forester/java/src/org/forester/surfacing/BasicDomainSimilarityCalculator.java

   1 // $Id:
   2 // Exp $
   3 //
   4 // FORESTER -- software libraries and applications
   5 // for evolutionary biology research and applications.
   6 //
   7 // Copyright (C) 2008-2009 Christian M. Zmasek
   8 // Copyright (C) 2008-2009 Burnham Institute for Medical Research
   9 // All rights reserved
  10 //
  11 // This library is free software; you can redistribute it and/or
  12 // modify it under the terms of the GNU Lesser General Public
  13 // License as published by the Free Software Foundation; either
  14 // version 2.1 of the License, or (at your option) any later version.
  15 //
  16 // This library is distributed in the hope that it will be useful,
  17 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  18 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  19 // Lesser General Public License for more details.
  20 //
  21 // You should have received a copy of the GNU Lesser General Public
  22 // License along with this library; if not, write to the Free Software
  23 // Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
  24 //
  25 // Contact: phylosoft @ gmail . com
  26 // WWW: www.phylosoft.org/forester
  27
  28 package org.forester.surfacing;
  29
  30 import java.util.ArrayList;
  31 import java.util.List;
  32 import java.util.SortedMap;
  33 import java.util.SortedSet;
  34 import java.util.TreeMap;
  35 import java.util.TreeSet;
  36
  37 import org.forester.util.BasicDescriptiveStatistics;
  38 import org.forester.util.DescriptiveStatistics;
  39
  40 public class BasicDomainSimilarityCalculator implements DomainSimilarityCalculator {
  41
  42     final DomainSimilarity.DomainSimilaritySortField _sort;
  43     private final boolean                            _sort_by_species_count_first;
  44     private final boolean                            _treat_as_binary_comparison;
  45
  46     public BasicDomainSimilarityCalculator( final DomainSimilarity.DomainSimilaritySortField sort,
  47                                             final boolean sort_by_species_count_first,
  48                                             final boolean treat_as_binary_comparison ) {
  49         _sort = sort;
  50         _sort_by_species_count_first = sort_by_species_count_first;
  51         _treat_as_binary_comparison = treat_as_binary_comparison;
  52     }
  53
  54     public SortedSet<DomainSimilarity> calculateSimilarities( final PairwiseDomainSimilarityCalculator pairwise_calculator,
  55                                                               final List<GenomeWideCombinableDomains> cdc_list,
  56                                                               final boolean ignore_domains_without_combinations_in_any_genome,
  57                                                               final boolean ignore_domains_specific_to_one_genome ) {
  58         if ( cdc_list.size() < 2 ) {
  59             throw new IllegalArgumentException( "attempt to calculate multiple combinable domains similarity for less than two combinale domains collections" );
  60         }
  61         final SortedSet<DomainSimilarity> similarities = new TreeSet<DomainSimilarity>();
  62         final SortedSet<DomainId> keys = new TreeSet<DomainId>();
  63         for( final GenomeWideCombinableDomains cdc : cdc_list ) {
  64             keys.addAll( ( cdc ).getAllCombinableDomainsIds().keySet() );
  65         }
  66         for( final DomainId key : keys ) {
  67             final List<CombinableDomains> same_id_cd_list = new ArrayList<CombinableDomains>( cdc_list.size() );
  68             final List<Species> species_with_key_id_domain = new ArrayList<Species>();
  69             for( final GenomeWideCombinableDomains cdc : cdc_list ) {
  70                 if ( cdc.contains( key ) ) {
  71                     same_id_cd_list.add( cdc.get( key ) );
  72                     species_with_key_id_domain.add( cdc.getSpecies() );
  73                 }
  74             }
  75             if ( ignore_domains_without_combinations_in_any_genome ) { //TODO: test me..........................................<<<<<<<<<<<<<
  76                 boolean without_combinations = true;
  77                 for( final CombinableDomains cd : same_id_cd_list ) {
  78                     if ( cd.getNumberOfCombinableDomains() > 0 ) {
  79                         without_combinations = false;
  80                         break;
  81                     }
  82                 }
  83                 if ( without_combinations ) {
  84                     continue;
  85                 }
  86             }
  87             // BIG CHANGE IN LOGIC: Tuesday July 08, 0;55
