refactoring of assemble_pairs

[biopieces.git] / bp_bin / BGB_upload

diff --git a/bp_bin/BGB_upload b/bp_bin/BGB_upload
index 916ffa803bd540c2847e2506292b89a68c19aadc..0d5415cc74632cc751b97d138d4359e99234d247 100755 (executable)
--- a/bp_bin/BGB_upload
+++ b/bp_bin/BGB_upload
@@ -21,7 +21,7 @@
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> DESCRIPTION <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
-# Write Biopiece records to the KISS Genome browser.
+# Write Biopiece records to the Biopieces Genome browser.
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
@@ -33,31 +33,36 @@ use Maasha::Common;
 use Maasha::KISS;
 use Maasha::Biopieces;
 use Maasha::Fasta;
+use Maasha::BGB::Track;
+use Maasha::BGB::Wiggle;
 use Maasha::Filesys;
 
 use constant {
     SEQ_NAME => 0,
     SEQ      => 1,
+
+    S_ID     => 0,
 };
 
 
 # >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
 
 
-my ( $data_dir, $user, $options, $path, $in, $out, $tmp_dir, %fh_hash, $fh_out, $record, $entry, $key, @dirs, $dir, $dst_dir, @nums, $num, $contig_dir );
+my ( $data_dir, $user, $options, $path, $in, $out, $tmp_dir, %fh_hash, $fh_out, $record, $entry, $key, $dst_dir, @nums, $num, $contig_dir, $wig );
 
-$data_dir = Maasha::Biopieces::biopiecesrc( "KISS_DATA_DIR" );
-$user     = Maasha::Biopieces::biopiecesrc( "KISS_USER" );
+$data_dir = Maasha::Biopieces::biopiecesrc( "BGB_DATA_DIR" );
+$user     = Maasha::Biopieces::biopiecesrc( "BGB_USER" );
 
 $options = Maasha::Biopieces::parse_options(
     [
-        { long => 'user',       short => 'u', type => 'string', mandatory => 'no',  default => $user, allowed => undef, disallowed => undef },
-        { long => 'clade',      short => 'c', type => 'string', mandatory => 'yes', default => undef, allowed => undef, disallowed => undef },
-        { long => 'genome',     short => 'g', type => 'string', mandatory => 'yes', default => undef, allowed => undef, disallowed => undef },
-        { long => 'assembly',   short => 'a', type => 'string', mandatory => 'yes', default => undef, allowed => undef, disallowed => undef },
-        { long => 'track_name', short => 't', type => 'string', mandatory => 'no',  default => undef, allowed => undef, disallowed => undef },
-        { long => 'force',      short => 'f', type => 'flag',   mandatory => 'no',  default => undef, allowed => undef, disallowed => undef },
-        { long => 'no_stream',  short => 'x', type => 'flag',   mandatory => 'no',  default => undef, allowed => undef, disallowed => undef },
+        { long => 'user',       short => 'u', type => 'string', mandatory => 'no',  default => $user,    allowed => undef,           disallowed => undef },
+        { long => 'clade',      short => 'c', type => 'string', mandatory => 'yes', default => undef,    allowed => undef,           disallowed => undef },
+        { long => 'genome',     short => 'g', type => 'string', mandatory => 'yes', default => undef,    allowed => undef,           disallowed => undef },
+        { long => 'assembly',   short => 'a', type => 'string', mandatory => 'yes', default => undef,    allowed => undef,           disallowed => undef },
+        { long => 'track_name', short => 't', type => 'string', mandatory => 'no',  default => undef,    allowed => undef,           disallowed => undef },
+        { long => 'track_type', short => 'T', type => 'string', mandatory => 'no',  default => 'linear', allowed => 'linear,wiggle', disallowed => undef },
+        { long => 'force',      short => 'f', type => 'flag',   mandatory => 'no',  default => undef,    allowed => undef,           disallowed => undef },
+        { long => 'no_stream',  short => 'x', type => 'flag',   mandatory => 'no',  default => undef,    allowed => undef,           disallowed => undef },
     ]   
 );
 
@@ -68,7 +73,7 @@ if ( $options->{ 'track_name' } )
 {
     $options->{ 'track_name' } =~ tr/ /_/;
 
