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 import java.util.regex.Pattern;
  34
  35 import javax.swing.JOptionPane;
  36
  37 import org.forester.archaeopteryx.MainFrameApplication;
  38 import org.forester.archaeopteryx.TreePanel;
  39 import org.forester.phylogeny.Phylogeny;
  40 import org.forester.phylogeny.PhylogenyNode;
  41 import org.forester.phylogeny.data.Accession;
  42 import org.forester.phylogeny.data.Identifier;
  43 import org.forester.phylogeny.data.Sequence;
  44 import org.forester.phylogeny.data.Taxonomy;
  45 import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
  46 import org.forester.util.ForesterUtil;
  47 import org.forester.ws.uniprot.SequenceDatabaseEntry;
  48 import org.forester.ws.uniprot.UniProtWsTools;
  49
  50 public final class SequenceDataRetriver implements Runnable {
  51
  52     // uniprot/expasy accession number format (6 chars):
  53     // letter digit letter-or-digit letter-or-digit letter-or-digit digit
  54     private final static Pattern       UNIPROT_AC_PATTERN = Pattern.compile( "[A-NR-ZOPQ]\\d[A-Z0-9]{3}\\d" );
  55     private final Phylogeny            _phy;
  56     private final MainFrameApplication _mf;
  57     private final TreePanel            _treepanel;
  58     private final static boolean       DEBUG              = true;
  59
  60     private enum Db {
  61         UNKNOWN, UNIPROT;
  62     }
  63
  64     public SequenceDataRetriver( final MainFrameApplication mf, final TreePanel treepanel, final Phylogeny phy ) {
  65         _phy = phy;
  66         _mf = mf;
  67         _treepanel = treepanel;
  68     }
  69
  70     private String getBaseUrl() {
  71         return UniProtWsTools.BASE_URL;
  72     }
  73
  74     private void execute() {
  75         _mf.getMainPanel().getCurrentTreePanel().setWaitCursor();
  76         SortedSet<String> not_found = null;
  77         try {
  78             not_found = obtainSeqInformation( _phy );
  79         }
  80         catch ( final UnknownHostException e ) {
  81             _mf.getMainPanel().getCurrentTreePanel().setArrowCursor();
  82             JOptionPane.showMessageDialog( _mf,
  83                                            "Could not connect to \"" + getBaseUrl() + "\"",
  84                                            "Network error during taxonomic information gathering",
  85                                            JOptionPane.ERROR_MESSAGE );
  86             return;
  87         }
  88         catch ( final IOException e ) {
  89             _mf.getMainPanel().getCurrentTreePanel().setArrowCursor();
  90             e.printStackTrace();
  91             JOptionPane.showMessageDialog( _mf,
  92                                            e.toString(),
  93                                            "Failed to obtain taxonomic information",
  94                                            JOptionPane.ERROR_MESSAGE );
  95             return;
  96         }
  97         finally {
  98             _mf.getMainPanel().getCurrentTreePanel().setArrowCursor();
  99         }
 100         _treepanel.setTree( _phy );
 101         _mf.showWhole();
 102         _treepanel.setEdited( true );
 103         if ( ( not_found != null ) && ( not_found.size() > 0 ) ) {
 104             int max = not_found.size();
 105             boolean more = false;
 106             if ( max > 20 ) {
 107                 more = true;
 108                 max = 20;
 109             }
 110             final StringBuffer sb = new StringBuffer();
 111             sb.append( "Not all identifiers could be resolved.\n" );
 112             if ( not_found.size() == 1 ) {
 113                 sb.append( "The following identifier was not found:\n" );
 114             }
 115             else {
 116                 sb.append( "The following identifiers were not found (total: " + not_found.size() + "):\n" );
 117             }
 118             int i = 0;
 119             for( final String string : not_found ) {
 120                 if ( i > 19 ) {
 121                     break;
 122                 }
 123                 sb.append( string );
 124                 sb.append( "\n" );
 125                 ++i;
 126             }
 127             if ( more ) {
 128                 sb.append( "..." );
 129             }
 130             try {
 131                 JOptionPane.showMessageDialog( _mf,
 132                                                sb.toString(),
 133                                                "UniProt Sequence Tool Completed",
 134                                                JOptionPane.WARNING_MESSAGE );
 135             }
 136             catch ( final Exception e ) {
 137                 // Not important if this fails, do nothing.