  88             // ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  89             // OLD: if ( same_id_cd_list.size() > 1 ) {
  90             if ( same_id_cd_list.size() > 0 ) {
  91                 if ( !ignore_domains_specific_to_one_genome || ( same_id_cd_list.size() > 1 ) ) {
  92                     final DomainSimilarity s = calculateSimilarity( pairwise_calculator, same_id_cd_list );
  93                     if ( s != null ) {
  94                         similarities.add( s );
  95                     }
  96                     else {
  97                         throw new RuntimeException( "similarity is null: this should not have happened" );
  98                     }
  99                 }
 100             }
 101             // ~~~ NEW:
 102             else {
 103                 throw new RuntimeException( "this should not have happened" );
 104             }
 105             // ~~~ OLD:
 106             // else if ( same_id_cd_list.size() == 1 ) {
 107             // TODO need to go in file
 108             // System.out.println( "only in one species [" +
 109             // species_with_key_id_domain.get( 0 ) + "]: " + key_id );
 110             //}
 111             //else {
 112             //    throw new RuntimeException( "this should not have happened" );
 113             // }
 114         }
 115         return similarities;
 116     }
 117
 118     private DomainSimilarity calculateSimilarity( final PairwiseDomainSimilarityCalculator pairwise_calculator,
 119                                                   final List<CombinableDomains> domains_list ) {
 120         if ( domains_list.size() == 1 ) {
 121             // BIG CHANGE IN LOGIC: Tuesday July 08, 0;55
 122             // ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 123             // ~~~OLD:
 124             //throw new IllegalArgumentException( "attempt to calculate multiple combinable domains similarity for less than two combinable domains" );
 125             // ~~~new:
 126             final SortedMap<Species, SpeciesSpecificDomainSimilariyData> species_data = new TreeMap<Species, SpeciesSpecificDomainSimilariyData>();
 127             species_data.put( domains_list.get( 0 ).getSpecies(),
 128                               createSpeciesSpecificDomainSimilariyData( domains_list.get( 0 ) ) );
 129             return new PrintableDomainSimilarity( domains_list.get( 0 ),
 130                                                   1.0,
 131                                                   1.0,
 132                                                   1.0,
 133                                                   1.0,
 134                                                   0.0,
 135                                                   0,
 136                                                   0,
 137                                                   0,
 138                                                   species_data,
 139                                                   getSort(),
 140                                                   isSortBySpeciesCountFirst(),
 141                                                   isTreatAsBinaryComparison() );
 142         }
 143         final DescriptiveStatistics stat = new BasicDescriptiveStatistics();
 144         final SortedMap<Species, SpeciesSpecificDomainSimilariyData> species_data = new TreeMap<Species, SpeciesSpecificDomainSimilariyData>();
 145         species_data.put( domains_list.get( 0 ).getSpecies(), createSpeciesSpecificDomainSimilariyData( domains_list
 146                 .get( 0 ) ) );
 147         int max_difference_in_counts = 0;
 148         int max_difference = 0;
 149         final boolean is_domain_combination_based = pairwise_calculator instanceof CombinationsBasedPairwiseDomainSimilarityCalculator;
 150         for( int i = 1; i < domains_list.size(); ++i ) {
 151             species_data.put( domains_list.get( i ).getSpecies(),
 152                               createSpeciesSpecificDomainSimilariyData( domains_list.get( i ) ) );
 153             final CombinableDomains domains_i = domains_list.get( i );
 154             for( int j = 0; j < i; ++j ) {
 155                 final PairwiseDomainSimilarity pairwise_similarity = pairwise_calculator
 156                         .calculateSimilarity( domains_i, domains_list.get( j ) );
 157                 final int difference_in_counts = pairwise_similarity.getDifferenceInCounts();
 158                 int difference = 0;
 159                 if ( is_domain_combination_based ) {
 160                     difference = ( ( CombinationsBasedPairwiseDomainSimilarity ) pairwise_similarity )
 161                             .getNumberOfDifferentDomains();
