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