refactoring of assemble_pairs

[biopieces.git] / bp_bin / find_genes

diff --git a/bp_bin/find_genes b/bp_bin/find_genes
index 4ee726e974b0c994285a6570e975a20120d7330f..88dacf2ddee0c7ec899a13cf6f008a7b6380e61b 100755 (executable)
--- a/bp_bin/find_genes
+++ b/bp_bin/find_genes
@@ -1,6 +1,6 @@
-#!/usr/bin/env perl
+#!/usr/bin/env ruby
 
-# Copyright (C) 2007-2009 Martin A. Hansen.
+# Copyright (C) 2007-2011 Martin A. Hansen.
 
 # This program is free software; you can redistribute it and/or
 # modify it under the terms of the GNU General Public License
@@ -18,153 +18,50 @@
 
 # http://www.gnu.org/copyleft/gpl.html
 
-
-# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> DESCRIPTION <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
-
-# Find genes in sequences in stream.
-
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
+# This program is part of the Biopieces framework (www.biopieces.org).
 
-use warnings;
-use strict;
-use Data::Dumper;
-use Maasha::Common;
-use Maasha::Biopieces;
-use Maasha::Filesys;
+# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> DESCRIPTION <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
+# Find genes in prokaryotic genomic sequences in the stream.
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
 
-my ( $options, $in, $out, $record, $tmp_dir, $tmp_file, $fh_out, $fh_in, $entry, $chunk, @lines, $line, $s_id, $type, $s_beg, $s_end, $strand, @fields, $def,
-    @commands, $command );
-
-$options = Maasha::Biopieces::parse_options(
-    [
-        { long => 'full', short => 'f', type => 'flag', mandatory => 'no', default => undef, allowed => undef, disallowed => undef },
-    ]   
-);
-
-$in  = Maasha::Biopieces::read_stream( $options->{ "stream_in" } );
-$out = Maasha::Biopieces::write_stream( $options->{ "stream_out" } );
-
-$tmp_dir  = Maasha::Biopieces::get_tmpdir();
-$tmp_file = "$tmp_dir/tmp.seq";
-$fh_out   = Maasha::Filesys::file_write_open( $tmp_file );
-
-while ( $record = Maasha::Biopieces::get_record( $in ) )
-{
-    if ( $entry = Maasha::Fasta::biopiece2fasta( $record ) ) {
-        Maasha::Fasta::put_entry( $entry, $fh_out );
-    }
-
-    Maasha::Biopieces::put_record( $record, $out );
-}
-
-close $fh_out;
-
-push @commands, "-c" if $options->{ 'full' };
-push @commands, "< $tmp_file > $tmp_file.out";
-push @commands, "2> /dev/null" unless $options->{ 'verbose' };
-
-$command = join " ",  @commands;
-
-Maasha::Common::run( "prodigal", $command );
-
-$fh_in = Maasha::Filesys::file_read_open( "$tmp_file.out" );
-
-$/ = "//\n";
+require 'maasha/biopieces'
+require 'maasha/fasta'
+require 'maasha/prodigal'
 
-while ( $chunk = <$fh_in> )
-{
-    chomp $chunk;
+casts = []
+casts << {:long=>'full',      :short=>'f', :type=>'flag',   :mandatory=>false, :default=>nil,      :allowed=>nil,           :disallowed=>nil}
+casts << {:long=>'procedure', :short=>'p', :type=>'string', :mandatory=>true,  :default=>'single', :allowed=>'single,meta', :disallowed=>nil}
 
-    @lines = split /\n/, $chunk;
+options = Biopieces.options_parse(ARGV, casts)
 
-    $line = shift @lines;
-
-    if ( $line =~ /^DEFINITION\s+(.+)/ )
-    {
-        $def  = $1;
-
-        if ( $def =~ /seqhdr="([^"]+)"/ ) {
-            $s_id = $1;
-        } else {
-            Maasha::Common::error( qq(BAD sequence header: $def) );
-        }
-
-        $line = shift @lines;
-
-        if ( $line =~ /^FEATURES/ )
-        {
-            foreach $line ( @lines )
-            {
-                next if $line =~ /.+\//;
-
-                @fields = split " ", $line;
-
-                $type   = $fields[ 0 ];
-
-                if ( $fields[ 1 ] =~ /complement/ ) {
-                    $strand = "-"; 
-                } else {
-                    $strand = "+"; 
-                }
-
-                if ( $fields[ 1 ] =~ /(\d+)\.\.>?(\d+)/ )
-                {
-                    $s_beg = $1;
-                    $s_end = $2;
-                }
-                else
-                {
-                    Maasha::Common::error( qq(BAD locator: $line) );
-                }
-
-                $record = {
-                    S_ID   => $s_id,
-                    S_BEG  => $s_beg - 1,
-                    S_END  => $s_end - 1,
-                    Q_ID   => $type,
-                    STRAND => $strand,
-                };
-
-                Maasha::Biopieces::put_record( $record, $out );
-            }
-        }
-        else
-        {
-            Maasha::Common::error( qq(BAD feature: $line) );
-        }
-    }
-    else
-    {
-        Maasha::Common::error( qq(BAD definition: $line) );
-    }
-}
-
-close $fh_in;
-
-$/ = "\n";
-
-Maasha::Biopieces::close_stream( $in );
-Maasha::Biopieces::close_stream( $out );
-
-
-# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
+tmpdir  = Biopieces.mktmpdir
+infile  = File.join(tmpdir, "in.fna")
+outfile = File.join(tmpdir, "out.prodigal")
 
+Biopieces.open(options[:stream_in], options[:stream_out]) do |input, output|
+  Fasta.open(infile, "w") do |fasta_io|
+    input.each_record do |record|
+      output.puts record
 
-BEGIN
-{
-    Maasha::Biopieces::status_set();
-}
+      if record[:SEQ_NAME] and record[:SEQ]
+        seq = Seq.new_bp(record)
+        fasta_io.puts seq.to_fasta
+      end
+    end
+  end
 
+  prodigal = Prodigal.new(infile, outfile, options)
+  prodigal.run
 
-END
-{
-    Maasha::Biopieces::status_log();
-}
+  prodigal.each do |record|
+    output.puts record
+  end
+end
 
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<