From: martinahansen <martinahansen@74ccb610-7750-0410-82ae-013aeee3265d>
Date: Mon, 14 Jul 2008 08:32:34 +0000 (+0000)
Subject: added phastcons to format_genome
X-Git-Url: https://git.donarmstrong.com/?a=commitdiff_plain;h=4f7a684fc91d1759b00d78ac98bc4a3fddff9748;p=biopieces.git

added phastcons to format_genome

git-svn-id: http://biopieces.googlecode.com/svn/trunk@149 74ccb610-7750-0410-82ae-013aeee3265d
---

diff --git a/code_perl/Maasha/Biopieces.pm b/code_perl/Maasha/Biopieces.pm
index d4ccc17..2a7ef7d 100644
--- a/code_perl/Maasha/Biopieces.pm
+++ b/code_perl/Maasha/Biopieces.pm
@@ -2020,7 +2020,7 @@ sub script_format_genome
 
     # Returns nothing.
 
-    my ( $dir, $genome, $fasta_dir, $fh_out, $record, $format, $index );
+    my ( $dir, $genome, $fasta_dir, $phastcons_dir, $vals, $fh_out, $record, $format, $index );
 
     $dir    = $options->{ 'dir' } || $ENV{ 'BP_DATA' };
     $genome = $options->{ 'genome' };
@@ -2029,40 +2029,57 @@ sub script_format_genome
     Maasha::Common::dir_create_if_not_exists( "$dir/genomes" );
     Maasha::Common::dir_create_if_not_exists( "$dir/genomes/$genome" );
 
-    if ( -f "$dir/genomes/$genome/fasta/$genome.fna" )
+    if ( grep { $_ =~ /fasta|blast|vmatch/i } @{ $options->{ "formats" } } )
     {
-        $fasta_dir = "$dir/genomes/$genome/fasta";
-    }
-    elsif ( grep { $_ =~ /fasta/i } @{ $options->{ "formats" } } )
-    {
-        Maasha::Common::dir_create_if_not_exists( "$dir/genomes/$genome/fasta" );
+        if ( -f "$dir/genomes/$genome/fasta/$genome.fna" )
+        {
+            $fasta_dir = "$dir/genomes/$genome/fasta";
+        }
+        else
+        {
+            Maasha::Common::dir_create_if_not_exists( "$dir/genomes/$genome/fasta" );
 
-        $fasta_dir = "$dir/genomes/$genome/fasta";
+            $fasta_dir = "$dir/genomes/$genome/fasta";
 
-        $fh_out = Maasha::Common::write_open( "$fasta_dir/$genome.fna" );
+            $fh_out = Maasha::Common::write_open( "$fasta_dir/$genome.fna" );
+        }
     }
-    else
+    elsif ( grep { $_ =~ /phastcons/i } @{ $options->{ "formats" } } )
     {
-        $fasta_dir = $BP_TMP;
+        Maasha::Common::dir_create_if_not_exists( "$dir/genomes/$genome/phastcons" );
+    
+        $phastcons_dir = "$dir/genomes/$genome/phastcons";
 
-        $fh_out = Maasha::Common::write_open( "$fasta_dir/$genome.fna" );
+        $fh_out = Maasha::Common::write_open( "$phastcons_dir/$genome.pp" );
     }
 
     while ( $record = get_record( $in ) ) 
     {
-        if ( $fh_out and $record->{ "SEQ_NAME" } and $record->{ "SEQ" } ) {
+        if ( $fh_out and $record->{ "SEQ_NAME" } and $record->{ "SEQ" } )
+        {
             Maasha::Fasta::put_entry( [ $record->{ "SEQ_NAME" }, $record->{ "SEQ" } ], $fh_out, $options->{ "wrap" } );
         }
+        elsif ( $fh_out and $record->{ "CHR" } and $record->{ "CHR_BEG" } and $record->{ "STEP" } and $record->{ "VALS" } )
+        {
+            print $fh_out "fixedStep chrom=$record->{ 'CHR' } start=$record->{ 'CHR_BEG' } step=$record->{ 'STEP' }\n";
+
+            $vals = $record->{ 'VALS' };
+
+            $vals =~ tr/,/\n/;
+
+            print $fh_out "$vals\n";
+        }
 
         put_record( $record, $out ) if not $options->{ "no_stream" };
     }
 
     foreach $format ( @{ $options->{ 'formats' } } )
     {
-        if    ( $format =~ /^fasta$/i )  { Maasha::Fasta::fasta_index( "$fasta_dir/$genome.fna", "$dir/genomes/$genome/fasta/$genome.index" ) }
-        elsif ( $format =~ /^blast$/i )  { Maasha::NCBI::blast_index( "$genome.fna", $fasta_dir, "$dir/genomes/$genome/blast", "dna", $genome ) }
-        elsif ( $format =~ /^blat$/i )   { print STDERR "BLAT FORMAT NOT IMPLEMENTED" }
-        elsif ( $format =~ /^vmatch$/i ) { Maasha::Match::vmatch_index( "$genome.fna", $fasta_dir, "$dir/genomes/$genome/vmatch", $BP_TMP ) }
+        if    ( $format =~ /^fasta$/i )     { Maasha::Fasta::fasta_index( "$fasta_dir/$genome.fna", "$dir/genomes/$genome/fasta/$genome.index" ) }
+        elsif ( $format =~ /^blast$/i )     { Maasha::NCBI::blast_index( "$genome.fna", $fasta_dir, "$dir/genomes/$genome/blast", "dna", $genome ) }
+        elsif ( $format =~ /^blat$/i )      { print STDERR "BLAT FORMAT NOT IMPLEMENTED" }
+        elsif ( $format =~ /^vmatch$/i )    { Maasha::Match::vmatch_index( "$genome.fna", $fasta_dir, "$dir/genomes/$genome/vmatch", $BP_TMP ) }
+        elsif ( $format =~ /^phastcons$/i ) { Maasha::UCSC::phastcons_index( "$genome.pp", $phastcons_dir ) }
     }
 
     close $fh_out if $fh_out;
diff --git a/code_perl/Maasha/UCSC.pm b/code_perl/Maasha/UCSC.pm
index 239dea1..dfc5255 100644
--- a/code_perl/Maasha/UCSC.pm
+++ b/code_perl/Maasha/UCSC.pm
@@ -1098,6 +1098,26 @@ sub fixedStep_index_lookup
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> PhastCons format <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
 
+sub phastcons_index
+{
+    # Martin A. Hansen, July 2008
+
+    # Create a fixedStep index for PhastCons data.
+
+    my ( $file,   # file to index
+         $dir,    # dir with file
+       ) = @_;
+
+    # Returns nothing.
+
+    my ( $index );
+
+    $index = fixedstep_index_create( "$dir/$file" );
+
+    fixedstep_index_store( "$dir/$file.index", $index );
+}
+
+
 sub phastcons_parse_entry
 {
     # Martin A. Hansen, December 2007.