forester/java/src/org/forester/ws/seqdb/UniProtTaxonomy.java

   1 // $Id:
   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.ws.seqdb;
  27
  28 import java.util.ArrayList;
  29 import java.util.List;
  30
  31 import org.forester.util.ForesterUtil;
  32
  33 public final class UniProtTaxonomy {
  34
  35     private static final String         ARCHAEA                  = "Archaea";
  36     private static final String         BACTERIA                 = "Bacteria";
  37     private static final String         EUKARYOTA                = "Eukaryota";
  38     private final List<String>          _lineage;
  39     private final String                _code;
  40     private final String                _scientific_name;
  41     private final String                _common_name;
  42     private final String                _synonym;
  43     private final String                _rank;
  44     private final String                _id;
  45     public final static String          CELLULAR_ORGANISMS       = "cellular organisms";
  46     public final static String          VIRUSES                  = "Viruses";
  47     public final static UniProtTaxonomy DROSOPHILA_GENUS         = new UniProtTaxonomy( new String[] {
  48             CELLULAR_ORGANISMS, EUKARYOTA, "Metazoa", "Ecdysozoa", "Arthropoda", "Hexapoda", "Insecta", "Pterygota",
  49             "Neoptera", "Endopterygota", "Diptera", "Brachycera", "Muscomorpha", "Ephydroidea", "Drosophilidae",
  50             "Drosophila"                                                               },
  51                                                                                         "",
  52                                                                                         "fruit flies",
  53                                                                                         "Drosophila",
  54                                                                                         "",
  55                                                                                         "genus",
  56                                                                                         "7215" );
  57     public final static UniProtTaxonomy XENOPUS_GENUS            = new UniProtTaxonomy( new String[] {
  58             CELLULAR_ORGANISMS, EUKARYOTA, "Metazoa", "Chordata", "Craniata", "Vertebrata", "Euteleostomi", "Amphibia",
  59             "Batrachia", "Anura", "Mesobatrachia", "Pipoidea", "Pipidae", "Xenopodinae", "Xenopus" },
  60                                                                                         "",
  61                                                                                         "",
  62                                                                                         "Xenopus",
  63                                                                                         "",
  64                                                                                         "genus",
  65                                                                                         "8353" );
  66     public final static UniProtTaxonomy CAPITELLA_TELATA_SPECIES = new UniProtTaxonomy( new String[] {
  67             CELLULAR_ORGANISMS, EUKARYOTA, "Metazoa", "Annelida", "Polychaeta", "Scolecida", "Capitellida",
  68             "Capitellidae", "Capitella", "Capitella teleta"                            },
  69                                                                                         "",
  70                                                                                         "",
  71                                                                                         "Capitella teleta",
  72                                                                                         "Capitella sp. I",
  73                                                                                         "species",
  74                                                                                         "283909" );
  75
  76     public UniProtTaxonomy( final String line ) {
  77         final String[] items = line.split( "\t" );
  78         if ( items.length < 5 ) {
  79             throw new IllegalArgumentException( "cannot parse uniprot taxonomy from: " + line );
  80         }
  81         _id = items[ 0 ].trim();
  82         _code = items[ 1 ].trim();
  83         _scientific_name = items[ 2 ].trim();
  84         _common_name = items[ 3 ].trim();
  85         _synonym = items[ 4 ].trim();
  86         if ( items.length > 6 ) {
  87             _rank = items[ 7 ].trim();
  88         }
  89         else {
  90             _rank = "";
  91         }
  92         String[] lin = null;
