forester/java/src/org/forester/tools/PhylogenyDecorator.java

   1 // $Id:
   2 // FORESTER -- software libraries and applications
   3 // for evolutionary biology research and applications.
   4 //
   5 // Copyright (C) 2008-2009 Christian M. Zmasek
   6 // Copyright (C) 2008-2009 Burnham Institute for Medical Research
   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.tools;
  27
  28 import java.io.File;
  29 import java.io.IOException;
  30 import java.util.HashMap;
  31 import java.util.Map;
  32 import java.util.regex.Matcher;
  33 import java.util.regex.Pattern;
  34
  35 import org.forester.archaeopteryx.AptxUtil;
  36 import org.forester.io.parsers.nhx.NHXFormatException;
  37 import org.forester.io.parsers.phyloxml.PhyloXmlDataFormatException;
  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.Annotation;
  42 import org.forester.phylogeny.data.DomainArchitecture;
  43 import org.forester.phylogeny.data.Identifier;
  44 import org.forester.phylogeny.data.Sequence;
  45 import org.forester.phylogeny.iterators.PhylogenyNodeIterator;
  46 import org.forester.util.BasicTable;
  47 import org.forester.util.BasicTableParser;
  48 import org.forester.util.ForesterUtil;
  49
  50 public final class PhylogenyDecorator {
  51
  52     // From evoruby/lib/evo/apps/tseq_taxonomy_processor.rb:
  53     final private static String  TP_TAXONOMY_CODE                   = "TAXONOMY_CODE";
  54     final private static String  TP_TAXONOMY_ID                     = "TAXONOMY_ID";
  55     final private static String  TP_TAXONOMY_ID_PROVIDER            = "TAXONOMY_ID_PROVIDER";
  56     final private static String  TP_TAXONOMY_SN                     = "TAXONOMY_SN";
  57     final private static String  TP_TAXONOMY_CN                     = "TAXONOMY_CN";
  58     final private static String  TP_TAXONOMY_SYN                    = "TAXONOMY_SYN";
  59     final private static String  TP_SEQ_SYMBOL                      = "SEQ_SYMBOL";
  60     final private static String  TP_SEQ_ACCESSION                   = "SEQ_ACCESSION";
  61     final private static String  TP_SEQ_ACCESSION_SOURCE            = "SEQ_ACCESSION_SOURCE";
  62     final private static String  TP_SEQ_ANNOTATION_DESC             = "SEQ_ANNOTATION_DESC";
  63     final private static String  TP_SEQ_ANNOTATION_REF              = "SEQ_ANNOTATION_REF";
  64     final private static String  TP_SEQ_MOL_SEQ                     = "SEQ_MOL_SEQ";
  65     final private static String  TP_SEQ_NAME                        = "SEQ_NAME";
  66     final private static String  TP_NODE_NAME                       = "NODE_NAME";
  67     final private static Pattern NODENAME_SEQNUMBER_TAXDOMAINNUMBER = Pattern
  68                                                                             .compile( "^([a-fA-Z0-9]{1,5})_([A-Z0-9]{2,4}[A-Z])(\\d{1,4})$" );
  69     public final static boolean  SANITIZE                           = false;
  70     public final static boolean  VERBOSE                            = true;
  71     private static final boolean CUT                                = true;
  72
  73     private PhylogenyDecorator() {
  74         // Not needed.
