t/SeqIO/fasta.t

   1 # -*-Perl-*- Test Harness script for Bioperl
   2 # $Id$
   3
   4 use strict;
   5
   6 BEGIN {
   7         use lib '.';
   8     use Bio::Root::Test;
   9
  10     test_begin(-tests               => 22,
  11                            -requires_modules    => [],
  12                            -requires_networking => 0,
  13                           );
  14
  15         use_ok('Bio::SeqIO::fasta');
  16 }
  17
  18 my $verbose = test_debug();
  19
  20 my $format = 'fasta';
  21 my $seqio_obj = Bio::SeqIO->new(-file   => test_input_file("test.$format"),
  22                                                         -format => $format);
  23
  24 isa_ok($seqio_obj, 'Bio::SeqIO');
  25
  26 my @methods = qw(next_seq write_seq);
  27 foreach my $method (@methods) {
  28         can_ok($seqio_obj, $method) ||
  29                 diag "$method method not implemented for $format";
  30 }
  31
  32 # checking the first sequence object
  33 my $seq_obj = $seqio_obj->next_seq();
  34 isa_ok($seq_obj, 'Bio::Seq');
  35 my %expected = ('seq'         => 'MVNSNQNQNGNSNGHDDDFPQDSITEPEHMRKLFIGGL' .
  36                                                                  'DYRTTDENLKAHEKWGNIVDVVVMKDPRTKRSRGFGFI' .
  37                                                                  'TYSHSSMIDEAQKSRPHKIDGRVEPKRAVPRQDIDSPN' .
  38                                                                  'AGATVKKLFVGALKDDHDEQSIRDYFQHFGNIVDNIVI' .
  39                                                                  'DKETGKKRGFAFVEFDDYDPVDKVVLQKQHQLNGKMVD' .
  40                                                                  'VKKALPKNDQQGGGGGRGGPGGRAGGNRGNMGGGNYGN' .
  41                                                                  'QNGGGNWNNGGNNWGNNRGNDNWGNNSFGGGGGGGGGY' .
  42                                                                  'GGGNNSWGNNNPWDNGNGGGNFGGGGNNWNGGNDFGGY' .
  43                                                                  'QQNYGGGPQRGGGNFNNNRMQPYQGGGGFKAGGGNQGN' .
  44                                                                  'YGNNQGFNNGGNNRRY',
  45                                 'length'      => '358',
  46                                 'primary_id'  => 'roa1_drome',
  47                                 'description' => qr(Rea guano receptor type III),
  48                            );
  49 is   ($seq_obj->seq(),         $expected{'seq'},         'sequence');
  50 is   ($seq_obj->length(),      $expected{'length'},      'length');
  51 is   ($seq_obj->primary_id(),  $expected{'primary_id'},  'primary_id');
  52 like ($seq_obj->description(), $expected{'description'}, 'description');
  53
  54
  55 # checking the second sequence object
  56 my $seq_obj2  = $seqio_obj->next_seq();
  57 isa_ok($seq_obj2, 'Bio::Seq');
  58 my %expected2 = ('seq'         => 'MVNSNQNQNGNSNGHDDDFPQDSITEPEHMRKLFIGGL' .
  59                                                                   'DYRTTDENLKAHEKWGNIVDVVVMKDPTSTSTSTSTST' .
  60                                                                   'STSTSTMIDEAQKSRPHKIDGRVEPKRAVPRQDIDSPN' .
  61                                                                   'AGATVKKLFVGALKDDHDEQSIRDYFQHLLLLLLLDLL' .
  62                                                                   'LLDLLLLDLLLFVEFDDYDPVDKVVLQKQHQLNGKMVD' .
  63                                                                   'VKKALPKNDQQGGGGGRGGPGGRAGGNRGNMGGGNYGN' .
  64                                                                   'QNGGGNWNNGGNNWGNNRGNDNWGNNSFGGGGGGGGGY' .
  65                                                                   'GGGNNSWGNNNPWDNGNGGGNFGGGGNNWNGGNDFGGY' .
  66                                                                   'QQNYGGGPQRGGGNFNNNRMQPYQGGGGFKAGGGNQGN' .
  67                                                                   'YGNNQGFNNGGNNRRY',
  68                                  'length'      => '358',
  69                                  'primary_id'  => 'roa2_drome',
  70                                  'description' => qr(Rea guano ligand),
  71                             );
  72 is   ($seq_obj2->seq(),         $expected2{'seq'},         'sequence');
  73 is   ($seq_obj2->length(),      $expected2{'length'},      'length');
  74 is   ($seq_obj2->primary_id(),  $expected2{'primary_id'},  'primary_id');
  75 like ($seq_obj2->description(), $expected2{'description'}, 'description');
  76
  77 # from testformats.pl
  78 SKIP: {
  79     test_skip(-tests => 4, -requires_modules => [qw(Algorithm::Diff
  80                                                     IO::ScalarArray
  81                                                     IO::String)]);
  82     use_ok('Algorithm::Diff');
  83     eval "use Algorithm::Diff qw(diff LCS);";
  84     use_ok('IO::ScalarArray');
  85     use_ok('IO::String');
  86
  87         my ($file, $type) = ("test.$format", $format);
  88     my $filename = test_input_file($file);
  89     print "processing file $filename\n" if $verbose;
  90     open(FILE, "< $filename") or die("cannot open $filename");
  91     my @datain = <FILE>;
  92     my $in = new IO::String(join('', @datain));
  93     my $seqin = new Bio::SeqIO( -fh => $in,
  94                 -format => $type);
  95     my $out = new IO::String;
  96     my $seqout = new Bio::SeqIO( -fh => $out,
  97                  -format => $type);
  98     my $seq;
  99     while( defined($seq = $seqin->next_seq) ) {
 100     $seqout->write_seq($seq);
 101     }
 102     $seqout->close();
 103     $seqin->close();
 104     my $strref = $out->string_ref;
 105     my @dataout = map { $_."\n"} split(/\n/, $$strref );
 106     my @diffs = &diff( \@datain, \@dataout);
 107     is(@diffs, 0, "$format format can round-trip");
 108
 109     if(@diffs && $verbose) {
 110         foreach my $d ( @diffs ) {
 111             foreach my $diff ( @$d ) {
 112                 chomp($diff->[2]);
 113                 print $diff->[0], $diff->[1], "\n>", $diff->[2], "\n";
 114             }
 115         }
 116         print "in is \n", join('', @datain), "\n";
 117         print "out is \n", join('',@dataout), "\n";
 118     }
 119
 120 }
 121
 122 # bug 1508
 123 # test genbank, gcg, ace against fasta (should throw an exception on each)
 124
 125 for my $file (qw(roa1.genbank test.gcg test.ace test.raw)) {
 126         my $in = Bio::SeqIO->new(-file   => test_input_file($file),
 127                                                          -format => 'fasta');
 128         throws_ok {$in->next_seq}
 129                 qr/The sequence does not appear to be FASTA format/, "dies with $file";
 130 }