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.FileNotFoundException;
  29 import java.io.IOException;
  30 import java.net.UnknownHostException;
  31 import java.util.SortedSet;
  32 import java.util.TreeSet;
  33
  34 import javax.swing.JOptionPane;
  35
  36 import org.forester.archaeopteryx.MainFrameApplication;
  37 import org.forester.archaeopteryx.TreePanel;
  38 import org.forester.phylogeny.Phylogeny;
  39 import org.forester.phylogeny.PhylogenyNode;
  40 import org.forester.phylogeny.data.Accession;
  41 import org.forester.phylogeny.data.Identifier;
  42 import org.forester.phylogeny.data.Sequence;
  43 import org.forester.phylogeny.data.Taxonomy;
  44 import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
  45 import org.forester.util.ForesterUtil;
  46 import org.forester.ws.uniprot.DatabaseTools;
  47 import org.forester.ws.uniprot.SequenceDatabaseEntry;
  48 import org.forester.ws.uniprot.UniProtWsTools;
  49
  50 public final class SequenceDataRetriver implements Runnable {
  51
  52     private final Phylogeny            _phy;
  53     private final MainFrameApplication _mf;
  54     private final TreePanel            _treepanel;
  55     private final static boolean       DEBUG = false;
  56
  57     private enum Db {
  58         UNKNOWN, UNIPROT, EMBL;
  59     }
  60
  61     public SequenceDataRetriver( final MainFrameApplication mf, final TreePanel treepanel, final Phylogeny phy ) {
  62         _phy = phy;
  63         _mf = mf;
  64         _treepanel = treepanel;
  65     }
  66
  67     private String getBaseUrl() {
  68         return UniProtWsTools.BASE_URL;
  69     }
  70
  71     private void execute() {
  72         _mf.getMainPanel().getCurrentTreePanel().setWaitCursor();
  73         SortedSet<String> not_found = null;
  74         try {
  75             not_found = obtainSeqInformation( _phy );
  76         }
  77         catch ( final UnknownHostException e ) {
  78             _mf.getMainPanel().getCurrentTreePanel().setArrowCursor();
  79             JOptionPane.showMessageDialog( _mf,
  80                                            "Could not connect to \"" + getBaseUrl() + "\"",
  81                                            "Network error during taxonomic information gathering",
  82                                            JOptionPane.ERROR_MESSAGE );
  83             return;
  84         }
  85         catch ( final IOException e ) {
  86             _mf.getMainPanel().getCurrentTreePanel().setArrowCursor();
  87             e.printStackTrace();
  88             JOptionPane.showMessageDialog( _mf,
  89                                            e.toString(),
  90                                            "Failed to obtain taxonomic information",
  91                                            JOptionPane.ERROR_MESSAGE );
  92             return;
  93         }
  94         finally {
  95             _mf.getMainPanel().getCurrentTreePanel().setArrowCursor();
  96         }
  97         _treepanel.setTree( _phy );
  98         _mf.showWhole();
  99         _treepanel.setEdited( true );
 100         if ( ( not_found != null ) && ( not_found.size() > 0 ) ) {
 101             int max = not_found.size();
 102             boolean more = false;
 103             if ( max > 20 ) {
 104                 more = true;
 105                 max = 20;
 106             }
 107             final StringBuffer sb = new StringBuffer();
 108             if ( not_found.size() == 1 ) {
 109                 sb.append( "Data for the following sequence identifier was not found:\n" );
 110             }
 111             else {
 112                 sb.append( "Data for the following sequence identifiers was not found (total: " + not_found.size()
 113                         + "):\n" );
 114             }
 115             int i = 0;
 116             for( final String string : not_found ) {
 117                 if ( i > 19 ) {
 118                     break;
 119                 }
 120                 sb.append( string );
 121                 sb.append( "\n" );
 122                 ++i;
 123             }
 124             if ( more ) {
 125                 sb.append( "..." );
 126             }
 127             try {
 128                 JOptionPane.showMessageDialog( _mf,
 129                                                sb.toString(),
 130                                                "Sequence Tool Completed",
 131                                                JOptionPane.WARNING_MESSAGE );
 132             }
 133             catch ( final Exception e ) {
 134                 // Not important if this fails, do nothing.
 135             }
 136         }
 137         else {
 138             try {
 139                 JOptionPane.showMessageDialog( _mf,
 140                                                "UniProt sequence tool successfully completed",
 141                                                "UniProt Sequence Tool Completed",
 142                                                JOptionPane.INFORMATION_MESSAGE );
 143             }
 144             catch ( final Exception e ) {
 145                 // Not important if this fails, do nothing.