  75     }
  76
  77     public static void decorate( final Phylogeny phylogeny,
  78                                  final Map<String, Map<String, String>> map,
  79                                  final boolean picky,
  80                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map )
  81             throws IllegalArgumentException, PhyloXmlDataFormatException {
  82         for( final PhylogenyNodeIterator iter = phylogeny.iteratorPostorder(); iter.hasNext(); ) {
  83             final PhylogenyNode node = iter.next();
  84             final String name = node.getName();
  85             if ( !ForesterUtil.isEmpty( name ) ) {
  86                 if ( map.containsKey( name ) || ( numbers_of_chars_allowed_to_remove_if_not_found_in_map > 0 ) ) {
  87                     Map<String, String> new_values = map.get( name );
  88                     int x = 0;
  89                     while ( ( new_values == null ) && ( numbers_of_chars_allowed_to_remove_if_not_found_in_map > 0 )
  90                             && ( x <= numbers_of_chars_allowed_to_remove_if_not_found_in_map ) ) {
  91                         new_values = map.get( name.substring( 0, name.length() - x ) );
  92                         ++x;
  93                     }
  94                     if ( new_values != null ) {
  95                         if ( new_values.containsKey( TP_TAXONOMY_CODE ) ) {
  96                             AptxUtil.ensurePresenceOfTaxonomy( node );
  97                             node.getNodeData().getTaxonomy().setTaxonomyCode( new_values.get( TP_TAXONOMY_CODE ) );
  98                         }
  99                         if ( new_values.containsKey( TP_TAXONOMY_ID )
 100                                 && new_values.containsKey( TP_TAXONOMY_ID_PROVIDER ) ) {
 101                             AptxUtil.ensurePresenceOfTaxonomy( node );
 102                             node.getNodeData()
 103                                     .getTaxonomy()
 104                                     .setIdentifier( new Identifier( new_values.get( TP_TAXONOMY_ID ),
 105                                                                     new_values.get( TP_TAXONOMY_ID_PROVIDER ) ) );
 106                         }
 107                         else if ( new_values.containsKey( TP_TAXONOMY_ID ) ) {
 108                             AptxUtil.ensurePresenceOfTaxonomy( node );
 109                             node.getNodeData().getTaxonomy()
 110                                     .setIdentifier( new Identifier( new_values.get( TP_TAXONOMY_ID ) ) );
 111                         }
 112                         if ( new_values.containsKey( TP_TAXONOMY_SN ) ) {
 113                             AptxUtil.ensurePresenceOfTaxonomy( node );
 114                             node.getNodeData().getTaxonomy().setScientificName( new_values.get( TP_TAXONOMY_SN ) );
 115                         }
 116                         if ( new_values.containsKey( TP_TAXONOMY_CN ) ) {
 117                             AptxUtil.ensurePresenceOfTaxonomy( node );
 118                             node.getNodeData().getTaxonomy().setCommonName( new_values.get( TP_TAXONOMY_CN ) );
 119                         }
 120                         if ( new_values.containsKey( TP_TAXONOMY_SYN ) ) {
 121                             AptxUtil.ensurePresenceOfTaxonomy( node );
 122                             node.getNodeData().getTaxonomy().getSynonyms().add( new_values.get( TP_TAXONOMY_SYN ) );
 123                         }
 124                         if ( new_values.containsKey( TP_SEQ_ACCESSION )
 125                                 && new_values.containsKey( TP_SEQ_ACCESSION_SOURCE ) ) {
 126                             AptxUtil.ensurePresenceOfSequence( node );
 127                             node.getNodeData()
 128                                     .getSequence()
 129                                     .setAccession( new Accession( new_values.get( TP_SEQ_ACCESSION ),
 130                                                                   new_values.get( TP_SEQ_ACCESSION_SOURCE ) ) );
 131                         }
 132                         if ( new_values.containsKey( TP_SEQ_ANNOTATION_DESC ) ) {
 133                             AptxUtil.ensurePresenceOfSequence( node );
 134                             final Annotation ann = new Annotation( "?" );
 135                             ann.setDesc( new_values.get( TP_SEQ_ANNOTATION_DESC ) );
 136                             node.getNodeData().getSequence().addAnnotation( ann );
 137                         }
 138                         if ( new_values.containsKey( TP_SEQ_ANNOTATION_REF ) ) {
 139                             AptxUtil.ensurePresenceOfSequence( node );
 140                             final Annotation ann = new Annotation( new_values.get( TP_SEQ_ANNOTATION_REF ) );
 141                             node.getNodeData().getSequence().addAnnotation( ann );
 142                         }
 143                         if ( new_values.containsKey( TP_SEQ_SYMBOL ) ) {
 144                             AptxUtil.ensurePresenceOfSequence( node );
 145                             node.getNodeData().getSequence().setSymbol( new_values.get( TP_SEQ_SYMBOL ) );
 146                         }
 147                         if ( new_values.containsKey( TP_SEQ_NAME ) ) {
 148                             AptxUtil.ensurePresenceOfSequence( node );
 149                             node.getNodeData().getSequence().setName( new_values.get( TP_SEQ_NAME ) );
 150                         }
 151                         if ( new_values.containsKey( TP_SEQ_MOL_SEQ ) ) {