 138             }
 139         }
 140         else {
 141             try {
 142                 JOptionPane.showMessageDialog( _mf,
 143                                                "UniProt sequence tool successfully completed",
 144                                                "UniProt Sequence Tool Completed",
 145                                                JOptionPane.INFORMATION_MESSAGE );
 146             }
 147             catch ( final Exception e ) {
 148                 // Not important if this fails, do nothing.
 149             }
 150         }
 151     }
 152
 153     synchronized public static SortedSet<String> obtainSeqInformation( final Phylogeny phy ) throws IOException {
 154         final SortedSet<String> not_found = new TreeSet<String>();
 155         for( final PhylogenyNodeIterator iter = phy.iteratorPostorder(); iter.hasNext(); ) {
 156             final PhylogenyNode node = iter.next();
 157             Sequence seq = null;
 158             Taxonomy tax = null;
 159             if ( node.getNodeData().isHasSequence() ) {
 160                 seq = node.getNodeData().getSequence();
 161             }
 162             else {
 163                 seq = new Sequence();
 164             }
 165             if ( node.getNodeData().isHasTaxonomy() ) {
 166                 tax = node.getNodeData().getTaxonomy();
 167             }
 168             else {
 169                 tax = new Taxonomy();
 170             }
 171             String query = null;
 172             Db db = Db.UNKNOWN;
 173             if ( node.getNodeData().isHasSequence() && ( node.getNodeData().getSequence().getAccession() != null )
 174                     && !ForesterUtil.isEmpty( node.getNodeData().getSequence().getAccession().getSource() )
 175                     && !ForesterUtil.isEmpty( node.getNodeData().getSequence().getAccession().getValue() )
 176                     && node.getNodeData().getSequence().getAccession().getValue().toLowerCase().startsWith( "uniprot" ) ) {
 177                 query = node.getNodeData().getSequence().getAccession().getValue();
 178                 db = Db.UNIPROT;
 179             }
 180             else if ( !ForesterUtil.isEmpty( node.getName() ) ) {
 181                 query = node.getName();
 182             }
 183             if ( !ForesterUtil.isEmpty( query ) ) {
 184                 if ( query.indexOf( '/' ) > 0 ) {
 185                     query = query.substring( 0, query.indexOf( '/' ) );
 186                 }
 187                 if ( query.indexOf( '.' ) > 0 ) {
 188                     query = query.substring( 0, query.indexOf( '.' ) );
 189                 }
 190                 if ( query.indexOf( '_' ) > 0 ) {
 191                     query = query.substring( 0, query.indexOf( '_' ) );
 192                 }
 193                 SequenceDatabaseEntry db_entry = null;
 194                 if ( ( db == Db.UNIPROT ) || UNIPROT_AC_PATTERN.matcher( query ).matches() ) {
 195                     if ( DEBUG ) {
 196                         System.out.println( "uniprot: " + query );
 197                     }
 198                     try {
 199                         db_entry = UniProtWsTools.obtainUniProtEntry( query, 200 );
 200                     }
 201                     catch ( final FileNotFoundException e ) {
 202                         // Ignore.
 203                     }
 204                 }
 205                 if ( db_entry != null ) {
 206                     if ( !ForesterUtil.isEmpty( db_entry.getAccession() ) ) {
 207                         seq.setAccession( new Accession( db_entry.getAccession(), "uniprot" ) );
 208                     }
 209                     if ( !ForesterUtil.isEmpty( db_entry.getSequenceName() ) ) {
 210                         seq.setName( db_entry.getSequenceName() );
 211                     }
 212                     if ( !ForesterUtil.isEmpty( db_entry.getSequenceSymbol() ) ) {
 213                         seq.setSymbol( db_entry.getSequenceSymbol() );
 214                     }
 215                     if ( !ForesterUtil.isEmpty( db_entry.getTaxonomyScientificName() ) ) {
 216                         tax.setScientificName( db_entry.getTaxonomyScientificName() );
 217                     }
 218                     if ( !ForesterUtil.isEmpty( db_entry.getTaxonomyIdentifier() ) ) {
 219                         tax.setIdentifier( new Identifier( db_entry.getTaxonomyIdentifier(), "uniprot" ) );
 220                     }
 221                     node.getNodeData().setTaxonomy( tax );
 222                     node.getNodeData().setSequence( seq );
 223                 }
 224                 else {
 225                     not_found.add( node.getName() );
 226                 }
 227             }
 228         }
 229         return not_found;
 230     }
 231
 232     @Override
 233     public void run() {
 234         execute();
 235     }
 236 }