|
0
|
1 #
|
|
|
2 # Ensembl module for Bio::EnsEMBL::Funcgen::ExperimentalSubset
|
|
|
3 #
|
|
|
4 # You may distribute this module under the same terms as Perl itself
|
|
|
5
|
|
|
6 =head1 NAME
|
|
|
7
|
|
|
8 Bio::EnsEMBL::ExperimentalSet - A module to represent ExperimentalSet object.
|
|
|
9
|
|
|
10
|
|
|
11 =head1 SYNOPSIS
|
|
|
12
|
|
|
13 use Bio::EnsEMBL::Funcgen::ExpeimentalSet;
|
|
|
14
|
|
|
15 my $data_set = Bio::EnsEMBL::Funcgen::ExperimentalSet->new(
|
|
|
16 -DBID => $dbID,
|
|
|
17 -ADAPTOR => $self,
|
|
|
18 -EXPERIMENT => $exp,
|
|
|
19 -FEATURE_TYPE => $ftype,
|
|
|
20 -CELL_TYPE => $ctype,
|
|
|
21 -FORMAT => 'READ_FORMAT',
|
|
|
22 -VENDOR => 'SOLEXA',
|
|
|
23 -NAME => 'ExpSet1',
|
|
|
24 );
|
|
|
25
|
|
|
26
|
|
|
27
|
|
|
28 =head1 DESCRIPTION
|
|
|
29
|
|
|
30 An ExperimentalSet object provides a generic container for any non-array based feature import,
|
|
|
31 allowing tracking of file import via the status table and integration into Data and FeatureSets to
|
|
|
32 provide traceability to the source experiment from a given FeatureSet.
|
|
|
33
|
|
|
34 =head1 AUTHOR
|
|
|
35
|
|
|
36 This module was created by Nathan Johnson.
|
|
|
37
|
|
|
38 This module is part of the Ensembl project: http://www.ensembl.org/
|
|
|
39
|
|
|
40 =head1 CONTACT
|
|
|
41
|
|
|
42 Post comments or questions to the Ensembl development list: ensembl-dev@ebi.ac.uk
|
|
|
43
|
|
|
44 =head1 METHODS
|
|
|
45
|
|
|
46 =cut
|
|
|
47
|
|
|
48 use strict;
|
|
|
49 use warnings;
|
|
|
50
|
|
|
51 package Bio::EnsEMBL::Funcgen::ExperimentalSet;
|
|
|
52
|
|
|
53 use Bio::EnsEMBL::Funcgen::ExperimentalSubset;
|
|
|
54 use Bio::EnsEMBL::Utils::Argument qw( rearrange );
|
|
|
55 use Bio::EnsEMBL::Utils::Exception qw( throw warning deprecate);
|
|
|
56 use Bio::EnsEMBL::Funcgen::Set;
|
|
|
57
|
|
|
58
|
|
|
59 use vars qw(@ISA);
|
|
|
60 @ISA = qw(Bio::EnsEMBL::Funcgen::Set);#change to Set once we have implemented analysis properly
|
|
|
61
|
|
|
62
|
|
|
63 =head2 new
|
|
|
64
|
|
|
65
|
|
|
66
|
|
|
67 Example : my $eset = Bio::EnsEMBL::Funcgen::ExperimentalSet->new(
|
|
|
68 -EXPERIMENT => $exp,
|
|
|
69 -FEATURE_TYPE => $ftype,
|
|
|
70 -CELL_TYPE => $ctype,
|
|
|
71 -FORMAT => 'READ_FORMAT',
|
|
|
72 -VENDOR => 'SOLEXA',
|
|
|
73 -NAME => 'ExpSet1',
|
|
|
74 -ANALYSIS => $anal,
|
|
|
75 );
|
|
|
76
|
|
|
77 Do we want to define subsets likes this or are we more likely to add them one by one?
|
|
|
78
|
|
|
79 Description: Constructor for ExperimentalSet objects.
|
|
|
80 Returntype : Bio::EnsEMBL::Funcgen::ExperimentalSet
|
|
|
81 Exceptions : Throws if no Experiment defined
|
|
|
82 Throws if CellType or FeatureType are not valid or stored
|
|
|
83 Caller : General
|
|
|
84 Status : At risk
|
|
|
85
|
|
|
86 =cut
|
|
|
87
|
|
|
88 sub new {
|
|
|
89 my $caller = shift;
|
|
|
90
|
|
|
91 my $class = ref($caller) || $caller;
|
|
|
92
|
|
|
93 my $self = $class->SUPER::new(@_);
|
|
|
94
|
|
|
95 #do we need to add $fg_ids to this? Currently maintaining one feature_group focus.(combi exps?)