 162                 }
 163                 else {
 164                     difference = difference_in_counts;
 165                 }
 166                 if ( Math.abs( difference_in_counts ) > Math.abs( max_difference_in_counts ) ) {
 167                     max_difference_in_counts = difference_in_counts;
 168                 }
 169                 if ( Math.abs( difference ) > Math.abs( max_difference ) ) {
 170                     max_difference = difference;
 171                 }
 172                 stat.addValue( pairwise_similarity.getSimilarityScore() );
 173             }
 174         }
 175         if ( stat.getN() < 1 ) {
 176             throw new AssertionError( "empty descriptive statistics: this should not have happened" );
 177         }
 178         if ( ( stat.getN() != 1 ) && isTreatAsBinaryComparison() ) {
 179             throw new IllegalArgumentException( "attmpt to treat similarity with N not equal to one as binary comparison" );
 180         }
 181         if ( ( /*stat.getN() != 1 ||*/!isTreatAsBinaryComparison() ) && ( max_difference_in_counts < 0 ) ) {
 182             max_difference_in_counts = Math.abs( max_difference_in_counts );
 183             if ( !is_domain_combination_based ) {
 184                 max_difference = Math.abs( max_difference ); //=max_difference_in_counts for !is_domain_combination_based.
 185             }
 186         }
 187         DomainSimilarity similarity = null;
 188         if ( stat.getN() == 1 ) {
 189             similarity = new PrintableDomainSimilarity( domains_list.get( 0 ),
 190                                                         stat.getMin(),
 191                                                         stat.getMax(),
 192                                                         stat.arithmeticMean(),
 193                                                         stat.median(),
 194                                                         0.0,
 195                                                         stat.getN(),
 196                                                         max_difference_in_counts,
 197                                                         max_difference,
 198                                                         species_data,
 199                                                         getSort(),
 200                                                         isSortBySpeciesCountFirst(),
 201                                                         isTreatAsBinaryComparison() );
 202         }
 203         else {
 204             similarity = new PrintableDomainSimilarity( domains_list.get( 0 ),
 205                                                         stat.getMin(),
 206                                                         stat.getMax(),
 207                                                         stat.arithmeticMean(),
 208                                                         stat.median(),
 209                                                         stat.sampleStandardDeviation(),
 210                                                         stat.getN(),
 211                                                         max_difference_in_counts,
 212                                                         max_difference,
 213                                                         species_data,
 214                                                         getSort(),
 215                                                         isSortBySpeciesCountFirst(),
 216                                                         isTreatAsBinaryComparison() );
 217         }
 218         return similarity;
 219     }
 220
 221     private DomainSimilarity.DomainSimilaritySortField getSort() {
 222         return _sort;
 223     }
 224
 225     private boolean isSortBySpeciesCountFirst() {
 226         return _sort_by_species_count_first;
 227     }
 228
 229     private boolean isTreatAsBinaryComparison() {
 230         return _treat_as_binary_comparison;
 231     }
 232
 233     private static SpeciesSpecificDomainSimilariyData createSpeciesSpecificDomainSimilariyData( final CombinableDomains cd ) {
 234         final SpeciesSpecificDomainSimilariyData sd = new PrintableSpeciesSpecificDomainSimilariyData( cd
 235                 .getKeyDomainProteinsCount(), cd.getKeyDomainCount(), cd.getNumberOfCombinableDomains(), cd
 236                 .getKeyDomainConfidenceDescriptiveStatistics() );
 237         for( final DomainId domain : cd.getCombinableDomains() ) {
 238             sd.addProteinsExhibitingCombinationCount( domain, cd.getNumberOfProteinsExhibitingCombination( domain ) );
 239         }
 240         return sd;
 241     }
 242 }