|
0
|
1 # $Id: FastHitEventBuilder.pm,v 1.6 2002/12/05 13:46:35 heikki Exp $
|
|
|
2 #
|
|
|
3 # BioPerl module for Bio::SearchIO::FastHitEventBuilder
|
|
|
4 #
|
|
|
5 # Cared for by Jason Stajich <jason@bioperl.org>
|
|
|
6 #
|
|
|
7 # Copyright Jason Stajich
|
|
|
8 #
|
|
|
9 # You may distribute this module under the same terms as perl itself
|
|
|
10
|
|
|
11 # POD documentation - main docs before the code
|
|
|
12
|
|
|
13 =head1 NAME
|
|
|
14
|
|
|
15 Bio::SearchIO::FastHitEventBuilder - Event Handler for SearchIO events.
|
|
|
16
|
|
|
17 =head1 SYNOPSIS
|
|
|
18
|
|
|
19 # Do not use this object directly, this object is part of the SearchIO
|
|
|
20 # event based parsing system.
|
|
|
21
|
|
|
22 # to use the FastHitEventBuilder do this
|
|
|
23
|
|
|
24 use Bio::SearchIO::FastHitEventBuilder;
|
|
|
25
|
|
|
26 my $searchio = new Bio::SearchIO(-format => $format, -file => $file);
|
|
|
27
|
|
|
28 $searchio->attach_EventHandler(new Bio::SearchIO::FastHitEventBuilder);
|
|
|
29
|
|
|
30 while( my $r = $searchio->next_result ) {
|
|
|
31 while( my $h = $r->next_hit ) {
|
|
|
32 # note that Hits will NOT have HSPs
|
|
|
33 }
|
|
|
34 }
|
|
|
35
|
|
|
36 =head1 DESCRIPTION
|
|
|
37
|
|
|
38 This object handles Search Events generated by the SearchIO classes
|
|
|
39 and build appropriate Bio::Search::* objects from them. This object
|
|
|
40 is intended for lightweight parsers which only want Hits and not deal
|
|
|
41 with the overhead of HSPs. It is a lot faster than the standard
|
|
|
42 parser event handler but of course you are getting less information
|
|
|
43 and less objects out.
|
|
|
44
|
|
|
45
|
|
|
46 =head1 FEEDBACK
|
|
|
47
|
|
|
48 =head2 Mailing Lists
|
|
|
49
|
|
|
50 User feedback is an integral part of the evolution of this and other
|
|
|
51 Bioperl modules. Send your comments and suggestions preferably to
|
|
|
52 the Bioperl mailing list. Your participation is much appreciated.
|
|
|
53
|
|
|
54 bioperl-l@bioperl.org - General discussion
|
|
|
55 http://bioperl.org/MailList.shtml - About the mailing lists
|
|
|
56
|
|
|
57 =head2 Reporting Bugs
|
|
|
58
|
|
|
59 Report bugs to the Bioperl bug tracking system to help us keep track
|
|
|
60 of the bugs and their resolution. Bug reports can be submitted via
|
|
|
61 email or the web:
|
|
|
62
|
|
|
63 bioperl-bugs@bioperl.org
|
|
|
64 http://bugzilla.bioperl.org/
|
|
|
65
|
|
|
66 =head1 AUTHOR - Jason Stajich
|
|
|
67
|
|
|
68 Email jason@bioperl.org
|
|
|
69
|
|
|
70 Describe contact details here
|
|
|
71
|
|
|
72 =head1 CONTRIBUTORS
|
|
|
73
|
|
|
74 Additional contributors names and emails here
|
|
|
75
|
|
|
76 =head1 APPENDIX
|
|
|
77
|
|
|
78 The rest of the documentation details each of the object methods.
|
|
|
79 Internal methods are usually preceded with a _
|
|
|
80
|
|
|
81 =cut
|
|
|
82
|
|
|
83
|
|
|
84 # Let the code begin...