|
|
|
96 my ($exp, $format, $vendor)
|
|
|
97 = rearrange(['EXPERIMENT', 'FORMAT', 'VENDOR'], @_);
|
|
|
98
|
|
|
99 if (! (ref $exp && $exp->isa('Bio::EnsEMBL::Funcgen::Experiment') && $exp->dbID())){
|
|
|
100 throw('Must specify a valid stored Bio::EnsEMBL::Funcgen::Experiment');
|
|
|
101 }
|
|
|
102
|
|
|
103
|
|
|
104 #These are set in Set, just validate here
|
|
|
105 throw ('Must provide a FeatureType') if(! defined $self->feature_type);
|
|
|
106 throw ('Must provide a CellType') if(! defined $self->cell_type);
|
|
|
107
|
|
|
108
|
|
|
109
|
|
|
110 if(! defined $self->analysis){
|
|
|
111 #default analysis hack for v47
|
|
|
112 #Set directly to avoid dbID boolean check
|
|
|
113 $self->{'analysis'} = Bio::EnsEMBL::Analysis->new(-logic_name => 'external',
|
|
|
114 -id => 0,#??someone needs to rewrite analysis
|
|
|
115 );
|
|
|
116 }
|
|
|
117
|
|
|
118 $self->format($format) if defined $format;
|
|
|
119 $self->vendor($vendor) if defined $vendor;
|
|
|
120 $self->{'experiment'} = $exp;
|
|
|
121 $self->{'subsets'} = {};
|
|
|
122
|
|
|
123 return $self;
|
|
|
124 }
|
|
|
125
|
|
|
126
|
|
|
127 =head2 add_new_subset
|
|
|
128
|
|
|
129 Arg [1] : string - sub set name e.g. the file name (not path as we're restricted to 30 chars)
|
|
|
130 Example : $expset->add_new_subset($ss_name, $exp_subset);
|
|
|
131 Description: Adds experimental_subset
|
|
|
132 Returntype : none
|
|
|
133 Exceptions : Throws if set is already present
|
|
|
134 Throws if ExperimentalSubset is not valid or stored
|
|
|
135 Caller : General
|
|
|
136 Status : At Risk
|
|
|
137
|
|
|
138 =cut
|
|
|
139
|
|
|
140 sub add_new_subset {
|
|
|
141 my ($self, $ss_name, $exp_sset) = @_;
|
|
|
142
|
|
|
143 if($self->get_subset_by_name($ss_name)){
|
|
|
144 throw("Subset $ss_name is already present in this ExperimentalSet, maybe you need to alter the filename?");
|
|
|
145 }
|
|
|
146
|
|
|
147 if(defined $exp_sset){
|
|
|
148
|
|
|
149 if(!(ref($exp_sset) && $exp_sset->isa('Bio::EnsEMBL::Funcgen::ExperimentalSubset') && $exp_sset->dbID())){
|
|
|
150 throw('ExperimentalSubsets must be valid and stored');
|
|
|
151 }
|
|
|
152 }
|
|
|
153 else{
|
|
|
154
|
|
|
155 $exp_sset = Bio::EnsEMBL::Funcgen::ExperimentalSubset->new(
|
|
|
156 -name => $ss_name,
|
|
|
157 -experimental_set => $self,
|
|
|
158 );
|
|
|
159 }
|
|
|
160
|
|
|
161 $self->{'subsets'}{$ss_name} = $exp_sset;
|
|
|
162
|
|
|
163 return $self->{'subsets'}{$ss_name};
