scripts/utilities/bp_sreformat.pl

   1 #!perl
   2 # Author:  Jason Stajich <jason-at-bioperl-dot-org>
   3 # Purpose: Bioperl implementation of Sean Eddy's sreformat
   4 #          We're not as clever as Sean's squid library though so
   5 #          you have to specify the input format rather than letting
   6 #          the application guess.
   7
   8 use strict;
   9 use warnings;
  10 use Bio::SeqIO;
  11 use Bio::AlignIO;
  12 use Getopt::Long;
  13
  14 my $USAGE = "bp_sreformat -if INFORMAT -of OUTFORMAT -i FILENAME -o output.FORMAT
  15
  16 -h/--help               Print this help
  17 -if/--informat          Specify the input format
  18 -of/--outformat         Specify the output format
  19 -i/--input              Specify the input file name
  20                         (to pass in data on STDIN use minus sign as filename)
  21 -o/--output             Specify the output file name
  22                         (to pass data out on STDOUT use minus sign as filename)
  23 --msa                   Specify this is multiple sequence alignment data
  24 --special=specialparams Specify special params supported by some formats
  25                         Comma or space separated please.
  26                         These include:
  27                         nointerleaved   -- for phylip,non-interleaved format
  28                         idlinebreak     -- for phylip, makes it molphy format
  29                         percentages     -- for clustalw, show % id per line
  30                         flat            -- don't show start-end in seqid
  31                         linelength      -- line length for clustalw
  32                         mrbayes         -- for MrBayes proper NEXUS output
  33 ";
  34
  35
  36 my ($input,$output,$informat,$outformat,$msa,$special);
  37
  38 GetOptions(
  39            'h|help'          => sub { print STDERR ($USAGE); exit(0) },
  40            'i|input:s'         => \$input,
  41            'o|output:s'        => \$output,
  42            'if|informat:s'     => \$informat,
  43            'of|outformat:s'    => \$outformat,
  44            'msa'               => \$msa,
  45            's|special:s'       => \$special,
  46            );
  47
  48 unless( defined $informat && defined $outformat ) {
  49     die(sprintf("Cannot proceed without a defined input and output you gave (%s,%s)\n",
  50                 defined $informat ? $informat : "''" ,
  51                 defined $outformat ? $outformat : "''"));
  52 }
  53
  54 my ($in,$out);
  55 my @extra;
  56 if( $special ) {
  57     @extra = map { my @rc;
  58                    if( /nointerleaved/) {
  59                        @rc = ('-interleaved' => '0');
  60                    } elsif( /mrbayes/ ) {
  61                        @rc = ('-show_symbols' => 0,
  62                               '-show_endblock' => 0);
  63                    } elsif( /(\S+)\=(\S+)/ ) { @rc = ( "-$1" => $2) }
  64                    else{ @rc = ("-$_" => 1) }
  65                    @rc;
  66                } split(/[\s,]/,$special);
  67 }
  68 # guess we're talking about MSA if any of the standard MSA names are used
  69 if( $informat =~ /nexus|phylip|clustal|maf|stockholm|bl2seq|msf/ ||
  70     $outformat =~ /nexus|phylip|clustal|maf|stockholm|bl2seq|msf/ ) {
  71     $msa = 1;
  72 }
  73
  74 if( $msa ) {
  75     eval {
  76         if( defined $input ) {
  77             $in = new Bio::AlignIO(-format => $informat, -file => $input);
  78         } else {
  79             $in = new Bio::AlignIO(-format => $informat, -fh => \*ARGV);
  80         }
  81     };
  82     if( $@ ) {
  83         die("Unknown MSA format to bioperl $informat\n");
  84     }
  85     eval {
  86         if( $output ) {
  87             $out = new Bio::AlignIO(-format => $outformat,
  88                                     -file => ">$output", @extra);
  89         } else {
  90             # default to STDOUT for output
  91             $out = new Bio::AlignIO(-format => $outformat,@extra);
  92         }
  93     };
  94     if( $@ ) {
  95         die("Unknown MSA format to bioperl $outformat\n");
  96     }
  97     while( my $aln = $in->next_aln) {
  98          if( $special =~ /flat/ ) {$aln->set_displayname_flat(1); }
  99          $out->write_aln($aln) }
 100
 101 } else {
 102     eval {
 103         if( defined $input ) {
 104             $in = new Bio::SeqIO(-format => $informat, -file => $input);
 105         } else {
 106             $in = new Bio::SeqIO(-format => $informat, -fh => \*ARGV);
 107         }
 108     };
 109     if( $@ ) {
 110         if( $@ =~ /Could not open/ ) {
 111             die("Could not open input file: $input\n");
 112         } else {
 113             die("Unknown sequence format to bioperl $informat\n");
 114         }
 115     }
 116     eval {
 117            if( $output ) {
 118                $out = new Bio::SeqIO(-format => $outformat,
 119                                      -file => ">$output");
 120            } else {
 121                # default to STDOUT for output
 122                $out = new Bio::SeqIO(-format => $outformat);
 123            }
 124        };
 125     if( $@ ) {
 126         if( $@ =~ /Could not open/ ) {
 127             die("Could not open output file: $output\n");
 128         } else {
 129             die("Unknown sequence format to bioperl $outformat: $@\n");
 130         }
 131     }
 132     while( my $seq = $in->next_seq ) {
 133         $out->write_seq($seq);
 134     }
 135 }
 136
 137 =head1 NAME
 138
 139 bpsreformat - convert sequence formats
 140
 141 =head1 DESCRIPTION
 142
 143 This script uses the SeqIO system that allows conversion of sequence
 144 formats either sequence data or multiple sequence alignment data.  The
 145 name comes from the fact that Sean Eddy's program sreformat (part of
 146 the HMMER pkg) already does this.  Sean's program tries to guess the
 147 input formats while in our code we currently require your to specify what
 148 the input and output formats are and if the data is from a multiple
 149 sequence alignment or from straight sequence files.
 150
 151 Usage:
 152
 153 bpsreformat -if INFORMAT -of OUTFORMAT -i FILENAME -o output.FORMAT
 154
 155   -h/--help        Print this help
 156
 157   -if/--informat   Specify the input format
 158
 159   -of/--outformat  Specify the output format
 160
 161   -i/--input       Specify the input file name
 162                    (to pass in data on STDIN use minus sign as filename)
 163   -o/--output      Specify the output file name
 164                    (to pass data out on STDOUT use minus sign as filename)
 165
 166   --msa            Specify this is multiple sequence alignment data
 167
 168   --special        Will pass on special parameters to the AlignIO/SeqIO
 169                    object -- most of these are for Bio::AlignIO objects
 170                    Comma separated list of the following
 171                    nointerleaved   -- for phylip,non-interleaved format
 172                    idlinebreak     -- for phylip, makes it molphy format
 173                    percentages     -- for clustalw, show % id per line
 174
 175 =cut