 146             }
 147         }
 148     }
 149
 150     public static SortedSet<String> obtainSeqInformation( final Phylogeny phy ) throws IOException {
 151         final SortedSet<String> not_found = new TreeSet<String>();
 152         for( final PhylogenyNodeIterator iter = phy.iteratorPostorder(); iter.hasNext(); ) {
 153             final PhylogenyNode node = iter.next();
 154             Sequence seq = null;
 155             Taxonomy tax = null;
 156             if ( node.getNodeData().isHasSequence() ) {
 157                 seq = node.getNodeData().getSequence();
 158             }
 159             else {
 160                 seq = new Sequence();
 161             }
 162             if ( node.getNodeData().isHasTaxonomy() ) {
 163                 tax = node.getNodeData().getTaxonomy();
 164             }
 165             else {
 166                 tax = new Taxonomy();
 167             }
 168             String query = null;
 169             Db db = Db.UNKNOWN;
 170             if ( node.getNodeData().isHasSequence() && ( node.getNodeData().getSequence().getAccession() != null )
 171                     && !ForesterUtil.isEmpty( node.getNodeData().getSequence().getAccession().getSource() )
 172                     && !ForesterUtil.isEmpty( node.getNodeData().getSequence().getAccession().getValue() )
 173                     && node.getNodeData().getSequence().getAccession().getValue().toLowerCase().startsWith( "uniprot" ) ) {
 174                 query = node.getNodeData().getSequence().getAccession().getValue();
 175                 db = Db.UNIPROT;
 176             }
 177             else if ( node.getNodeData().isHasSequence()
 178                     && ( node.getNodeData().getSequence().getAccession() != null )
 179                     && !ForesterUtil.isEmpty( node.getNodeData().getSequence().getAccession().getSource() )
 180                     && !ForesterUtil.isEmpty( node.getNodeData().getSequence().getAccession().getValue() )
 181                     && ( node.getNodeData().getSequence().getAccession().getValue().toLowerCase().startsWith( "embl" ) || node
 182                             .getNodeData().getSequence().getAccession().getValue().toLowerCase().startsWith( "ebi" ) ) ) {
 183                 query = node.getNodeData().getSequence().getAccession().getValue();
 184                 db = Db.EMBL;
 185             }
 186             else if ( !ForesterUtil.isEmpty( node.getName() ) ) {
 187                 if ( ( query = UniProtWsTools.parseUniProtAccessor( node.getName() ) ) != null ) {
 188                     db = Db.UNIPROT;
 189                 }
 190                 else if ( ( query = DatabaseTools.parseGenbankAccessor( node.getName() ) ) != null ) {
 191                     db = Db.EMBL;
 192                 }
 193             }
 194             if ( !ForesterUtil.isEmpty( query ) ) {
 195                 SequenceDatabaseEntry db_entry = null;
 196                 if ( db == Db.UNIPROT ) {
 197                     if ( DEBUG ) {
 198                         System.out.println( "uniprot: " + query );
 199                     }
 200                     try {
 201                         db_entry = UniProtWsTools.obtainUniProtEntry( query, 200 );
 202                     }
 203                     catch ( final FileNotFoundException e ) {
 204                         // Ignore.
 205                     }
 206                 }
 207                 if ( ( db == Db.EMBL ) || ( ( db == Db.UNIPROT ) && ( db_entry == null ) ) ) {
 208                     if ( DEBUG ) {
 209                         System.out.println( "embl: " + query );
 210                     }
 211                     try {
 212                         db_entry = UniProtWsTools.obtainEmblEntry( query, 200 );
 213                     }
 214                     catch ( final FileNotFoundException e ) {
 215                         // Ignore.
 216                     }
 217                     if ( ( db == Db.UNIPROT ) && ( db_entry != null ) ) {
 218                         db = Db.EMBL;
 219                     }
 220                 }
 221                 if ( ( db_entry != null ) && !db_entry.isEmpty() ) {
 222                     if ( !ForesterUtil.isEmpty( db_entry.getAccession() ) ) {
 223                         String type = null;
 224                         if ( db == Db.EMBL ) {
 225                             type = "embl";
 226                         }
 227                         else if ( db == Db.UNIPROT ) {
 228                             type = "uniprot";
 229                         }
 230                         seq.setAccession( new Accession( db_entry.getAccession(), type ) );
 231                     }
 232                     if ( !ForesterUtil.isEmpty( db_entry.getSequenceName() ) ) {
 233                         seq.setName( db_entry.getSequenceName() );
 234                     }
 235                     if ( !ForesterUtil.isEmpty( db_entry.getSequenceSymbol() ) ) {
 236                         seq.setSymbol( db_entry.getSequenceSymbol() );
 237                     }
 238                     if ( !ForesterUtil.isEmpty( db_entry.getTaxonomyScientificName() ) ) {
 239                         tax.setScientificName( db_entry.getTaxonomyScientificName() );
 240                     }
 241                     if ( !ForesterUtil.isEmpty( db_entry.getTaxonomyIdentifier() ) ) {
 242                         tax.setIdentifier( new Identifier( db_entry.getTaxonomyIdentifier(), "uniprot" ) );
 243                     }
 244                     node.getNodeData().setTaxonomy( tax );
 245                     node.getNodeData().setSequence( seq );
 246                 }
 247                 else {
 248                     not_found.add( node.getName() );
 249                 }
 250             }
 251         }
 252         return not_found;
 253     }
 254
 255     @Override
 256     public void run() {
 257         execute();
 258     }
 259 }