|
|
|
85
|
|
|
86
|
|
|
87 package Bio::SearchIO::FastHitEventBuilder;
|
|
|
88 use vars qw(@ISA %KNOWNEVENTS);
|
|
|
89 use strict;
|
|
|
90
|
|
|
91 use Bio::Root::Root;
|
|
|
92 use Bio::SearchIO::EventHandlerI;
|
|
|
93 use Bio::Search::HSP::HSPFactory;
|
|
|
94 use Bio::Search::Hit::HitFactory;
|
|
|
95 use Bio::Search::Result::ResultFactory;
|
|
|
96
|
|
|
97 @ISA = qw(Bio::Root::Root Bio::SearchIO::EventHandlerI);
|
|
|
98
|
|
|
99 =head2 new
|
|
|
100
|
|
|
101 Title : new
|
|
|
102 Usage : my $obj = new Bio::SearchIO::FastHitEventBuilder();
|
|
|
103 Function: Builds a new Bio::SearchIO::FastHitEventBuilder object
|
|
|
104 Returns : Bio::SearchIO::FastHitEventBuilder
|
|
|
105 Args : -hit_factory => Bio::Factory::ObjectFactoryI
|
|
|
106 -result_factory => Bio::Factory::ObjectFactoryI
|
|
|
107
|
|
|
108 See L<Bio::Factory::ObjectFactoryI> for more information
|
|
|
109
|
|
|
110 =cut
|
|
|
111
|
|
|
112 sub new {
|
|
|
113 my ($class,@args) = @_;
|
|
|
114 my $self = $class->SUPER::new(@args);
|
|
|
115 my ($hspF,$hitF,$resultF) = $self->_rearrange([qw(HIT_FACTORY
|
|
|
116 RESULT_FACTORY)],@args);
|
|
|
117 $self->register_factory('hit', $hitF || Bio::Search::Hit::HitFactory->new());
|
|
|
118 $self->register_factory('result', $resultF || Bio::Search::Result::ResultFactory->new());
|
|
|
119
|
|
|
120 return $self;
|
|
|
121 }
|
|
|
122
|
|
|
123 # new comes from the superclass
|
|
|
124
|
|
|
125 =head2 will_handle
|
|
|
126
|
|
|
127 Title : will_handle
|
|
|
128 Usage : if( $handler->will_handle($event_type) ) { ... }
|
|
|
129 Function: Tests if this event builder knows how to process a specific event
|
|
|
130 Returns : boolean
|
|
|
131 Args : event type name
|
|
|
132
|
|
|
133
|
|
|
134 =cut
|
|
|
135
|
|
|
136 sub will_handle{
|
|
|
137 my ($self,$type) = @_;
|
|
|
138 # these are the events we recognize
|
|
|
139 return ( $type eq 'hit' || $type eq 'result' );
|
|
|
140 }
|
|
|
141
|
|
|
142 =head2 SAX methods
|
|
|
143
|
|
|
144 =cut
|
|
|
145
|
|
|
146 =head2 start_result
|
|
|
147
|
|
|
148 Title : start_result
|
|
|
149 Usage : $handler->start_result($resulttype)
|
|
|
150 Function: Begins a result event cycle
|
|
|
151 Returns : none
|
|
|
152 Args : Type of Report
|
|
|
153
|
|
|
154 =cut
|
|
|
155
|
|
|
156 sub start_result {
|
|
|
157 my ($self,$type) = @_;
|
|
|
158 $self->{'_resulttype'} = $type;
|
|
|
159 $self->{'_hits'} = [];
|
|
|
160 return;
|
|
|
161 }
|
|
|
162
|
|
|
163 =head2 end_result
|
|
|
164
|
|
|
165 Title : end_result
|
|
|
166 Usage : my @results = $parser->end_result
|
|
|
167 Function: Finishes a result handler cycle Returns : A Bio::Search::Result::ResultI
|
|
|
168 Args : none
|
|
|
169
|
|
|
170 =cut
|
|
|
171
|
|
|
172 sub end_result {
|
|
|
173 my ($self,$type,$data) = @_;
|
|
|
174 if( defined $data->{'runid'} &&
|
|
|
175 $data->{'runid'} !~ /^\s+$/ ) {
|
|
|
176
|
|
|
177 if( $data->{'runid'} !~ /^lcl\|/) {
|
|
|
178 $data->{"RESULT-query_name"}= $data->{'runid'};
|
|
|
179 } else {
|
|
|
180 ($data->{"RESULT-query_name"},$data->{"RESULT-query_description"}) = split(/\s+/,$data->{"RESULT-query_description"},2);
|
|
|
181 }
|
|
|
182
|
|
|
183 if( my @a = split(/\|/,$data->{'RESULT-query_name'}) ) {
|
|
|
184 my $acc = pop @a ; # this is for accession |1234|gb|AAABB1.1|AAABB1
|
|
|
