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.SequenceDatabaseEntry;
  47 import org.forester.ws.uniprot.UniProtWsTools;
  48
  49 public final class SequenceDataRetriver implements Runnable {
  50
  51     private final Phylogeny            _phy;
  52     private final MainFrameApplication _mf;
  53     private final TreePanel            _treepanel;
  54     private final static boolean       DEBUG = true;
  55
  56     private enum Db {
  57         UNKNOWN, UNIPROT;
  58     }
  59
  60     public SequenceDataRetriver( final MainFrameApplication mf, final TreePanel treepanel, final Phylogeny phy ) {
  61         _phy = phy;
  62         _mf = mf;
  63         _treepanel = treepanel;
  64     }
  65
  66     private String getBaseUrl() {
  67         return UniProtWsTools.BASE_URL;
  68     }
  69
  70     private void execute() {
  71         _mf.getMainPanel().getCurrentTreePanel().setWaitCursor();
  72         SortedSet<String> not_found = null;
  73         try {
  74             not_found = obtainSeqInformation( _phy );
  75         }
  76         catch ( final UnknownHostException e ) {
  77             _mf.getMainPanel().getCurrentTreePanel().setArrowCursor();
  78             JOptionPane.showMessageDialog( _mf,
  79                                            "Could not connect to \"" + getBaseUrl() + "\"",
  80                                            "Network error during taxonomic information gathering",
  81                                            JOptionPane.ERROR_MESSAGE );
  82             return;
  83         }
  84         catch ( final IOException e ) {
  85             _mf.getMainPanel().getCurrentTreePanel().setArrowCursor();
  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             _mf.getMainPanel().getCurrentTreePanel().setArrowCursor();
  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             sb.append( "Not all identifiers could be resolved.\n" );
 108             if ( not_found.size() == 1 ) {
 109                 sb.append( "The following identifier was not found:\n" );
 110             }
 111             else {
 112                 sb.append( "The following identifiers were not found (total: " + not_found.size() + "):\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                                                "UniProt 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 ( !ForesterUtil.isEmpty( node.getName() ) ) {
 177                 query = UniProtWsTools.parseUniProtAccessor( node.getName() );
 178                 if ( !ForesterUtil.isEmpty( query ) ) {
 179                     db = Db.UNIPROT;
 180                 }
 181             }
 182             if ( !ForesterUtil.isEmpty( query ) ) {
 183                 SequenceDatabaseEntry db_entry = null;
 184                 if ( db == Db.UNIPROT ) {
 185                     if ( DEBUG ) {
 186                         System.out.println( "uniprot: " + query );
 187                     }
 188                     try {
 189                         db_entry = UniProtWsTools.obtainUniProtEntry( query, 200 );
 190                     }
 191                     catch ( final FileNotFoundException e ) {
 192                         // Ignore.
 193                     }
 194                 }
 195                 if ( db_entry != null ) {
 196                     if ( !ForesterUtil.isEmpty( db_entry.getAccession() ) ) {
 197                         seq.setAccession( new Accession( db_entry.getAccession(), "uniprot" ) );
 198                     }
 199                     if ( !ForesterUtil.isEmpty( db_entry.getSequenceName() ) ) {
 200                         seq.setName( db_entry.getSequenceName() );
 201                     }
 202                     if ( !ForesterUtil.isEmpty( db_entry.getSequenceSymbol() ) ) {
 203                         seq.setSymbol( db_entry.getSequenceSymbol() );
 204                     }
 205                     if ( !ForesterUtil.isEmpty( db_entry.getTaxonomyScientificName() ) ) {
 206                         tax.setScientificName( db_entry.getTaxonomyScientificName() );
 207                     }
 208                     if ( !ForesterUtil.isEmpty( db_entry.getTaxonomyIdentifier() ) ) {
 209                         tax.setIdentifier( new Identifier( db_entry.getTaxonomyIdentifier(), "uniprot" ) );
 210                     }
 211                     node.getNodeData().setTaxonomy( tax );
 212                     node.getNodeData().setSequence( seq );
 213                 }
 214                 else {
 215                     not_found.add( node.getName() );
 216                 }
 217             }
 218         }
 219         return not_found;
 220     }
 221
 222     @Override
 223     public void run() {
 224         execute();
 225     }
 226 }