t/Assembly.t

   1 # -*-Perl-*- Test Harness script for Bioperl
   2 # $Id$
   3
   4 use strict;
   5
   6 BEGIN {
   7     use lib 't/lib';
   8     use BioperlTest;
   9
  10     test_begin(-tests => 35,
  11         -requires_module => 'DB_File');
  12
  13         use_ok('Bio::Assembly::IO');
  14 }
  15
  16 #
  17 # Testing IO
  18 #
  19
  20
  21 my $in = Bio::Assembly::IO->new
  22         (-file => test_input_file("consed_project","edit_dir","test_project.phrap.out"));
  23
  24 isa_ok($in, 'Bio::Assembly::IO');
  25
  26 my $sc = $in->next_assembly;
  27 isa_ok($sc, 'Bio::Assembly::Scaffold');
  28
  29 #
  30 # Testing Scaffold
  31 #
  32
  33
  34 is $sc->id, "NoName";
  35 is $sc->id('test'), "test";
  36
  37 isa_ok($sc->annotation, 'Bio::AnnotationCollectionI');
  38 is $sc->annotation->get_all_annotation_keys, 0,"no annotations in Annotation collection?";
  39 is $sc->get_nof_contigs, 1;
  40 is $sc->get_nof_sequences_in_contigs, 2;
  41
  42 TODO: {
  43         local $TODO = "get_nof_singlets() should return a number";
  44         is($sc->get_nof_singlets, 1, "get_nof_singlets");
  45 }
  46 is($sc->get_seq_ids, 2, "get_seq_ids");
  47 is($sc->get_contig_ids, 1, "get_contig_ids");
  48 TODO: {
  49         local $TODO = "get_singlet_ids() should return a list";
  50         isnt $sc->get_singlet_ids, 0;
  51 }
  52
  53 #
  54 # Testing Contig
  55 #
  56
  57 #
  58 # Testing ContigAnalysis
  59 #
  60
  61 #
  62 # Testing Ace
  63 #
  64
  65 my $aio = Bio::Assembly::IO->new(
  66         -file=>test_input_file("consed_project","edit_dir","test_project.fasta.screen.ace.2"),
  67         -format=>'ace',
  68 );
  69
  70 my $assembly = $aio->next_assembly();
  71 my @contigs = $assembly->all_contigs();
  72
  73 my $direction = $contigs[0]->strand;
  74 is $direction, 1;
  75
  76 my $features =  $contigs[0]->get_features_collection;
  77 my @contig_features = $features->get_all_features;
  78 is @contig_features, 8;
  79
  80 my @annotations = grep {$_->primary_tag eq 'Annotation'} @contig_features;
  81 is @annotations, 2;
  82 my $had_tag = 0;
  83 foreach my $an (@annotations) {
  84         if ($an->has_tag('extra_info')) {
  85                 $had_tag++;
  86                 is (($an->get_tag_values('extra_info'))[0], "contig extra\ninfo\n");
  87         }
  88         elsif ($an->has_tag('comment')){
  89                 $had_tag++;
  90                 is (($an->get_tag_values('comment'))[0], "contig tag\ncomment\n");
  91         }
  92 }
  93 is $had_tag, 2;
  94
  95 #
  96 # Testing TIGR format
  97 #
  98
  99 # Importing an assembly
 100
 101 my $asm_in = Bio::Assembly::IO->new(
 102     -file => test_input_file("sample_dataset.tasm "),
 103     -format=>'tigr'
 104 );
 105 my $scaf_in = $asm_in->next_assembly;
 106
 107 isa_ok($scaf_in, 'Bio::Assembly::Scaffold');