 152                             AptxUtil.ensurePresenceOfSequence( node );
 153                             node.getNodeData().getSequence().setMolecularSequence( new_values.get( TP_SEQ_MOL_SEQ ) );
 154                         }
 155                         if ( new_values.containsKey( TP_NODE_NAME ) ) {
 156                             node.setName( new_values.get( TP_NODE_NAME ) );
 157                         }
 158                     } // if ( new_values != null )
 159                 } // if ( map.containsKey( name ) || ( numbers_of_chars_allowed_to_remove_if_not_found_in_map > 0 ) )
 160                 else if ( picky ) {
 161                     throw new IllegalArgumentException( "\"" + name + "\" not found in name map" );
 162                 }
 163             }
 164         }
 165     }
 166
 167     /**
 168      *
 169      *
 170      *
 171      *
 172      *
 173      * @param phylogeny
 174      * @param map
 175      *            maps names (in phylogeny) to new values
 176      * @param field
 177      * @param picky
 178      * @throws IllegalArgumentException
 179      * @throws NHXFormatException
 180      * @throws PhyloXmlDataFormatException
 181      */
 182     public static void decorate( final Phylogeny phylogeny,
 183                                  final Map<String, String> map,
 184                                  final FIELD field,
 185                                  final boolean extract_bracketed_scientific_name,
 186                                  final boolean picky,
 187                                  final boolean cut_name_after_space,
 188                                  final boolean process_name_intelligently,
 189                                  final boolean process_similar_to,
 190                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 191                                  final boolean move_domain_numbers_at_end_to_middle ) throws IllegalArgumentException,
 192             NHXFormatException, PhyloXmlDataFormatException {
 193         PhylogenyDecorator.decorate( phylogeny,
 194                                      map,
 195                                      field,
 196                                      extract_bracketed_scientific_name,
 197                                      picky,
 198                                      null,
 199                                      cut_name_after_space,
 200                                      process_name_intelligently,
 201                                      process_similar_to,
 202                                      numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 203                                      move_domain_numbers_at_end_to_middle );
 204     }
 205
 206     /**
 207      *
 208      *
 209      *
 210      * @param phylogeny
 211      * @param map
 212      *            maps names (in phylogeny) to new values if intermediate_map is
 213      *            null otherwise maps intermediate value to new value
 214      * @param field
 215      * @param picky
 216      * @param intermediate_map
 217      *            maps name (in phylogeny) to a intermediate value
 218      * @throws IllegalArgumentException
 219      * @throws PhyloXmlDataFormatException
 220      */
 221     public static void decorate( final Phylogeny phylogeny,
 222                                  final Map<String, String> map,
 223                                  final FIELD field,
 224                                  final boolean extract_bracketed_scientific_name,
 225                                  final boolean picky,
 226                                  final Map<String, String> intermediate_map,
 227                                  final boolean cut_name_after_space,
 228                                  final boolean process_name_intelligently,
 229                                  final boolean process_similar_to,
 230                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 231                                  final boolean move_domain_numbers_at_end_to_middle ) throws IllegalArgumentException,
 232             PhyloXmlDataFormatException {
 233         if ( extract_bracketed_scientific_name && ( field == FIELD.TAXONOMY_SCIENTIFIC_NAME ) ) {
 234             throw new IllegalArgumentException( "Attempt to extract bracketed scientific name together with data field pointing to scientific name" );
 235         }
 236         for( final PhylogenyNodeIterator iter = phylogeny.iteratorPostorder(); iter.hasNext(); ) {
 237             final PhylogenyNode node = iter.next();
 238             String name = node.getName();
 239             if ( !ForesterUtil.isEmpty( name ) ) {
 240                 if ( intermediate_map != null ) {
 241                     name = PhylogenyDecorator.extractIntermediate( intermediate_map, name );
 242                 }
 243                 // int space_index = name.indexOf( " " );
 244                 //                if ( CUT && space_index > 0 ) {
 245                 //                    int y = name.lastIndexOf( "|" );
 246                 //                    name = name.substring( y + 1, space_index );
 247                 //                }
 248                 //                String new_value = null;
 249                 //                for( String key : map.keySet() ) {
 250                 //                    if ( key.indexOf( name ) >= 0 ) {
 251                 //                        if ( new_value == null ) {
 252                 //                            new_value = map.get( key );