185 # this is for |123|gb|ABC1.1|
|
|
|
186 $acc = pop @a if( ! defined $acc || $acc =~ /^\s+$/);
|
|
|
187 $data->{"RESULT-query_accession"}= $acc;
|
|
|
188 }
|
|
|
189 delete $data->{'runid'};
|
|
|
190 }
|
|
|
191 my %args = map { my $v = $data->{$_}; s/RESULT//; ($_ => $v); }
|
|
|
192 grep { /^RESULT/ } keys %{$data};
|
|
|
193
|
|
|
194 $args{'-algorithm'} = uc( $args{'-algorithm_name'} || $type);
|
|
|
195 $args{'-hits'} = $self->{'_hits'};
|
|
|
196 my $result = $self->factory('result')->create(%args);
|
|
|
197 $self->{'_hits'} = [];
|
|
|
198 return $result;
|
|
|
199 }
|
|
|
200
|
|
|
201 =head2 start_hit
|
|
|
202
|
|
|
203 Title : start_hit
|
|
|
204 Usage : $handler->start_hit()
|
|
|
205 Function: Starts a Hit event cycle
|
|
|
206 Returns : none
|
|
|
207 Args : type of event and associated hashref
|
|
|
208
|
|
|
209
|
|
|
210 =cut
|
|
|
211
|
|
|
212 sub start_hit{
|
|
|
213 my ($self,$type) = @_;
|
|
|
214 return;
|
|
|
215 }
|
|
|
216
|
|
|
217
|
|
|
218 =head2 end_hit
|
|
|
219
|
|
|
220 Title : end_hit
|
|
|
221 Usage : $handler->end_hit()
|
|
|
222 Function: Ends a Hit event cycle
|
|
|
223 Returns : Bio::Search::Hit::HitI object
|
|
|
224 Args : type of event and associated hashref
|
|
|
225
|
|
|
226
|
|
|
227 =cut
|
|
|
228
|
|
|
229 sub end_hit{
|
|
|
230 my ($self,$type,$data) = @_;
|
|
|
231 my %args = map { my $v = $data->{$_}; s/HIT//; ($_ => $v); } grep { /^HIT/ } keys %{$data};
|
|
|
232 $args{'-algorithm'} = uc( $args{'-algorithm_name'} || $type);
|
|
|
233 $args{'-query_len'} = $data->{'RESULT-query_length'};
|
|
|
234 my ($hitrank) = scalar @{$self->{'_hits'}} + 1;
|
|
|
235 $args{'-rank'} = $hitrank;
|
|
|
236 my $hit = $self->factory('hit')->create(%args);
|
|
|
237 push @{$self->{'_hits'}}, $hit;
|
|
|
238 $self->{'_hsps'} = [];
|
|
|
239 return $hit;
|
|
|
240 }
|
|
|
241
|
|
|
242 =head2 Factory methods
|
|
|
243
|
|
|
244 =cut
|
|
|
245
|
|
|
246 =head2 register_factory
|
|
|
247
|
|
|
248 Title : register_factory
|
|
|
249 Usage : $handler->register_factory('TYPE',$factory);
|
|
|
250 Function: Register a specific factory for a object type class
|
|
|
251 Returns : none
|
|
|
252 Args : string representing the class and
|
|
|
253 Bio::Factory::ObjectFactoryI
|
|
|
254
|
|
|
255 See L<Bio::Factory::ObjectFactoryI> for more information
|
|
|
256
|
|
|
257 =cut
|
|
|
258
|
|
|
259 sub register_factory{
|
|
|
260 my ($self, $type,$f) = @_;
|
|
|
261 if( ! defined $f || ! ref($f) ||
|
|
|
262 ! $f->isa('Bio::Factory::ObjectFactoryI') ) {
|
|
|
263 $self->throw("Cannot set factory to value $f".ref($f)."\n");
|
|
|
264 }
|
|
|
265 $self->{'_factories'}->{lc($type)} = $f;
|
|
|
266 }
|
|
|
267
|
|
|
268
|
|
|
269 =head2 factory
|
|
|
270
|
|
|
271 Title : factory
|
|
|
272 Usage : my $f = $handler->factory('TYPE');
|
|
|
273 Function: Retrieves the associated factory for requested 'TYPE'
|
|
|
274 Returns : a Bio::Factory::ObjectFactoryI or undef if none registered
|
|
|
275 Args : name of factory class to retrieve
|
|
|
276
|
|
|
277 See L<Bio::Factory::ObjectFactoryI> for more information
|
|
|
278
|
|
|
279 =cut
|
|
|
280
|
|
|
281 sub factory{
|
|
|
282 my ($self,$type) = @_;
|
|
|
283 return $self->{'_factories'}->{lc($type)} || $self->throw("No factory registered for $type");
|
|
|
284 }
|
|
|
285
|
|
|
286 1;
|
