X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bp_bin%2Fget_genome_seq;h=e219c5188a993fd54e0af1b2c363abbb45ba9720;hb=5de6112b70b59420b245ce636a8b2e3c90acbe00;hp=cde18f90d07b9f6f51fb592443867554646b0f57;hpb=9854397f83abe8d5a8c614b799ef3107d3f85f80;p=biopieces.git

diff --git a/bp_bin/get_genome_seq b/bp_bin/get_genome_seq
index cde18f9..e219c51 100755
--- a/bp_bin/get_genome_seq
+++ b/bp_bin/get_genome_seq
@@ -1,4 +1,4 @@
-#!/usr/bin/env perl -w
+#!/usr/bin/env perl
 
 # Copyright (C) 2007-2009 Martin A. Hansen.
 
@@ -26,28 +26,31 @@
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
 
+use warnings;
 use strict;
+use Data::Dumper;
 use Maasha::Biopieces;
 use Maasha::Filesys;
+use Maasha::Fasta;
 use Maasha::Seq;
 
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
 
-my ( $run_time_beg, $run_time_end, $options, $in, $out, $record, $genome_file, $index_file, $fh, $genome,
-     $beg, $end, $len, $index_beg, $index_len, @begs, @lens, $index, $seq, $i, %lookup_hash );
+my ( $options, $in, $out, $record, $genome_file, $index_file, $fh, $genome,
+     $beg, $end, $len, $index_beg, $index_len, @begs, @lens, $index, $seq, $i );
 
 $options = Maasha::Biopieces::parse_options(
     [
-        { long => 'genome', short => 'g', type => 'genome', mandatory => 'no', default => undef, allowed => undef, disallowed => undef },
-        { long => 'chr',    short => 'c', type => 'string', mandatory => 'no', default => undef, allowed => undef, disallowed => undef },
-        { long => 'beg',    short => 'b', type => 'uint',   mandatory => 'no', default => undef, allowed => undef, disallowed => 0 },
-        { long => 'end',    short => 'e', type => 'uint',   mandatory => 'no', default => undef, allowed => undef, disallowed => 0 },
-        { long => 'len',    short => 'l', type => 'uint',   mandatory => 'no', default => undef, allowed => undef, disallowed => 0 },
-        { long => 'flank',  short => 'f', type => 'uint',   mandatory => 'no', default => 0,     allowed => undef, disallowed => undef },
-        { long => 'mask',   short => 'm', type => 'flag',   mandatory => 'no', default => undef, allowed => undef, disallowed => undef },
-        { long => 'splice', short => 's', type => 'flag',   mandatory => 'no', default => undef, allowed => undef, disallowed => undef },
+        { long => 'genome', short => 'g', type => 'genome', mandatory => 'yes', default => undef, allowed => undef, disallowed => undef },
+        { long => 'chr',    short => 'c', type => 'string', mandatory => 'no',  default => undef, allowed => undef, disallowed => undef },
+        { long => 'beg',    short => 'b', type => 'uint',   mandatory => 'no',  default => undef, allowed => undef, disallowed => 0 },
+        { long => 'end',    short => 'e', type => 'uint',   mandatory => 'no',  default => undef, allowed => undef, disallowed => 0 },
+        { long => 'len',    short => 'l', type => 'uint',   mandatory => 'no',  default => undef, allowed => undef, disallowed => 0 },
+        { long => 'flank',  short => 'f', type => 'uint',   mandatory => 'no',  default => 0,     allowed => undef, disallowed => undef },
+        { long => 'mask',   short => 'm', type => 'flag',   mandatory => 'no',  default => undef, allowed => undef, disallowed => undef },
+        { long => 'splice', short => 's', type => 'flag',   mandatory => 'no',  default => undef, allowed => undef, disallowed => undef },
     ]   
 );
 
@@ -64,13 +67,9 @@ if ( $options->{ "genome" } )
     $fh          = Maasha::Filesys::file_read_open( $genome_file );
     $index       = Maasha::Fasta::index_retrieve( $index_file );
 
-    shift @{ $index }; # Get rid of the file size info
-
-    map { $lookup_hash{ $_->[ 0 ] } = [ $_->[ 1 ], $_->[ 2 ] ] } @{ $index };
-
-    if ( exists $lookup_hash{ $options->{ "chr" } } and defined $options->{ "beg" } and ( defined $options->{ "end" } or defined $options->{ "len" } ) )
+    if ( defined $options->{ "chr" } and exists $index->{ $options->{ "chr" } } and defined $options->{ "beg" } and ( defined $options->{ "end" } or defined $options->{ "len" } ) )
     {
-        ( $index_beg, $index_len ) = @{ $lookup_hash{ $options->{ "chr" } } };
+        ( $index_beg, $index_len ) = @{ $index->{ $options->{ "chr" } } };
 