 108 is($scaf_in->id, 'NoName');
 109 is($scaf_in->get_nof_contigs, 13);
 110 is($scaf_in->get_nof_sequences_in_contigs, 36);
 111 is($scaf_in->get_nof_singlets, 0);
 112 my @seqids = sort qw(sdsu|SDSU1_RFPERU_001_A09.x01.phd.1
 113 sdsu|SDSU1_RFPERU_001_B03.x01.phd.1 sdsu|SDSU1_RFPERU_001_B04.x01.phd.1
 114 sdsu|SDSU1_RFPERU_001_E04.x01.phd.1 sdsu|SDSU_RFPERU_002_A01.x01.phd.1
 115 sdsu|SDSU_RFPERU_002_B07.x01.phd.1 sdsu|SDSU_RFPERU_002_C12.x01.phd.1
 116 sdsu|SDSU_RFPERU_002_D08.x01.phd.1 sdsu|SDSU_RFPERU_002_H12.x01.phd.1
 117 sdsu|SDSU_RFPERU_003_G09.x01.phd.1 sdsu|SDSU_RFPERU_004_H12.x01.phd.1
 118 sdsu|SDSU_RFPERU_005_F02.x01.phd.1 sdsu|SDSU_RFPERU_006_D03.x01.phd.1
 119 sdsu|SDSU_RFPERU_006_E04.x01.phd.1 sdsu|SDSU_RFPERU_006_E05.x01.phd.1
 120 sdsu|SDSU_RFPERU_006_H08.x01.phd.1 sdsu|SDSU_RFPERU_007_E09.x01.phd.1
 121 sdsu|SDSU_RFPERU_007_F06.x01.phd.1 sdsu|SDSU_RFPERU_008_B02.x01.phd.1
 122 sdsu|SDSU_RFPERU_009_E07.x01.phd.1 sdsu|SDSU_RFPERU_010_B05.x01.phd.1
 123 sdsu|SDSU_RFPERU_010_B06.x01.phd.1 sdsu|SDSU_RFPERU_010_C09.x01.phd.1
 124 sdsu|SDSU_RFPERU_010_D10.x01.phd.1 sdsu|SDSU_RFPERU_012_H02.x01.phd.1
 125 sdsu|SDSU_RFPERU_013_B05.x01.phd.1 sdsu|SDSU_RFPERU_013_C07.x01.phd.1
 126 sdsu|SDSU_RFPERU_013_C08.x01.phd.1 sdsu|SDSU_RFPERU_013_G10.x01.phd.1
 127 sdsu|SDSU_RFPERU_013_H05.x01.phd.1 sdsu|SDSU_RFPERU_014_H06.x01.phd.1
 128 sdsu|SDSU_RFPERU_015_A05.x01.phd.1 sdsu|SDSU_RFPERU_015_C06.x01.phd.1
 129 sdsu|SDSU_RFPERU_015_E04.x01.phd.1 sdsu|SDSU_RFPERU_015_G04.x01.phd.1
 130 sdsu|SDSU_RFPERU_015_H03.x01.phd.1);
 131 my @contigids = sort qw(106 144 148 17 185 2 210 36 453 500 613 668 93);
 132 is_deeply([sort $scaf_in->get_seq_ids], \@seqids);
 133 is_deeply([sort $scaf_in->get_contig_ids], \@contigids);
 134 is_deeply([$scaf_in->get_singlet_ids], []);
 135 isa_ok($scaf_in->get_seq_by_id('sdsu|SDSU1_RFPERU_001_A09.x01.phd.1'),'Bio::LocatableSeq');
 136 my $contig = $scaf_in->get_contig_by_id('106');
 137 isa_ok($contig,'Bio::Assembly::Contig');
 138 # check Contig object SeqFeature::Collection
 139 # should add more specific Contig tests...
 140 my @sfs = $contig->get_features_collection->get_all_features;
 141 is(scalar(@sfs), 5);
 142 my %primary_tags = map { $_->primary_tag => 1 } @sfs;
 143 ok exists $primary_tags{'_aligned_coord:sdsu|SDSU_RFPERU_006_E04.x01.phd.1'};
 144 is($sfs[1]->seq_id(), undef); # should this be undef?
 145
 146 isa_ok($scaf_in->annotation, 'Bio::AnnotationCollectionI');
 147 is($scaf_in->annotation->get_all_annotation_keys, 0,"no annotations in Annotation collection?");
 148
 149 # Exporting an assembly
 150
 151 my $asm_outfile = test_output_file();
 152 my $asm_out = Bio::Assembly::IO->new(
 153     -file=> ">$asm_outfile",
 154     -format=>'tigr'
 155 );
 156
 157 ok $asm_out->write_assembly( -scaffold => $scaf_in);
 158