adding bzip2 support in ruby

[biopieces.git] / code_perl / Maasha / NCBI.pm

diff --git a/code_perl/Maasha/NCBI.pm b/code_perl/Maasha/NCBI.pm
index a94dc4a88550236db861eb83630198ed6240e36c..331dea351f929fd0f192f2137d04563dfe45a07e 100644 (file)
--- a/code_perl/Maasha/NCBI.pm
+++ b/code_perl/Maasha/NCBI.pm
@@ -28,10 +28,12 @@ package Maasha::NCBI;
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
 
+use warnings;
 use strict;
 use Data::Dumper;
 use LWP::Simple;
 use Maasha::Common;
+use Maasha::Filesys;
 use Maasha::Seq;
 
 use vars qw( @ISA @EXPORT );
@@ -107,134 +109,26 @@ sub get_seq
 }
 
 
-sub soft_parse
-{
-    # Martin A. Hansen, February 2008.
-
-    # !!! NOT USED ANYMORE !!! #
-
-    # Reads in and parses a file in SOFT format.
-
-    my ( $path,   # full path to SOFT file
-       ) = @_;
-
-    # Returns a list.
-
-    my ( $fh, @lines, $i, $c, $num, %key_hash, @fields, %id_hash, $id, $seq, $count, $record, @records, $platform_id, $sample_id, $sample_title );
-
-    $fh = Maasha::Common::read_open( $path );
-
-    @lines = <$fh>;
-
-    close $fh;
-
-    chomp @lines;
-
-    $i = 0;
-
-    $num = 1;
-
-    while ( $i < @lines ) 
-    {
-        if ( $lines[ $i ] =~ /^\^PLATFORM = (.+)/ )
-        {
-            $platform_id = $1;
-        }
-        elsif ( $lines[ $i ] =~ /^!platform_table_begin$/ )
-        {
-            @fields = split "\t", $lines[ $i + 1 ];
-
-            for ( $c = 0; $c < @fields; $c++ ) {
-                $key_hash{ $fields[ $c ] } = $c;
-            }
-
-            $c = $i + 2;
-
-            while ( $lines[ $c ] !~ /^!platform_table_end$/ )
-            {
-                @fields = split "\t", $lines[ $c ];
-
-                $id_hash{ $fields[ $key_hash{ "ID" } ] } = $fields[ $key_hash{ "SEQUENCE" } ];
-
-                $c++;
-            }
-
-            $i = $c;
-        }
-        elsif ( $lines[ $i ] =~ /^\^SAMPLE = (.+)/ )
-        {
-            $sample_id = $1;
-        }
-        elsif ( $lines[ $i ] =~ /^!Sample_title = (.+)/ )
-        {
-            $sample_title = $1;
-        }
-        elsif ( $lines[ $i ] =~ /^!sample_table_begin/ )
-        {
-            undef %key_hash;
-
-            @fields = split "\t", $lines[ $i + 1 ];
-
-            for ( $c = 0; $c < @fields; $c++ ) {
-                $key_hash{ $fields[ $c ] } = $c;
-            }
-
-            $c = $i + 2;
-
-            while ( $lines[ $c ] !~ /^!sample_table_end$/ )
-            {
-                undef $record;
-
-                @fields = split "\t", $lines[ $c ];
-    
-                $id    = $fields[ $key_hash{ "ID_REF" } ];
-                $seq   = $id_hash{ $id };
-                $count = $fields[ $key_hash{ "VALUE" } ];
-
-                $seq =~ tr/./N/;
-
-                $record->{ "SAMPLE_TITLE" } = $sample_title;
-                $record->{ "SEQ" }          = $seq;
-                $record->{ "SEQ_NAME" }     = join( "_", $platform_id, $sample_id, $num, $count );
-            
-                push @records, $record;
-
-                $c++;
-                $num++;
-            }
-
-            $i = $c;
-       
-            $num = 1;
-        }
-
-        $i++;
-    }
-
-    return wantarray ? @records : \@records;
-}
-
-
 sub soft_index_file
 {
     # Martin A. Hansen, June 2008.
 
-    # Create a index with linenumbers of the different tables
+    # Create a index with line numbers of the different tables
     # in a soft file. The index is returned.
 
     my ( $file,   # file to index
        ) = @_;
 
-    # Returns
+    # Returns a list
     
     my ( $fh, $line, $i, $c, @index, $first );
 
-    $fh = Maasha::Common::read_open( $file );
+    $fh = Maasha::Filesys::file_read_open( $file );
 
     $first = 1;
 
-    $i      = 0;
-    $c      = 0;
+    $i     = 0;
+    $c     = 0;
 
     while ( $line = <$fh> )
     {
@@ -317,8 +211,12 @@ sub soft_get_platform
             {
                 @fields = split "\t", $lines[ $c ];
 
-                $id_hash{ $fields[ $key_hash{ "ID" } ] } = $fields[ $key_hash{ "SEQUENCE" } ] || $fields[ $key_hash{ "ID" } ];
-            
+                if ( defined $key_hash{ "SEQUENCE" } ) {
+                    $id_hash{ $fields[ $key_hash{ "ID" } ] } = $fields[ $key_hash{ "SEQUENCE" } ];
+                } else {
+                    $id_hash{ $fields[ $key_hash{ "ID" } ] } = $fields[ $key_hash{ "ID" } ];
+                }
+
                 $c++;
             }
 
@@ -394,7 +292,11 @@ sub soft_get_sample
 
             @fields = split "\t", $lines[ $i + 1 ];
 
-            for ( $c = 0; $c < @fields; $c++ ) {
+            for ( $c = 0; $c < @fields; $c++ )
+            {
+                $fields[ $c ] = "ID_REF" if $fields[ $c ] eq "SEQUENCE";
+                $fields[ $c ] = "VALUE"  if $fields[ $c ] eq "COUNT";
+
                 $key_hash{ $fields[ $c ] } = $c;
             }
 
@@ -405,9 +307,8 @@ sub soft_get_sample
                 undef $record;
 
                 @fields = split "\t", $lines[ $c ];
-    
                 $id    = $fields[ $key_hash{ "ID_REF" } ];
-                $seq   = $plat_table->{ $id };
+                $seq   = $plat_table->{ $id } || $id;
                 $count = $fields[ $key_hash{ "VALUE" } ];
 
                 $seq =~ tr/./N/;
@@ -458,7 +359,7 @@ sub blast_index
         $type = "F";
     }
 
-    Maasha::Common::dir_create_if_not_exists( $dst_dir );
+    Maasha::Filesys::dir_create_if_not_exists( $dst_dir );
     Maasha::Common::run( "formatdb", "-p $type -i $src_dir/$file -t $index_name -l /dev/null" );
     Maasha::Common::run( "mv", "$src_dir/*hr $dst_dir" );
     Maasha::Common::run( "mv", "$src_dir/*in $dst_dir" );