         $beg = $index_beg + $options->{ "beg" } - 1;
 
@@ -89,7 +88,9 @@ if ( $options->{ "genome" } )
             $beg = $index_beg;
         }
 
-        $len = $index_beg + $index_len - $beg if $beg + $len > $index_beg + $index_len;
+        if ( $beg + $len > $index_beg + $index_len ) {
+            $len = $index_beg + $index_len - $beg;
+        }
 
         next if $beg > $index_beg + $index_len;
 
@@ -104,28 +105,34 @@ if ( $options->{ "genome" } )
     }   
 }
 
-
 while ( $record = Maasha::Biopieces::get_record( $in ) )
 {
     if ( $options->{ "genome" } and not $record->{ "SEQ" } )
     {
-        if ( $record->{ "REC_TYPE" } eq "BED" and exists $lookup_hash{ $record->{ "CHR" } } )
+        if ( $record->{ "REC_TYPE" } eq "BED" and exists $index->{ $record->{ "CHR" } } )
         {
-            ( $index_beg, $index_len ) = @{ $lookup_hash{ $record->{ "CHR" } } };
+            ( $index_beg, $index_len ) = @{ $index->{ $record->{ "CHR" } } };
         
             $beg = $record->{ "CHR_BEG" } + $index_beg;
             $len = $record->{ "CHR_END" } - $record->{ "CHR_BEG" } + 1;
         }
-        elsif ( $record->{ "REC_TYPE" } eq "PSL" and exists $lookup_hash{ $record->{ "S_ID" } } )
+        elsif ( $record->{ "REC_TYPE" } eq "PSL" and exists $index->{ $record->{ "S_ID" } } )
         {
-            ( $index_beg, $index_len ) = @{ $lookup_hash{ $record->{ "S_ID" } } };
+            ( $index_beg, $index_len ) = @{ $index->{ $record->{ "S_ID" } } };
         
             $beg = $record->{ "S_BEG" } + $index_beg;
             $len = $record->{ "S_END" } - $record->{ "S_BEG" } + 1;
         }
-        elsif ( $record->{ "REC_TYPE" } eq "BLAST" and exists $lookup_hash{ $record->{ "S_ID" } } )
+        elsif ( $record->{ "REC_TYPE" } eq "VMATCH" and exists $index->{ $record->{ "S_ID" } } )
         {
-            ( $index_beg, $index_len ) = @{ $lookup_hash{ $record->{ "S_ID" } } };
+            ( $index_beg, $index_len ) = @{ $index->{ $record->{ "S_ID" } } };
+        
+            $beg = $record->{ "S_BEG" } + $index_beg;
+            $len = $record->{ "S_END" } - $record->{ "S_BEG" } + 1;
+        }
+        elsif ( $record->{ "REC_TYPE" } eq "BLAST" and exists $index->{ $record->{ "S_ID" } } )
+        {
+            ( $index_beg, $index_len ) = @{ $index->{ $record->{ "S_ID" } } };
         
             $beg = $record->{ "S_BEG" } + $index_beg;
             $len = $record->{ "S_END" } - $record->{ "S_BEG" } + 1;
@@ -191,26 +198,22 @@ while ( $record = Maasha::Biopieces::get_record( $in ) )
     Maasha::Biopieces::put_record( $record, $out );
 }
 
+Maasha::Biopieces::close_stream( $in );
+Maasha::Biopieces::close_stream( $out );
+
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
 
 BEGIN
 {
-    $run_time_beg = Maasha::Biopieces::run_time();
-
-    Maasha::Biopieces::log_biopiece();
+    Maasha::Biopieces::status_set();
 }
 
 
 END
 {
-    Maasha::Biopieces::close_stream( $in );
-    Maasha::Biopieces::close_stream( $out );
-
-    $run_time_end = Maasha::Biopieces::run_time();
-
-    Maasha::Biopieces::run_time_print( $run_time_beg, $run_time_end, $options );
+    Maasha::Biopieces::status_log();
 }
 
 
@@ -218,4 +221,3 @@ END
 
 
 __END__
-