEFAULT_LINES_TO_RETURN );
 195                 }
 196                 if ( ( db == Db.EMBL ) || ( ( db == Db.UNIPROT ) && ( db_entry == null ) ) ) {
 197                     if ( DEBUG ) {
 198                         System.out.println( "embl: " + query );
 199                     }
 200                     db_entry = SequenceDbWsTools.obtainEmblEntry(  new Identifier( query ), DEFAULT_LINES_TO_RETURN );
 201                     if ( ( db == Db.UNIPROT ) && ( db_entry != null ) ) {
 202                         db = Db.EMBL;
 203                     }
 204                 }
 205             }
 206             else if ( ( db == Db.REFSEQ ) && ( id != null ) ) {
 207                 db_entry = SequenceDbWsTools.obtainRefSeqEntryFromEmbl( id, DEFAULT_LINES_TO_RETURN );
 208             }
 209             else if ( ( db == Db.NCBI ) && ( id != null ) ) {
 210                 db_entry = SequenceDbWsTools.obtainEmblEntry( id, DEFAULT_LINES_TO_RETURN );
 211             }
 212             if ( ( db_entry != null ) && !db_entry.isEmpty() ) {
 213                 if ( !ForesterUtil.isEmpty( db_entry.getAccession() ) ) {
 214                     String type = null;
 215                     if ( db == Db.EMBL ) {
 216                         type = "embl";
 217                     }
 218                     else if ( db == Db.UNIPROT ) {
 219                         type = "uniprot";
 220                     }
 221                     else if ( db == Db.NCBI ) {
 222                         type = "ncbi";
 223                     }
 224                     else if ( db == Db.REFSEQ ) {
 225                         type = "refseq";
 226                     }
 227                     seq.setAccession( new Accession( db_entry.getAccession(), type ) );
 228                 }
 229                 if ( !ForesterUtil.isEmpty( db_entry.getSequenceName() ) ) {
 230                     seq.setName( db_entry.getSequenceName() );
 231                 }
 232                 if ( !ForesterUtil.isEmpty( db_entry.getSequenceSymbol() ) ) {
 233                     seq.setSymbol( db_entry.getSequenceSymbol() );
 234                 }
 235                 if ( !ForesterUtil.isEmpty( db_entry.getTaxonomyScientificName() ) ) {
 236                     tax.setScientificName( db_entry.getTaxonomyScientificName() );
 237                 }
 238                 if ( !ForesterUtil.isEmpty( db_entry.getTaxonomyIdentifier() ) ) {
 239                     tax.setIdentifier( new Identifier( db_entry.getTaxonomyIdentifier(), "uniprot" ) );
 240                 }
 241                 node.getNodeData().setTaxonomy( tax );
 242                 node.getNodeData().setSequence( seq );
 243             }
 244             else if ( db != Db.NONE ) {
 245                 not_found.add( node.getName() );
 246             }
 247         }
 248         return not_found;
 249     }
 250
 251     @Override
 252     public void run() {
 253         execute();
 254     }
 255 }