  93         if ( items.length > 8 ) {
  94             lin = items[ 8 ].split( "; " );
  95         }
  96         _lineage = new ArrayList<String>();
  97         if ( ( lin != null ) && ( lin.length > 0 ) ) {
  98             final List<String> temp = new ArrayList<String>();
  99             for( final String t : lin ) {
 100                 if ( !ForesterUtil.isEmpty( t ) ) {
 101                     temp.add( t.trim() );
 102                 }
 103             }
 104             for( int i = 0; i < temp.size(); ++i ) {
 105                 if ( ( i == 0 )
 106                         && ( temp.get( i ).equalsIgnoreCase( EUKARYOTA ) || temp.get( i ).equalsIgnoreCase( BACTERIA ) || temp
 107                                 .get( i ).equalsIgnoreCase( ARCHAEA ) ) ) {
 108                     _lineage.add( CELLULAR_ORGANISMS );
 109                 }
 110                 _lineage.add( temp.get( i ) );
 111             }
 112         }
 113         if ( _lineage.isEmpty()
 114                 && ( _scientific_name.equalsIgnoreCase( EUKARYOTA ) || _scientific_name.equalsIgnoreCase( BACTERIA ) || _scientific_name
 115                         .equalsIgnoreCase( ARCHAEA ) ) ) {
 116             _lineage.add( CELLULAR_ORGANISMS );
 117         }
 118         _lineage.add( _scientific_name );
 119         if ( _lineage.isEmpty() ) {
 120             throw new IllegalArgumentException( "lineage in a UniProt taxonomy can not be empty\n: " + line );
 121         }
 122     }
 123
 124     public UniProtTaxonomy( final List<String> lineage,
 125                             final String code,
 126                             final String common_name,
 127                             final String scientific_name,
 128                             final String synonym,
 129                             final String rank,
 130                             final String id ) {
 131         _lineage = lineage;
 132         _code = code;
 133         _scientific_name = scientific_name;
 134         _common_name = common_name;
 135         _synonym = synonym;
 136         _rank = rank;
 137         _id = id;
 138         if ( ( _lineage != null ) && !_lineage.get( _lineage.size() - 1 ).equalsIgnoreCase( _scientific_name ) ) {
 139             _lineage.add( _scientific_name );
 140         }
 141     }
 142
 143     public UniProtTaxonomy( final String[] lineage,
 144                             final String code,
 145                             final String common_name,
 146                             final String scientific_name,
 147                             final String synonym,
 148                             final String rank,
 149                             final String id ) {
 150         _lineage = new ArrayList<String>();
 151         if ( lineage != null ) {
 152             for( final String l : lineage ) {
 153                 _lineage.add( l );
 154             }
 155         }
 156         _code = code;
 157         _scientific_name = scientific_name;
 158         _common_name = common_name;
 159         _synonym = synonym;
 160         _rank = rank;
 161         _id = id;
 162         if ( ( _lineage != null ) && !_lineage.get( _lineage.size() - 1 ).equalsIgnoreCase( _scientific_name ) ) {
 163             _lineage.add( _scientific_name );
 164         }
 165     }
 166
 167     /**
 168      * Creates deep copy for all fields, except lineage.
 169      *
 170      * @return
 171      */
 172     public UniProtTaxonomy copy() {
 173         return new UniProtTaxonomy( getLineage(),
 174                                     getCode() != null ? new String( getCode() ) : null,
 175                                     getCommonName() != null ? new String( getCommonName() ) : null,
 176                                     getScientificName() != null ? new String( getScientificName() ) : null,
 177                                     getSynonym() != null ? new String( getSynonym() ) : null,
 178                                     getRank() != null ? new String( getRank() ) : null,
 179                                     getId() != null ? new String( getId() ) : null );
 180     }
 181
 182     public String getCode() {
 183         return _code;
 184     }
 185
 186     public String getCommonName() {
 187         return _common_name;
 188     }
 189
 190     public String getId() {
 191         return _id;
 192     }
 193
 194     public List<String> getLineage() {
 195         return _lineage;
 196     }
 197
 198     public String getRank() {
 199         return _rank;
 200     }
 201
 202     public String getScientificName() {
 203         return _scientific_name;
 204     }
 205
 206     public String getSynonym() {
 207         return _synonym;
 208     }
 209 }