branch-1-6/examples/revcom_dir.pl

   1 #!/usr/bin/perl -w
   2 #
   3 ################################################################################
   4 #11-17-2001
   5 #Jianwen Fang (jwfang1999@yahoo.com)
   6 #
   7 #THis program returns reverse complement sequences of all sequences in the current directory
   8 #and save them in the same directory, using the same name with extension ".rev"
   9 ###############################################################################
  10
  11
  12 use strict;
  13 use Bio::Seq;
  14 use Bio::SeqIO;
  15
  16 my @files = ();
  17 my $folder = '.';
  18 my $inputFormat;
  19 my $outputFormat;
  20 my $numSeq;
  21
  22    #Fasta       FASTA format
  23    #EMBL        EMBL format
  24    #GenBank     GenBank format
  25    #GCG         GCG format
  26    #raw         Raw format (one sequence per line, no ID)
  27
  28 my @format = ('Fasta', 'EMBL', 'GenBank', 'GCG', 'Raw');
  29
  30 print("\nWhat is the format of the original sequence files?\n");
  31 print("type 0 for Fasta; 1 for EMBL; 2 for GenBank; 3 for GCG; 4 for Raw\n");
  32 $inputFormat = <STDIN>;
  33 chomp ($inputFormat);
  34
  35 print("\nWhat is the format of the reverse complement sequence files you want?\n");
  36 print("type 0 for Fasta; 1 for EMBL; 2 for GenBank; 3 for GCG; 4 for Raw\n");
  37 $outputFormat = <STDIN>;
  38 chomp ($outputFormat);
  39
  40 unless(opendir(FOLDER, $folder))
  41 {
  42         print "cannot open folder $folder!\n";
  43         exit;
  44 }
  45
  46 @files = grep(!/^\.\.?$/, readdir(FOLDER));
  47
  48 foreach my $file (@files)
  49         {
  50            if($file =~ /seq/i)
  51             {
  52                     getRevcom($file);
  53                 $numSeq++;
  54             }
  55         }
  56
  57 print "$numSeq reverse complement sequences have been saved in current directory\n";
  58 exit;
  59
  60 ############################################################################
  61 #subroutine getRevcom take an backward sequence file name(should with .seq extension) as parameter
  62 #return its revcom sequence using the same name with the extension replaced with rev
  63 ############################################################################
  64 sub getRevcom
  65 {
  66         my $seqFile = $_[0];
  67         my $in = Bio::SeqIO->new('-file'=>$seqFile, '-format'=>$format[$inputFormat]);
  68         my $seq = $in->next_seq();
  69         my $revcomSeq = $seq->revcom();
  70         my @outSeqFile = split (/\./, $seqFile);
  71         pop @outSeqFile;
  72         push(@outSeqFile, 'rev');
  73         my $outSeqFile = join('.', @outSeqFile);
  74         print "$outSeqFile\n";
  75         my $out = Bio::SeqIO->new('-file'=>">$outSeqFile", '-format'=>$format[$outputFormat]);
  76         $out->write_seq($revcomSeq);
  77 }
  78
  79