 253                 //                        }
 254                 //                        else {
 255                 //                            System.out.println( name + " is not unique" );
 256                 //                            System.exit( -1 );
 257                 //                        }
 258                 //                    }
 259                 //                }
 260                 // if ( new_value != null ) {
 261                 if ( map.containsKey( name ) || ( numbers_of_chars_allowed_to_remove_if_not_found_in_map > 0 ) ) {
 262                     String new_value = map.get( name );
 263                     int x = 0;
 264                     while ( ( new_value == null ) && ( numbers_of_chars_allowed_to_remove_if_not_found_in_map > 0 )
 265                             && ( x <= numbers_of_chars_allowed_to_remove_if_not_found_in_map ) ) {
 266                         new_value = map.get( name.substring( 0, name.length() - x ) );
 267                         ++x;
 268                     }
 269                     if ( new_value != null ) {
 270                         new_value = new_value.trim();
 271                         new_value.replaceAll( "/\\s+/", " " );
 272                         if ( extract_bracketed_scientific_name && new_value.endsWith( "]" ) ) {
 273                             extractBracketedScientificNames( node, new_value );
 274                         }
 275                         switch ( field ) {
 276                             case SEQUENCE_ANNOTATION_DESC:
 277                                 if ( PhylogenyDecorator.VERBOSE ) {
 278                                     System.out.println( name + ": " + new_value );
 279                                 }
 280                                 if ( !node.getNodeData().isHasSequence() ) {
 281                                     node.getNodeData().setSequence( new Sequence() );
 282                                 }
 283                                 final Annotation annotation = new Annotation( "?" );
 284                                 annotation.setDesc( new_value );
 285                                 node.getNodeData().getSequence().addAnnotation( annotation );
 286                                 break;
 287                             case DOMAIN_STRUCTURE:
 288                                 if ( PhylogenyDecorator.VERBOSE ) {
 289                                     System.out.println( name + ": " + new_value );
 290                                 }
 291                                 if ( !node.getNodeData().isHasSequence() ) {
 292                                     node.getNodeData().setSequence( new Sequence() );
 293                                 }
 294                                 node.getNodeData().getSequence()
 295                                         .setDomainArchitecture( new DomainArchitecture( new_value ) );
 296                                 break;
 297                             case TAXONOMY_CODE:
 298                                 if ( PhylogenyDecorator.VERBOSE ) {
 299                                     System.out.println( name + ": " + new_value );
 300                                 }
 301                                 AptxUtil.ensurePresenceOfTaxonomy( node );
 302                                 node.getNodeData().getTaxonomy().setTaxonomyCode( new_value );
 303                                 break;
 304                             case TAXONOMY_SCIENTIFIC_NAME:
 305                                 if ( PhylogenyDecorator.VERBOSE ) {
 306                                     System.out.println( name + ": " + new_value );
 307                                 }
 308                                 AptxUtil.ensurePresenceOfTaxonomy( node );
 309                                 node.getNodeData().getTaxonomy().setScientificName( new_value );
 310                                 break;
 311                             case SEQUENCE_NAME:
 312                                 if ( PhylogenyDecorator.VERBOSE ) {
 313                                     System.out.println( name + ": " + new_value );
 314                                 }
 315                                 if ( !node.getNodeData().isHasSequence() ) {
 316                                     node.getNodeData().setSequence( new Sequence() );
 317                                 }
 318                                 node.getNodeData().getSequence().setName( new_value );
 319                                 break;
 320                             case NODE_NAME:
 321                                 if ( PhylogenyDecorator.VERBOSE ) {
 322                                     System.out.print( name + " -> " );
 323                                 }
 324                                 if ( cut_name_after_space ) {
 325                                     if ( PhylogenyDecorator.VERBOSE ) {
 326                                         System.out.print( new_value + " -> " );
 327                                     }
 328                                     new_value = PhylogenyDecorator.deleteAtFirstSpace( new_value );
 329                                 }
 330                                 else if ( process_name_intelligently ) {
 331                                     if ( PhylogenyDecorator.VERBOSE ) {
 332                                         System.out.print( new_value + " -> " );
 333                                     }
 334                                     new_value = PhylogenyDecorator.processNameIntelligently( new_value );
 335                                 }
