inprogress
[jalview.git] / 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: https://sites.google.com/site/cmzmasek/home/software/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.species.Species;
38 import org.forester.util.BasicDescriptiveStatistics;
39 import org.forester.util.DescriptiveStatistics;
40
41 public class BasicDomainSimilarityCalculator implements DomainSimilarityCalculator {
42
43     final DomainSimilarity.DomainSimilaritySortField _sort;
44     private final boolean                            _sort_by_species_count_first;
45     private final boolean                            _treat_as_binary_comparison;
46
47     public BasicDomainSimilarityCalculator( final DomainSimilarity.DomainSimilaritySortField sort,
48                                             final boolean sort_by_species_count_first,
49                                             final boolean treat_as_binary_comparison ) {
50         _sort = sort;
51         _sort_by_species_count_first = sort_by_species_count_first;
52         _treat_as_binary_comparison = treat_as_binary_comparison;
53     }
54
55     @Override
56     public SortedSet<DomainSimilarity> calculateSimilarities( final PairwiseDomainSimilarityCalculator pairwise_calculator,
57                                                               final List<GenomeWideCombinableDomains> cdc_list,
58                                                               final boolean ignore_domains_without_combinations_in_any_genome,
59                                                               final boolean ignore_domains_specific_to_one_genome ) {
60         if ( cdc_list.size() < 2 ) {
61             throw new IllegalArgumentException( "attempt to calculate multiple combinable domains similarity for less than two combinale domains collections" );
62         }
63         final SortedSet<DomainSimilarity> similarities = new TreeSet<DomainSimilarity>();
64         final SortedSet<String> keys = new TreeSet<String>();
65         for( final GenomeWideCombinableDomains cdc : cdc_list ) {
66             keys.addAll( ( cdc ).getAllCombinableDomainsIds().keySet() );
67         }
68         for( final String key : keys ) {
69             final List<CombinableDomains> same_id_cd_list = new ArrayList<CombinableDomains>( cdc_list.size() );
70             final List<Species> species_with_key_id_domain = new ArrayList<Species>();
71             for( final GenomeWideCombinableDomains cdc : cdc_list ) {
72                 if ( cdc.contains( key ) ) {
73                     same_id_cd_list.add( cdc.get( key ) );
74                     species_with_key_id_domain.add( cdc.getSpecies() );
75                 }
76             }
77             if ( ignore_domains_without_combinations_in_any_genome ) { //TODO: test me..........................................<<<<<<<<<<<<<
78                 boolean without_combinations = true;
79                 for( final CombinableDomains cd : same_id_cd_list ) {
80                     if ( cd.getNumberOfCombinableDomains() > 0 ) {
81                         without_combinations = false;
82                         break;
83                     }
84                 }
85                 if ( without_combinations ) {
86                     continue;
87                 }
88             }
89             // BIG CHANGE IN LOGIC: Tuesday July 08, 0;55
90             // ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
91             // OLD: if ( same_id_cd_list.size() > 1 ) {
92             if ( same_id_cd_list.size() > 0 ) {
93                 if ( !ignore_domains_specific_to_one_genome || ( same_id_cd_list.size() > 1 ) ) {
94                     final DomainSimilarity s = calculateSimilarity( pairwise_calculator, same_id_cd_list );
95                     if ( s != null ) {
96                         similarities.add( s );
97                     }
98                     else {
99                         throw new RuntimeException( "similarity is null: this should not have happened" );
100                     }
101                 }
102             }
103             // ~~~ NEW:
104             else {
105                 throw new RuntimeException( "this should not have happened" );
106             }
107             // ~~~ OLD:
108             // else if ( same_id_cd_list.size() == 1 ) {
109             // TODO need to go in file
110             // System.out.println( "only in one species [" +
111             // species_with_key_id_domain.get( 0 ) + "]: " + key_id );
112             //}
113             //else {
114             //    throw new RuntimeException( "this should not have happened" );
115             // }
116         }
117         return similarities;
118     }
119
120     private DomainSimilarity calculateSimilarity( final PairwiseDomainSimilarityCalculator pairwise_calculator,
121                                                   final List<CombinableDomains> domains_list ) {
122         if ( domains_list.size() == 1 ) {
123             // BIG CHANGE IN LOGIC: Tuesday July 08, 0;55
124             // ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
125             // ~~~OLD:
126             //throw new IllegalArgumentException( "attempt to calculate multiple combinable domains similarity for less than two combinable domains" );
127             // ~~~new: 
128             final SortedMap<Species, SpeciesSpecificDcData> species_data = new TreeMap<Species, SpeciesSpecificDcData>();
129             species_data.put( domains_list.get( 0 ).getSpecies(),
130                               createSpeciesSpecificDomainSimilariyData( domains_list.get( 0 ) ) );
131             return new PrintableDomainSimilarity( domains_list.get( 0 ),