|
|
|
164 }
|
|
|
165
|
|
|
166
|
|
|
167 =head2 get_Experiment
|
|
|
168
|
|
|
169 Example : my $exp = $exp_set->get_Experiment();
|
|
|
170 Description: Getter for the Experiment of this DataSet.
|
|
|
171 Returntype : Bio::EnsEMBL::Fuuncgen::Experiment
|
|
|
172 Exceptions : None
|
|
|
173 Caller : General
|
|
|
174 Status : At Risk
|
|
|
175
|
|
|
176 =cut
|
|
|
177
|
|
|
178 sub get_Experiment{
|
|
|
179 my $self = shift;
|
|
|
180
|
|
|
181 return $self->{'experiment'};
|
|
|
182 }
|
|
|
183
|
|
|
184
|
|
|
185 =head2 get_subsets
|
|
|
186
|
|
|
187 Example : my @subsets = @{$exp_set->get_subsets()};
|
|
|
188 Description: Getter for the subsets for this ExperimentalSet.
|
|
|
189 Returntype : Arrayref
|
|
|
190 Exceptions : None
|
|
|
191 Caller : General
|
|
|
192 Status : At Risk
|
|
|
193
|
|
|
194 =cut
|
|
|
195
|
|
|
196 sub get_subsets{
|
|
|
197 my ($self) = shift;
|
|
|
198
|
|
|
199 return [ values %{$self->{'subsets'}} ];
|
|
|
200 }
|
|
|
201
|
|
|
202 =head2 get_subset_by_name
|
|
|
203
|
|
|
204 Example : my $subsets = $exp_set->get_subset_by_name('subset1');
|
|
|
205 Description: Getter for the subset of a given name for this ExperimentalSet.
|
|
|
206 Returntype : Bio::EnsEMBL::Funcgen::ExpeirmentalSubset
|
|
|
207 Exceptions : None
|
|
|
208 Caller : General
|
|
|
209 Status : At Risk
|
|
|
210
|
|
|
211 =cut
|
|
|
212
|
|
|
213 sub get_subset_by_name{
|
|
|
214 my ($self, $name) = @_;
|
|
|
215
|
|
|
216 return (exists $self->{'subsets'}{$name}) ? $self->{'subsets'}{$name} : undef;
|
|
|
217 }
|
|
|
218
|
|
|
219 =head2 get_subset_names
|
|
|
220
|
|
|
221 Example : my @subset_names = @{$exp_set->get_subset_names()};
|
|
|
222 Description: Getter for the subset names for this ExperimentalSet.
|
|
|
223 Returntype : Arrayref
|
|
|
224 Exceptions : None
|
|
|
225 Caller : General
|
|
|
226 Status : At Risk
|
|
|
227
|
|
|
228 =cut
|
|
|
229
|
|
|
230 sub get_subset_names{
|
|
|
231 my ($self) = shift;
|
|
|
232
|
|
|
233 return [ keys %{$self->{'subsets'}} ];
|
|
|
234 }
|
|
|
235
|
|
|
236
|
|
|
237
|
|
|
238
|
|
|
239 =head2 vendor
|
|
|
240
|
|
|
241 Arg[1] : string - vendor
|
|
|
242 Example : my $eset->vendor('SOLEXA');
|
|
|
243 Description: Getter/Setter for the vendor attribute of this DataSet.
|
|
|
244 Returntype : string
|
|
|
245 Exceptions : None
|
|
|
246 Caller : General
|
|
|
247 Status : At Risk
|
|
|
248
|
|
|
249 =cut
|
|
|
250
|
|
|
251 sub vendor {
|
|
|
252 my $self = shift;
|
|
|
253
|
|
|
254 $self->{'vendor'} = shift if @_;
|
|
|
255
|
|
|
256 return $self->{'vendor'};
|
|
|
257 }
|
|
|
258
|
|
|
259
|
|
|
260 =head2 format
|
|
|
261
|
|
|
262 Arg[1] : string - format i.e. product type/format
|
|
|
263 Example : my $eset->format('DATASET1');
|
|
|
264 Description: Getter/Setter for the format attribute of this ExperimentalSet.
|
|
|
265 Returntype : string
|
|
|
266 Exceptions : None
|
|
|
267 Caller : General
|
|
|
268 Status : At Risk
|
|
|
269
|
|
|
270 =cut
|
|
|
271
|
|
|
272 sub format {
|
|
|
273 my $self = shift;
|
|
|
274
|
|
|
275 $self->{'format'} = shift if @_;
|
|
|
276
|
|
|
277 return $self->{'format'};
|
|
|
278 }
|
|
|
279
|
|
|
280 1;
|
|
|
281
|
