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