132                                                   1.0,
133                                                   1.0,
134                                                   1.0,
135                                                   1.0,
136                                                   0.0,
137                                                   0,
138                                                   0,
139                                                   0,
140                                                   species_data,
141                                                   isSortBySpeciesCountFirst(),
142                                                   isTreatAsBinaryComparison() );
143         }
144         final DescriptiveStatistics stat = new BasicDescriptiveStatistics();
145         final SortedMap<Species, SpeciesSpecificDcData> species_data = new TreeMap<Species, SpeciesSpecificDcData>();
146         species_data.put( domains_list.get( 0 ).getSpecies(),
147                           createSpeciesSpecificDomainSimilariyData( domains_list.get( 0 ) ) );
148         int max_difference_in_counts = 0;
149         int max_difference = 0;
150         final boolean is_domain_combination_based = pairwise_calculator instanceof CombinationsBasedPairwiseDomainSimilarityCalculator;
151         for( int i = 1; i < domains_list.size(); ++i ) {
152             species_data.put( domains_list.get( i ).getSpecies(),
153                               createSpeciesSpecificDomainSimilariyData( domains_list.get( i ) ) );
154             final CombinableDomains domains_i = domains_list.get( i );
155             for( int j = 0; j < i; ++j ) {
156                 final PairwiseDomainSimilarity pairwise_similarity = pairwise_calculator
157                         .calculateSimilarity( domains_i, domains_list.get( j ) );
158                 final int difference_in_counts = pairwise_similarity.getDifferenceInCounts();
159                 int difference = 0;
160                 if ( is_domain_combination_based ) {
161                     difference = ( ( CombinationsBasedPairwiseDomainSimilarity ) pairwise_similarity )
162                             .getNumberOfDifferentDomains();
163                 }
164                 else {
165                     difference = difference_in_counts;
166                 }
167                 if ( Math.abs( difference_in_counts ) > Math.abs( max_difference_in_counts ) ) {
168                     max_difference_in_counts = difference_in_counts;
169                 }
170                 if ( Math.abs( difference ) > Math.abs( max_difference ) ) {
171                     max_difference = difference;
172                 }
173                 stat.addValue( pairwise_similarity.getSimilarityScore() );
174             }
175         }
176         if ( stat.getN() < 1 ) {
177             throw new AssertionError( "empty descriptive statistics: this should not have happened" );
178         }
179         if ( ( stat.getN() != 1 ) && isTreatAsBinaryComparison() ) {
180             throw new IllegalArgumentException( "attmpt to treat similarity with N not equal to one as binary comparison" );
181         }
182         if ( ( /*stat.getN() != 1 ||*/!isTreatAsBinaryComparison() ) && ( max_difference_in_counts < 0 ) ) {
183             max_difference_in_counts = Math.abs( max_difference_in_counts );
184             if ( !is_domain_combination_based ) {
185                 max_difference = Math.abs( max_difference ); //=max_difference_in_counts for !is_domain_combination_based.
186             }
187         }
188         DomainSimilarity similarity = null;
189         if ( stat.getN() == 1 ) {
190             similarity = new PrintableDomainSimilarity( domains_list.get( 0 ),
191                                                         stat.getMin(),
192                                                         stat.getMax(),
193                                                         stat.arithmeticMean(),
194                                                         stat.median(),
195                                                         0.0,
196                                                         stat.getN(),
197                                                         max_difference_in_counts,
198                                                         max_difference,
199                                                         species_data,
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                                                         isSortBySpeciesCountFirst(),
215                                                         isTreatAsBinaryComparison() );
216         }
217         return similarity;
218     }
219
220     private boolean isSortBySpeciesCountFirst() {
221         return _sort_by_species_count_first;
222     }
223
224     private boolean isTreatAsBinaryComparison() {
225         return _treat_as_binary_comparison;
226     }
227
228     private static SpeciesSpecificDcData createSpeciesSpecificDomainSimilariyData( final CombinableDomains cd ) {
229         final SpeciesSpecificDcData sd = new PrintableSpeciesSpecificDcData( cd.getKeyDomainCount(),
230                                                                              cd.getNumberOfCombinableDomains() );
231         for( final String prot : cd.getKeyDomainProteins() ) {
232             sd.addKeyDomainProtein( prot );
233         }
234         for( final String domain : cd.getCombinableDomains() ) {
235             sd.addProteinsExhibitingCombinationCount( domain, cd.getNumberOfProteinsExhibitingCombination( domain ) );
236         }
237         return sd;
238     }
239 }