t/SeqIO/gcg.t

   1 # -*-Perl-*- Test Harness script for Bioperl
   2 # $Id$
   3
   4 use strict;
   5
   6 BEGIN {
   7     use Bio::Root::Test;
   8
   9     test_begin(-tests               => 17,
  10                            -requires_modules    => [],
  11                            -requires_networking => 0,
  12                           );
  13
  14         use_ok('Bio::SeqIO::gcg');
  15 }
  16
  17 my $verbose = test_debug();
  18
  19 my $format = 'gcg';
  20 my $seqio_obj = Bio::SeqIO->new(-file   => test_input_file("test.$format"),
  21                                                         -format => $format);
  22
  23 isa_ok($seqio_obj, 'Bio::SeqIO');
  24
  25 my @methods = qw(next_seq write_seq);
  26 foreach my $method (@methods) {
  27         can_ok($seqio_obj, $method) ||
  28                 diag "$method method not implemented for $format";
  29 }
  30
  31 # checking the first sequence object
  32 my $seq_obj = $seqio_obj->next_seq();
  33 isa_ok($seq_obj, 'Bio::Seq');
  34 isa_ok($seq_obj, 'Bio::Seq::RichSeq');
  35 my %expected = ('seq'         => 'MVNSNQNQNGNSNGHDDDFPQDSITEPEHMRKLFIGGL' .
  36                                                                  'DYRTTDENLKAHEKWGNIVDVVVMKDPRTKRSRGFGFI' .
  37                                                                  'TYSHSSMIDEAQKSRPHKIDGRVEPKRAVPRQDIDSPN' .
  38                                                                  'AGATVKKLFVGALKDDHDEQSIRDYFQHFGNIVDNIVI' .
  39                                                                  'DKETGKKRGFAFVEFDDYDPVDKVVLQKQHQLNGKMVD' .
  40                                                                  'VKKALPKNDQQGGGGGRGGPGGRAGGNRGNMGGGNYGN' .
  41                                                                  'QNGGGNWNNGGNNWGNNRGNDNWGNNSFGGGGGGGGGY' .
  42                                                                  'GGGNNSWGNNNPWDNGNGGGNFGGGGNNWNGGNDFGGY' .
  43                                                                  'QQNYGGGPQRGGGNFNNNRMQPYQGGGGFKAGGGNQGN' .
  44                                                                  'YGNNQGFNNGGNNRRY',
  45                                 'length'      => '358',
  46                                 'primary_id'  => 'roa1_drome',
  47                                 'description' => qr(Rea guano receptor type III),
  48                            );
  49 is   ($seq_obj->seq(),         $expected{'seq'},         'sequence');
  50 is   ($seq_obj->length(),      $expected{'length'},      'length');
  51 TODO: {
  52         local $TODO = 'possible bug: RichSeq not setting primary_id?';
  53         is   ($seq_obj->primary_id(),  $expected{'primary_id'},  'primary_id');
  54 }
  55 like ($seq_obj->description(), $expected{'description'}, 'description');
  56
  57 # test DOS linefeeds in gcg parser
  58 my $str = Bio::SeqIO->new(-file => test_input_file('test_badlf.gcg'),
  59                                                                   -verbose => $verbose,
  60                                                                   -format => 'GCG');
  61 ok($str);
  62 my $seq = $str->next_seq();
  63 isa_ok ($seq, 'Bio::SeqI');
  64 is(length($seq->seq), $seq->length);
  65 print "Sequence 1 of 1 from GCG stream:\n", $seq->seq, "\n" if( $verbose);
  66
  67
  68
  69 # from testformats.pl
  70 SKIP: {
  71     test_skip(-tests => 4, -requires_modules => [qw(Algorithm::Diff
  72                                                     IO::ScalarArray
  73                                                     IO::String)]);
  74     use_ok('Algorithm::Diff');
  75     eval "use Algorithm::Diff qw(diff LCS);";
  76     use_ok('IO::ScalarArray');
  77     use_ok('IO::String');
  78
  79     my ($file, $type) = ("test.$format", $format);
  80     my $filename = test_input_file($file);
  81     print "processing file $filename\n" if $verbose;
  82     open my $FILE, '<', $filename or die "Could not read file '$filename': $!\n";
  83     my @datain = <$FILE>;
  84     close $FILE;
  85
  86     my $in = IO::String->new(join('', @datain));
  87     my $seqin = Bio::SeqIO->new( -fh => $in,
  88                 -format => $type);
  89     my $out = IO::String->new;
  90     my $seqout = Bio::SeqIO->new( -fh => $out,
  91                  -format => $type);
  92     my $seq;
  93     while( defined($seq = $seqin->next_seq) ) {
  94     $seqout->write_seq($seq);
  95     }
  96     $seqout->close();
  97     $seqin->close();
  98     my $strref = $out->string_ref;
  99     my @dataout = map { $_."\n"} split(/\n/, $$strref );
 100     my @diffs = &diff( \@datain, \@dataout);
 101     is(@diffs, 0, "$format format can round-trip");
 102
 103     if(@diffs && $verbose) {
 104         foreach my $d ( @diffs ) {
 105             foreach my $diff ( @$d ) {
 106                 chomp($diff->[2]);
 107                 print $diff->[0], $diff->[1], "\n>", $diff->[2], "\n";
 108             }
 109         }
 110         print "in is \n", join('', @datain), "\n";
 111         print "out is \n", join('',@dataout), "\n";
 112     }
 113
 114 }