-    $path = join "/", $data_dir, "Users/", $options->{ 'user' }, $options->{ 'clade' }, $options->{ 'genome' }, $options->{ 'assembly' };
+    $path = join "/", $data_dir, "Users", $options->{ 'user' }, $options->{ 'clade' }, $options->{ 'genome' }, $options->{ 'assembly' };
 
     Maasha::Common::error( qq(Path not found: "$path") ) if not -d $path;
 
@@ -78,13 +83,13 @@ if ( $options->{ 'track_name' } )
     {
         if ( $entry = Maasha::KISS::biopiece2kiss( $record ) )
         {
-            $entry->{ 'S_ID' } = ( split / /, $entry->{ 'S_ID' } )[ 0 ];
+            $entry->[ S_ID ] = ( split / /, $entry->[ S_ID ] )[ 0 ];
 
-            if ( not exists $fh_hash{ $entry->{ 'S_ID' } } ) {
-                $fh_hash{ $entry->{ 'S_ID' } } = Maasha::Filesys::file_write_open( "$tmp_dir/$entry->{ 'S_ID' }" );
+            if ( not exists $fh_hash{ $entry->[ S_ID ] } ) {
+                $fh_hash{ $entry->[ S_ID ] } = Maasha::Filesys::file_write_open( "$tmp_dir/$entry->[ S_ID ]" );
             }
 
-            $fh_out = $fh_hash{ $entry->{ 'S_ID' } };
+            $fh_out = $fh_hash{ $entry->[ S_ID ] };
 
             Maasha::KISS::kiss_entry_put( $entry, $fh_out );
         }
@@ -92,6 +97,9 @@ if ( $options->{ 'track_name' } )
         Maasha::Biopieces::put_record( $record, $out ) if not $options->{ "no_stream" };
     }
 
+    $num = Maasha::BGB::Track::max_track( $options->{ 'user' }, $options->{ 'clade' }, $options->{ 'genome' }, $options->{ 'assembly' } );
+    $num = sprintf( "%04d", $num + 10 );
+
     foreach $key ( keys %fh_hash )
     {
         close $fh_hash{ $key };
@@ -99,31 +107,21 @@ if ( $options->{ 'track_name' } )
         $dst_dir = Maasha::Filesys::dir_create_if_not_exists( "$path/$key" );
         $dst_dir = Maasha::Filesys::dir_create_if_not_exists( "$dst_dir/Tracks" );
 
-        @dirs = Maasha::Filesys::ls_dirs( $dst_dir );
-
-        foreach $dir ( @dirs )
-        {
-            $dir = ( split "/", $dir )[ -1 ];
-
-            if ( $dir =~ /^(\d+)/ ) {
-                push @nums, $1;
-            }
-        }
-
-        @nums = sort { $a <=> $b } @nums;
-
-        $num = $nums[ -1 ] || 0;
-
-        $num += 10;
-
         $dst_dir = "$dst_dir/$num" . "_$options->{ 'track_name' }";
 
         Maasha::Filesys::dir_create( $dst_dir );
 
-        Maasha::Filesys::file_copy( "$tmp_dir/$key", "$dst_dir/track_data.kiss" );
+        if ( $options->{ 'track_type' } eq 'linear' )
+        {
+            Maasha::Filesys::file_copy( "$tmp_dir/$key", "$dst_dir/track_data.kiss" );
 
-        Maasha::KISS::kiss_sort( "$dst_dir/track_data.kiss" );
-        Maasha::KISS::kiss_index( "$dst_dir/track_data.kiss" );
+            Maasha::KISS::kiss_sort( "$dst_dir/track_data.kiss" );
+        }
+        elsif ( $options->{ 'track_type' } eq 'wiggle' )
+        {
+            $wig = Maasha::BGB::Wiggle::wiggle_encode( "$tmp_dir/$key" );
+            Maasha::BGB::Wiggle::wiggle_store( "$dst_dir/track_data.wig", $wig );
+        }
 
         unlink "$tmp_dir/$key";
     }
@@ -134,7 +132,7 @@ else
 
     Maasha::Common::error( qq(Path not found: "$path") ) if not -d $path;
 
-    $dst_dir = Maasha::Filesys::dir_create( "$path/$options->{ 'genome' }" );
+    $dst_dir = Maasha::Filesys::dir_create_if_not_exists( "$path/$options->{ 'genome' }" );
     $dst_dir = Maasha::Filesys::dir_create( "$dst_dir/$options->{ 'assembly' }" );
 
     while ( $record = Maasha::Biopieces::get_record( $in ) )