 336                                 else if ( process_similar_to ) {
 337                                     if ( PhylogenyDecorator.VERBOSE ) {
 338                                         System.out.print( new_value + " -> " );
 339                                     }
 340                                     new_value = PhylogenyDecorator.processSimilarTo( new_value );
 341                                 }
 342                                 if ( PhylogenyDecorator.SANITIZE ) {
 343                                     new_value = PhylogenyDecorator.sanitize( new_value );
 344                                 }
 345                                 if ( PhylogenyDecorator.VERBOSE ) {
 346                                     System.out.println( new_value );
 347                                 }
 348                                 node.setName( new_value );
 349                                 break;
 350                             default:
 351                                 throw new RuntimeException( "unknown field \"" + field + "\"" );
 352                         }
 353                         if ( move_domain_numbers_at_end_to_middle && ( field != FIELD.NODE_NAME ) ) {
 354                             node.setName( moveDomainNumbersAtEnd( node.getName() ) );
 355                         }
 356                     }
 357                 }
 358                 else if ( picky ) {
 359                     throw new IllegalArgumentException( "\"" + name + "\" not found in name map" );
 360                 }
 361             }
 362         }
 363     }
 364
 365     public static void decorate( final Phylogeny[] phylogenies,
 366                                  final Map<String, Map<String, String>> map,
 367                                  final boolean picky,
 368                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map )
 369             throws IllegalArgumentException, NHXFormatException, PhyloXmlDataFormatException {
 370         for( int i = 0; i < phylogenies.length; ++i ) {
 371             PhylogenyDecorator.decorate( phylogenies[ i ],
 372                                          map,
 373                                          picky,
 374                                          numbers_of_chars_allowed_to_remove_if_not_found_in_map );
 375         }
 376     }
 377
 378     public static void decorate( final Phylogeny[] phylogenies,
 379                                  final Map<String, String> map,
 380                                  final FIELD field,
 381                                  final boolean extract_bracketed_scientific_name,
 382                                  final boolean picky,
 383                                  final boolean cut_name_after_space,
 384                                  final boolean process_name_intelligently,
 385                                  final boolean process_similar_to,
 386                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 387                                  final boolean move_domain_numbers_at_end_to_middle ) throws IllegalArgumentException,
 388             NHXFormatException, PhyloXmlDataFormatException {
 389         for( int i = 0; i < phylogenies.length; ++i ) {
 390             PhylogenyDecorator.decorate( phylogenies[ i ],
 391                                          map,
 392                                          field,
 393                                          extract_bracketed_scientific_name,
 394                                          picky,
 395                                          cut_name_after_space,
 396                                          process_name_intelligently,
 397                                          process_similar_to,
 398                                          numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 399                                          move_domain_numbers_at_end_to_middle );
 400         }
 401     }
 402
 403     public static void decorate( final Phylogeny[] phylogenies,
 404                                  final Map<String, String> map,
 405                                  final FIELD field,
 406                                  final boolean extract_bracketed_scientific_name,
 407                                  final boolean picky,
 408                                  final Map<String, String> intermediate_map,
 409                                  final boolean cut_name_after_space,
 410                                  final boolean process_name_intelligently,
 411                                  final boolean process_similar_to,
 412                                  final int numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 413                                  final boolean move_domain_numbers_at_end_to_middle ) throws IllegalArgumentException,
 414             NHXFormatException, PhyloXmlDataFormatException {
 415         for( int i = 0; i < phylogenies.length; ++i ) {
 416             PhylogenyDecorator.decorate( phylogenies[ i ],
 417                                          map,
 418                                          field,
 419                                          extract_bracketed_scientific_name,
 420                                          picky,
 421                                          intermediate_map,
 422                                          cut_name_after_space,
 423                                          process_name_intelligently,
 424                                          process_similar_to,
 425                                          numbers_of_chars_allowed_to_remove_if_not_found_in_map,
