X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bp_bin%2Fassemble_tag_contigs;h=2a8c04378ac3dd5d7819d69d0ffe586f7507cc84;hb=48bea5c28b89dc5586d0bddb338ccd6ba23aa1f9;hp=9236586f351b7ad6bc999a1ce25fc5de95c5611e;hpb=66938be910a5642b50ddf28991649858b5c36580;p=biopieces.git

diff --git a/bp_bin/assemble_tag_contigs b/bp_bin/assemble_tag_contigs
index 9236586..2a8c043 100755
--- a/bp_bin/assemble_tag_contigs
+++ b/bp_bin/assemble_tag_contigs
@@ -1,4 +1,4 @@
-#!/usr/bin/env perl -w
+#!/usr/bin/env perl
 
 # Copyright (C) 2007-2009 Martin A. Hansen.
 
@@ -26,17 +26,32 @@
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
 
+use warnings;
 use strict;
 use Maasha::Biopieces;
 use Maasha::Filesys;
 use Maasha::UCSC::BED;
 
+use constant {
+    chrom       => 0,   # BED field names
+    chromStart  => 1,
+    chromEnd    => 2,
+    name        => 3,
+    score       => 4,
+    strand      => 5,
+    thickStart  => 6,
+    thickEnd    => 7,
+    itemRgb     => 8,
+    blockCount  => 9,
+    blockSizes  => 10,
+    blockStarts => 11,
+};
+
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
 
-my ( $options, $in, $out, $tmp_dir, $bed_file, $tag_file, $fh_in, $fh_out, 
-     $cols, $record, $bed_entry, $file_hash, $chr, $strand );
+my ( $options, $in, $out, $tmp_dir, $fh_in, $fh_out, $cols, $record, $bed_entry, %fh_hash, %file_hash, $file, $strand, $id, $tag_file, $bed_file );
 
 $options = Maasha::Biopieces::parse_options(
     [
@@ -47,42 +62,53 @@ $options = Maasha::Biopieces::parse_options(
 $in  = Maasha::Biopieces::read_stream( $options->{ "stream_in" } );
 $out = Maasha::Biopieces::write_stream( $options->{ "stream_out" } );
 
+`cd "$ENV{ BP_C }/Maasha/src" && make`; # FIXME OMG this is crufty!
+
 $tmp_dir = Maasha::Biopieces::get_tmpdir();
     
-$bed_file = "$tmp_dir/assemble_tag_contigs.bed";
-$fh_out   = Maasha::Filesys::file_write_open( $bed_file );
-$cols     = 6; # we only need the first 6 BED columns
+$cols = 6; # we only need the first 6 BED columns
 
 while ( $record = Maasha::Biopieces::get_record( $in ) ) 
 {
     if ( $bed_entry = Maasha::UCSC::BED::biopiece2bed( $record, $cols ) )
     {
-        $strand = $record->{ 'STRAND' } || '+';
+        if ( not exists $fh_hash{ $bed_entry->[ chrom ] . $bed_entry->[ strand ] } )
+        {
+            $file = "$tmp_dir/$bed_entry->[ chrom ]$bed_entry->[ strand ]";
+
+            $file_hash{ $bed_entry->[ chrom ] . $bed_entry->[ strand ] } = $file;
+            $fh_hash{   $bed_entry->[ chrom ] . $bed_entry->[ strand ] } = Maasha::Filesys::file_write_open( $file );
+        }
+
+        $fh_out = $fh_hash{ $bed_entry->[ chrom ] . $bed_entry->[ strand ] };
 
         Maasha::UCSC::BED::bed_entry_put( $bed_entry, $fh_out, $cols, $options->{ 'check' } );
     }
 }
 
-close $fh_out;
-
-$file_hash = Maasha::UCSC::BED::bed_file_split_on_chr( $bed_file, $tmp_dir, $cols );
+$id = 0;
 
-unlink $bed_file;
-
-foreach $chr ( sort keys %{ $file_hash } )
+foreach $file ( sort keys %file_hash )
 {
-    $bed_file = $file_hash->{ $chr };
+    $bed_file = $file_hash{ $file };
     $tag_file = "$bed_file.tc";
 
+    $strand = substr $file, -1, 1;
+
     Maasha::Common::run( "bed2tag_contigs", "< $bed_file > $tag_file" );
 
     $fh_in = Maasha::Filesys::file_read_open( $tag_file );
 
     while ( $bed_entry = Maasha::UCSC::BED::bed_entry_get( $fh_in, $cols, $options->{ 'check' } ) )
     {
+        $bed_entry->[ name ]   = sprintf( "TC%08d", $id );
+        $bed_entry->[ strand ] = $strand;
+
         if ( $record = Maasha::UCSC::BED::bed2biopiece( $bed_entry ) ) {
             Maasha::Biopieces::put_record( $record, $out );
         }
+
+        $id++;
     }
 
     close $fh_in;
@@ -91,8 +117,6 @@ foreach $chr ( sort keys %{ $file_hash } )
     unlink $tag_file;
 }
 
-Maasha::Filesys::dir_remove( $tmp_dir );
-
 Maasha::Biopieces::close_stream( $in );
 Maasha::Biopieces::close_stream( $out );