refactoring of assemble_pairs

[biopieces.git] / bp_bin / analyze_seq

diff --git a/bp_bin/analyze_seq b/bp_bin/analyze_seq
index 14188894cdbd07a88da5282a3b12b108518b923a..2c38a274beeb593394ec5d33663d4dec1129de23 100755 (executable)
--- a/bp_bin/analyze_seq
+++ b/bp_bin/analyze_seq
@@ -1,6 +1,6 @@
-#!/usr/bin/env perl
+#!/usr/bin/env ruby
 
-# Copyright (C) 2007-2009 Martin A. Hansen.
+# Copyright (C) 2007-2010 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
@@ -18,117 +18,42 @@
 
 # http://www.gnu.org/copyleft/gpl.html
 
-
-# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> DESCRIPTION <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
-
-
-# Analyze sequences in the stream.
-
-
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
+# This program is part of the Biopieces framework (www.biopieces.org).
 
-use warnings;
-use strict;
-use Maasha::Biopieces;
-use Maasha::Common;
-use Maasha::Seq;
-
-
-# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
-
-
-my ( $options, $in, $out, $record, $analysis );
-
-$options = Maasha::Biopieces::parse_options();
-
-$in  = Maasha::Biopieces::read_stream( $options->{ "stream_in" } );
-$out = Maasha::Biopieces::write_stream( $options->{ "stream_out" } );
-
-while ( $record = Maasha::Biopieces::get_record( $in ) ) 
-{
-    seq_analyze( $record ) if $record->{ "SEQ" };
-
-    Maasha::Biopieces::put_record( $record, $out );
-}
-
-Maasha::Biopieces::close_stream( $in );
-Maasha::Biopieces::close_stream( $out );
+# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> DESCRIPTION <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
+# Analyze sequences in the stream.
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
 
-sub seq_analyze
-{
-    # Martin A. Hansen, July 2009.
+require 'maasha/biopieces'
+require 'maasha/seq'
 
-    # Analyzes the composition of the string in the record and appends
-    # the analysis to the record.
+casts = []
 
-    my ( $record,   # Biopiece record with a SEQ entry.
-       ) = @_;
-
-    # Returns nothing.
-
-    my ( %char_hash, @indels, @alph, $char, $gc, $at, $lc, $max, $indels );
-
-    %char_hash = Maasha::Common::str_analyze( $record->{ 'SEQ' } );
-
-    $record->{ 'SEQ_TYPE' } = Maasha::Seq::seq_guess_type( $record->{ 'SEQ' } );
-    $record->{ 'SEQ_LEN' }  = length $record->{ 'SEQ' };
-
-    @alph   = Maasha::Seq::seq_alph( $record->{ 'SEQ_TYPE' } . "_AMBI" );
-    @indels = qw( - ~ . _ );
-
-    $max = 0;
-
-    foreach $char ( @alph )
-    {
-        $char_hash{ $char } += $char_hash{ lc $char } || 0;
-
-        $record->{ "RES[$char]" } = $char_hash{ $char };
-
-        $max = $char_hash{ $char } if $char_hash{ $char } > $max;
-
-        $record->{ "RES_SUM" } += $char_hash{ $char };
-    }
-
-    $indels = 0;
-
-    map { $record->{ "RES[$_]" } = $char_hash{ $_ }; $indels += $char_hash{ $_ } } @indels;
-
-    if ( $record->{ "SEQ_TYPE" } =~ /DNA|RNA/i )
-    {
-        $gc = $char_hash{ "G" } + $char_hash{ "C" };
-        $at = $char_hash{ "A" } + $char_hash{ "T" } + $char_hash{ "U" };
-
-        $lc = 0;
-
-        map { $lc += $char_hash{ lc $_ } || 0 } @alph;
-
-        $record->{ "MIX_INDEX" }  = sprintf( "%.2f", $max / ( $record->{ "SEQ_LEN" } - $indels ) );
-        $record->{ "GC%" }        = sprintf( "%.2f", 100 * $gc / ( $record->{ "SEQ_LEN" } - $indels ) );
-        $record->{ "SOFT_MASK%" } = sprintf( "%.2f", 100 * $lc / ( $record->{ "SEQ_LEN" } - $indels ) );
-        $record->{ "HARD_MASK%" } = sprintf( "%.2f", 100 * ( $char_hash{ "n" } + $char_hash{ "N" } ) / ( $record->{ "SEQ_LEN" } - $indels ) );
-        $record->{ "MELT_TEMP" }  = sprintf( "%.2f", 4 * $gc + 2 * $at );
-    }
-}
-
-
-# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
+options = Biopieces.options_parse(ARGV, casts)
 
+Biopieces.open(options[:stream_in], options[:stream_out]) do |input, output|
+  input.each_record do |record|
+    if record[:SEQ]
+      seq  = Seq.new_bp(record)
+      comp = seq.composition
 
-BEGIN
-{
-    Maasha::Biopieces::status_set();
-}
+      comp.each_pair do |key,val|
+        record["RES[#{key}]"] = val
+      end
 
+      record["SOFT_MASK%"] = seq.soft_mask
+      record["HARD_MASK%"] = (comp["N"].to_f / (seq.len - seq.indels).to_f * 100.0).round(2)
+      record["GC%"]        = ((comp["G"] + comp["C"]).to_f / (seq.len - seq.indels).to_f * 100.0).round(2)
+    end
 
-END
-{
-    Maasha::Biopieces::status_log();
-}
+    output.puts record
+  end
+end
 
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<