 426                                          move_domain_numbers_at_end_to_middle );
 427         }
 428     }
 429
 430     private static String deleteAtFirstSpace( final String name ) {
 431         final int first_space = name.indexOf( " " );
 432         if ( first_space > 1 ) {
 433             return name.substring( 0, first_space ).trim();
 434         }
 435         return name;
 436     }
 437
 438     private static void extractBracketedScientificNames( final PhylogenyNode node, final String new_value ) {
 439         final int i = new_value.lastIndexOf( "[" );
 440         final String scientific_name = new_value.substring( i + 1, new_value.length() - 1 );
 441         AptxUtil.ensurePresenceOfTaxonomy( node );
 442         node.getNodeData().getTaxonomy().setScientificName( scientific_name );
 443     }
 444
 445     private static String extractIntermediate( final Map<String, String> intermediate_map, final String name ) {
 446         String new_name = null;
 447         if ( PhylogenyDecorator.VERBOSE ) {
 448             System.out.print( name + " => " );
 449         }
 450         if ( intermediate_map.containsKey( name ) ) {
 451             new_name = intermediate_map.get( name );
 452             if ( ForesterUtil.isEmpty( new_name ) ) {
 453                 throw new IllegalArgumentException( "\"" + name + "\" maps to null or empty string in secondary map" );
 454             }
 455         }
 456         else {
 457             throw new IllegalArgumentException( "\"" + name + "\" not found in name secondary map" );
 458         }
 459         if ( PhylogenyDecorator.VERBOSE ) {
 460             System.out.println( new_name + "  " );
 461         }
 462         return new_name;
 463     }
 464
 465     private static String moveDomainNumbersAtEnd( final String node_name ) {
 466         final Matcher m = NODENAME_SEQNUMBER_TAXDOMAINNUMBER.matcher( node_name );
 467         if ( m.matches() ) {
 468             final String seq_number = m.group( 1 );
 469             final String tax = m.group( 2 );
 470             final String domain_number = m.group( 3 );
 471             return seq_number + "_[" + domain_number + "]_" + tax;
 472         }
 473         else {
 474             return node_name;
 475         }
 476     }
 477
 478     public static Map<String, Map<String, String>> parseMappingTable( final File mapping_table_file )
 479             throws IOException {
 480         final Map<String, Map<String, String>> map = new HashMap<String, Map<String, String>>();
 481         BasicTable<String> mapping_table = null;
 482         mapping_table = BasicTableParser.parse( mapping_table_file, "\t", false );
 483         for( int row = 0; row < mapping_table.getNumberOfRows(); ++row ) {
 484             final Map<String, String> row_map = new HashMap<String, String>();
 485             String name = null;
 486             for( int col = 0; col < mapping_table.getNumberOfColumns(); ++col ) {
 487                 final String table_cell = mapping_table.getValue( col, row );
 488                 if ( col == 0 ) {
 489                     name = table_cell;
 490                 }
 491                 else if ( table_cell != null ) {
 492                     final String key = table_cell.substring( 0, table_cell.indexOf( ':' ) );
 493                     final String val = table_cell.substring( table_cell.indexOf( ':' ) + 1, table_cell.length() );
 494                     row_map.put( key, val );
 495                 }
 496             }
 497             map.put( name, row_map );
 498         }
 499         return map;
 500     }
 501
 502     private static String processNameIntelligently( final String name ) {
 503         final String[] s = name.split( " " );
 504         if ( s.length < 2 ) {
 505             return name;
 506         }
 507         else if ( ( s[ 0 ].indexOf( "_" ) > 0 ) && ( s[ 0 ].indexOf( "|" ) > 0 ) ) {
 508             return s[ 0 ];
 509         }
 510         else if ( ( s[ 1 ].indexOf( "_" ) > 0 ) && ( s[ 1 ].indexOf( "|" ) > 0 ) ) {
 511             return s[ 1 ];
 512         }
 513         else if ( ( s[ 0 ].indexOf( "_" ) > 0 ) && ( s[ 0 ].indexOf( "." ) > 0 ) ) {
 514             return s[ 0 ];
 515         }
 516         else if ( ( s[ 1 ].indexOf( "_" ) > 0 ) && ( s[ 1 ].indexOf( "." ) > 0 ) ) {
 517             return s[ 1 ];
 518         }
 519         else if ( s[ 0 ].indexOf( "_" ) > 0 ) {
 520             return s[ 0 ];
 521         }
 522         else if ( s[ 1 ].indexOf( "_" ) > 0 ) {
 523             return s[ 1 ];
 524         }
 525         else {
 526             return s[ 0 ];
 527         }
 528     }
 529
 530     private static String processSimilarTo( final String name ) {
 531         final int i = name.toLowerCase().indexOf( "similar to" );
 532         String similar_to = "";
 533         if ( i >= 0 ) {
 534             similar_to = " similarity=" + name.substring( i + 10 ).trim();
 535         }
 536         final String pi = processNameIntelligently( name );
 537         return pi + similar_to;
 538     }
 539
 540     private static String sanitize( String s ) {
 541         s = s.replace( ' ', '_' );
 542         s = s.replace( '(', '{' );
 543         s = s.replace( ')', '}' );
 544         s = s.replace( '[', '{' );
 545         s = s.replace( ']', '}' );
 546         s = s.replace( ',', '_' );
 547         return s;
 548     }
 549
 550     public static enum FIELD {
 551         NODE_NAME, SEQUENCE_ANNOTATION_DESC, DOMAIN_STRUCTURE, TAXONOMY_CODE, TAXONOMY_SCIENTIFIC_NAME, SEQUENCE_NAME;
 552     }
 553 }