forester/java/src/org/forester/archaeopteryx/tools/UniProtSequenceObtainer.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.List;
  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.UniProtEntry;
  47 import org.forester.ws.uniprot.UniProtWsTools;
  48
  49 public class UniProtSequenceObtainer implements Runnable {
  50
  51     private final Phylogeny            _phy;
  52     private final MainFrameApplication _mf;
  53     private final TreePanel            _treepanel;
  54
  55     public UniProtSequenceObtainer( final MainFrameApplication mf, final TreePanel treepanel, final Phylogeny phy ) {
  56         _phy = phy;
  57         _mf = mf;
  58         _treepanel = treepanel;
  59     }
  60
  61     private String getBaseUrl() {
  62         return UniProtWsTools.BASE_URL;
  63     }
  64
  65     private void execute() {
  66         _mf.getMainPanel().getCurrentTreePanel().setWaitCursor();
  67         SortedSet<String> not_found = null;
  68         try {
  69             not_found = obtainSeqInformation( _phy );
  70         }
  71         catch ( final UnknownHostException e ) {
  72             _mf.getMainPanel().getCurrentTreePanel().setArrowCursor();
  73             JOptionPane.showMessageDialog( _mf,
  74                                            "Could not connect to \"" + getBaseUrl() + "\"",
  75                                            "Network error during taxonomic information gathering",
  76                                            JOptionPane.ERROR_MESSAGE );
  77             return;
  78         }
  79         catch ( final IOException e ) {
  80             _mf.getMainPanel().getCurrentTreePanel().setArrowCursor();
  81             e.printStackTrace();
  82             JOptionPane.showMessageDialog( _mf,
  83                                            e.toString(),
  84                                            "Failed to obtain taxonomic information",
  85                                            JOptionPane.ERROR_MESSAGE );
  86             return;
  87         }
  88         finally {
  89             _mf.getMainPanel().getCurrentTreePanel().setArrowCursor();
  90         }
  91         _treepanel.setTree( _phy );
  92         _mf.showWhole();
  93         _treepanel.setEdited( true );
  94         if ( ( not_found != null ) && ( not_found.size() > 0 ) ) {
  95             int max = not_found.size();
  96             boolean more = false;
  97             if ( max > 20 ) {
  98                 more = true;
  99                 max = 20;
 100             }
 101             final StringBuffer sb = new StringBuffer();
 102             sb.append( "Not all identifiers could be resolved.\n" );
 103             if ( not_found.size() == 1 ) {
 104                 sb.append( "The following identifier was not found:\n" );
 105             }
 106             else {
 107                 sb.append( "The following identifiers were not found (total: " + not_found.size() + "):\n" );
 108             }
 109             int i = 0;
 110             for( final String string : not_found ) {
 111                 if ( i > 19 ) {
 112                     break;
 113                 }
 114                 sb.append( string );
 115                 sb.append( "\n" );
 116                 ++i;
 117             }
 118             if ( more ) {
 119                 sb.append( "..." );
 120             }
 121             try {
 122                 JOptionPane.showMessageDialog( _mf,
 123                                                sb.toString(),
 124                                                "UniProt Sequence Tool Completed",
 125                                                JOptionPane.WARNING_MESSAGE );
 126             }
 127             catch ( final Exception e ) {
 128                 // Not important if this fails, do nothing.
 129             }
 130         }
 131         else {
 132             try {
 133                 JOptionPane.showMessageDialog( _mf,
 134                                                "UniProt sequence tool successfully completed",
 135                                                "UniProt Sequence Tool Completed",
 136                                                JOptionPane.INFORMATION_MESSAGE );
 137             }
 138             catch ( final Exception e ) {
 139                 // Not important if this fails, do nothing.
 140             }
 141         }
 142     }
 143
 144     synchronized public static SortedSet<String> obtainSeqInformation( final Phylogeny phy ) throws IOException {
 145         final SortedSet<String> not_found = new TreeSet<String>();
 146         for( final PhylogenyNodeIterator iter = phy.iteratorPostorder(); iter.hasNext(); ) {
 147             final PhylogenyNode node = iter.next();
 148             if ( node.getNodeData().isHasSequence() ) {
 149                 //TODO  Do something
 150             }
 151             //  else if ( !ForesterUtil.isEmpty( node.getName() ) ) {
 152             //     not_found.add( node.getName() );
 153             //  }
 154             else if ( !ForesterUtil.isEmpty( node.getName() ) ) {
 155                 String query = node.getName();
 156                 if ( query.indexOf( '/' ) > 0 ) {
 157                     query = query.substring( 0, query.indexOf( '/' ) );
 158                 }
 159                 if ( query.indexOf( '.' ) > 0 ) {
 160                     query = query.substring( 0, query.indexOf( '.' ) );
 161                 }
 162                 if ( query.indexOf( '_' ) > 0 ) {
 163                     query = query.substring( 0, query.indexOf( '_' ) );
 164                 }
 165                 final UniProtEntry upe = obtainUniProtEntry( query );
 166                 if ( upe != null ) {
 167                     final Sequence seq = new Sequence();
 168                     final Taxonomy tax = new Taxonomy();
 169                     if ( !ForesterUtil.isEmpty( upe.getAc() ) ) {
 170                         seq.setAccession( new Accession( upe.getAc(), "uniprot" ) );
 171                     }
 172                     if ( !ForesterUtil.isEmpty( upe.getRecName() ) ) {
 173                         seq.setName( upe.getRecName() );
 174                     }
 175                     if ( !ForesterUtil.isEmpty( upe.getSymbol() ) ) {
 176                         seq.setSymbol( upe.getSymbol() );
 177                     }
 178                     if ( !ForesterUtil.isEmpty( upe.getOsScientificName() ) ) {
 179                         tax.setScientificName( upe.getOsScientificName() );
 180                     }
 181                     if ( !ForesterUtil.isEmpty( upe.getTaxId() ) ) {
 182                         tax.setIdentifier( new Identifier( upe.getTaxId(), "uniprot" ) );
 183                     }
 184                     node.getNodeData().setTaxonomy( tax );
 185                     node.getNodeData().setSequence( seq );
 186                 }
 187                 else {
 188                     not_found.add( node.getName() );
 189                 }
 190                 //}
 191             }
 192         }
 193         return not_found;
 194     }
 195
 196     static UniProtEntry obtainUniProtEntry( final String query ) throws IOException {
 197         final List<String> lines = UniProtWsTools.queryUniprot( "uniprot/" + query + ".txt", 200 );
 198         return UniProtEntry.createInstanceFromPlainText( lines );
 199     }
 200
 201     @Override
 202     public void run() {
 203         execute();
 204     }
 205 }