t/SeqIO/gcg.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               => 17,
  11                            -requires_modules    => [],
  12                            -requires_networking => 0,
  13                           );
  14
  15         use_ok('Bio::SeqIO::gcg');
  16 }
  17
  18 my $verbose = test_debug();
  19
  20 my $format = 'gcg';
  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 isa_ok($seq_obj, 'Bio::Seq::RichSeq');
  36 my %expected = ('seq'         => 'MVNSNQNQNGNSNGHDDDFPQDSITEPEHMRKLFIGGL' .
  37                                                                  'DYRTTDENLKAHEKWGNIVDVVVMKDPRTKRSRGFGFI' .
  38                                                                  'TYSHSSMIDEAQKSRPHKIDGRVEPKRAVPRQDIDSPN' .
  39                                                                  'AGATVKKLFVGALKDDHDEQSIRDYFQHFGNIVDNIVI' .
  40                                                                  'DKETGKKRGFAFVEFDDYDPVDKVVLQKQHQLNGKMVD' .
  41                                                                  'VKKALPKNDQQGGGGGRGGPGGRAGGNRGNMGGGNYGN' .
  42                                                                  'QNGGGNWNNGGNNWGNNRGNDNWGNNSFGGGGGGGGGY' .
  43                                                                  'GGGNNSWGNNNPWDNGNGGGNFGGGGNNWNGGNDFGGY' .
  44                                                                  'QQNYGGGPQRGGGNFNNNRMQPYQGGGGFKAGGGNQGN' .
  45                                                                  'YGNNQGFNNGGNNRRY',
  46                                 'length'      => '358',
  47                                 'primary_id'  => 'roa1_drome',
  48                                 'description' => qr(Rea guano receptor type III),
  49                            );
  50 is   ($seq_obj->seq(),         $expected{'seq'},         'sequence');
  51 is   ($seq_obj->length(),      $expected{'length'},      'length');
  52 TODO: {
  53         local $TODO = 'possible bug: RichSeq not setting primary_id?';
  54         is   ($seq_obj->primary_id(),  $expected{'primary_id'},  'primary_id');
  55 }
  56 like ($seq_obj->description(), $expected{'description'}, 'description');
  57
  58 # test DOS linefeeds in gcg parser
  59 my $str = Bio::SeqIO->new(-file => test_input_file('test_badlf.gcg'),
  60                                                                   -verbose => $verbose,
  61                                                                   -format => 'GCG');
  62 ok($str);
  63 my $seq = $str->next_seq();
  64 isa_ok ($seq, 'Bio::SeqI');
  65 is(length($seq->seq), $seq->length);
  66 print "Sequence 1 of 1 from GCG stream:\n", $seq->seq, "\n" if( $verbose);
  67
  68
  69
  70 # from testformats.pl
  71 SKIP: {
  72     test_skip(-tests => 4, -requires_modules => [qw(Algorithm::Diff
  73                                                     IO::ScalarArray
  74                                                     IO::String)]);
  75     use_ok('Algorithm::Diff');
  76     eval "use Algorithm::Diff qw(diff LCS);";
  77     use_ok('IO::ScalarArray');
  78     use_ok('IO::String');
  79
  80     my ($file, $type) = ("test.$format", $format);
  81     my $filename = test_input_file($file);
  82     print "processing file $filename\n" if $verbose;
  83     open my $FILE, '<', $filename or die "Could not read file '$filename': $!\n";
  84     my @datain = <$FILE>;
  85     close $FILE;
  86
  87     my $in = new IO::String(join('', @datain));
  88     my $seqin = new Bio::SeqIO( -fh => $in,
  89                 -format => $type);
  90     my $out = new IO::String;
  91     my $seqout = new Bio::SeqIO( -fh => $out,
  92                  -format => $type);
  93     my $seq;
  94     while( defined($seq = $seqin->next_seq) ) {
  95     $seqout->write_seq($seq);
  96     }
  97     $seqout->close();
  98     $seqin->close();
  99     my $strref = $out->string_ref;
 100     my @dataout = map { $_."\n"} split(/\n/, $$strref );
 101     my @diffs = &diff( \@datain, \@dataout);
 102     is(@diffs, 0, "$format format can round-trip");
 103
 104     if(@diffs && $verbose) {
 105         foreach my $d ( @diffs ) {
 106             foreach my $diff ( @$d ) {
 107                 chomp($diff->[2]);
 108                 print $diff->[0], $diff->[1], "\n>", $diff->[2], "\n";
 109             }
 110         }
 111         print "in is \n", join('', @datain), "\n";
 112         print "out is \n", join('',@dataout), "\n";
 113     }
 114
 115 }