move refSeqs.json into the seq/ subdirectory, leaving just trackList.…

…json in the top level of the data
Arabidopsis-Information-Portal · Feb 29, 2012 · cc4b505 · cc4b505
1 parent de19dbd
commit cc4b505
Show file tree

Hide file tree

Showing 12 changed files with 115 additions and 70 deletions.
diff --git a/bin/bam-to-json.pl b/bin/bam-to-json.pl
@@ -103,7 +103,9 @@ =head1 OPTIONS
     $nclChunk *= 4 if $compress;
 }
 
-my @refSeqs = @{JSON::from_json( slurp("$outdir/refSeqs.json") || '[]' )}
+my $gdb = GenomeDB->new( $outdir );
+
+my @refSeqs = @{ $gdb->refSeqs }
   or die "Run prepare-refseqs.pl to define reference sequences before running this script.\n";
 
 my %config = (
@@ -127,7 +129,6 @@ =head1 OPTIONS
 my $hdr = $bam->header;
 
 
-my $gdb = GenomeDB->new( $outdir );
 my $track = $gdb->getTrack( $trackLabel )
             || $gdb->createFeatureTrack( $trackLabel,
                                          \%config,

diff --git a/bin/biodb-to-json.pl b/bin/biodb-to-json.pl
@@ -98,9 +98,10 @@ =head2 OPTIONS
     $nclChunk *= 4 if $compress;
 }
 
+my $gdb = GenomeDB->new( $outdir );
 
 # determine which reference sequences we'll be operating on
-my @refSeqs = @{JsonGenerator::readJSON("$outdir/refSeqs.json", [], 1)};
+my @refSeqs = @{ $gdb->refSeqs };
 if (defined $refid) {
     @refSeqs = grep { $_->{id} eq $refid } @refSeqs;
     die "Didn't find a refseq with ID $refid (have you run prepare-refseqs.pl to supply information about your reference sequences?)" if $#refSeqs < 0;
@@ -125,7 +126,6 @@ =head2 OPTIONS
 $db->strict_bounds_checking(1) if $db->can('strict_bounds_checking');
 $db->absolute(1)               if $db->can('absolute');
 
-my $gdb = GenomeDB->new( $outdir );
 
 foreach my $seg (@refSeqs) {
     my $segName = $seg->{name};

diff --git a/bin/flatfile-to-json.pl b/bin/flatfile-to-json.pl
@@ -203,12 +203,12 @@ =head2 BED-SPECIFIC
 
 pod2usage( -verbose => 2 ) if $help;
 
-my %refSeqs =
-    map {
-        $_->{name} => $_
-    } @{JsonGenerator::readJSON("$outdir/refSeqs.json", [], 1)};
+my $gdb = GenomeDB->new( $outdir );
+
+my %refSeqs = map { $_->{name} => $_ } @{ $gdb->refSeqs };
 
-die "run prepare-refseqs.pl first to supply information about your reference sequences" unless (scalar keys %refSeqs);
+scalar keys %refSeqs
+    or die "run prepare-refseqs.pl first to supply information about your reference sequences";
 
 pod2usage( "Must provide a --tracklabel parameter." ) unless defined $trackLabel;
 pod2usage( "You must supply either a --gff or --bed parameter." )
@@ -367,7 +367,6 @@ =head2 BED-SPECIFIC
 
 ################################
 
-my $gdb = GenomeDB->new( $outdir );
 my $track = $gdb->getTrack( $trackLabel )
             || $gdb->createFeatureTrack( $trackLabel,
                                          \%config,

diff --git a/bin/generate-names.pl b/bin/generate-names.pl
@@ -51,7 +51,7 @@ =head1 OPTIONS
 use lib catdir($Bin, updir(), "lib");
 
 use LazyPatricia;
-use JsonGenerator qw/ readJSON writeJSON /;
+use GenomeDB;
 
 my %trackHash;
 my @tracksWithNames;
@@ -76,24 +76,12 @@ =head1 OPTIONS
 OUTDIR
 }
 
+my $gdb = GenomeDB->new( $outDir );
 my $nameDir = catdir($outDir, "names");
 mkdir($nameDir) unless (-d $nameDir);
 
-sub partitionCallback {
-    my ($subtreeRoot, $prefix, $thisChunk, $total) = @_;
-    # output subtree
-    writeJSON(catfile($nameDir, "lazy-$prefix.json"),
-              $subtreeRoot,
-              {pretty => 0});
-    printf STDERR "subtree for %15s   has %10d  in chunk, %10d  total\n",
-      $prefix, $thisChunk, $total
-        if $verbose;
-}
-
-my @refSeqs = @{readJSON(catfile($outDir, "refSeqs.json"), [], 1)};
-my $trackList = readJSON(catfile($outDir, "trackList.json"),
-                           {"tracks" => []}, 1);
-my @tracks = @{$trackList->{tracks}};
+my @refSeqs  = @{ $gdb->refSeqs   };
+my @tracks   = @{ $gdb->trackList };
 
 # open the root file; we lock this file while we're
 # reading the name lists, deleting all the old lazy-*
@@ -106,59 +94,74 @@ sub partitionCallback {
 # read the name list for each track that has one
 my %nameHash;
 my $trackNum = 0;
-my $OLDSEP = $/;
-undef $/;
 foreach my $ref (@refSeqs) {
     foreach my $track (@tracks) {
-        my $infile = catfile($outDir,
-                             "tracks",
-                             $track->{label},
-                             $ref->{name},
-                             "names.json");
+        my $infile = catfile( $outDir,
+                              "tracks",
+                              $track->{label},
+                              $ref->{name},
+                              "names.json");
         next unless -e $infile;
-        open JSON, "<$infile"
-            or die "couldn't open $: $!";
-        my $names = JSON::from_json(<JSON>);
-        close JSON
-            or die "couldn't close $infile: $!";
-        foreach my $nameinfo (@$names) {
-            foreach my $alias (@{$nameinfo->[0]}) {
+
+        my $names = do {
+            local $/;
+            open my $j, '<', $infile or die "$! reading $infile";
+            JSON::from_json(<$j>);
+        };
+
+        foreach my $nameinfo ( @$names) {
+            foreach my $alias ( @{$nameinfo->[0]} ) {
                 my $track = $nameinfo->[1];
-                if (!defined($trackHash{$track})) {
+                unless( defined $trackHash{$track} ) {
                     $trackHash{$track} = $trackNum++;
                     push @tracksWithNames, $track;
                 }
 
-                push @{$nameHash{lc $alias}}, [$trackHash{$track},
-                                               @{$nameinfo}[2..$#{$nameinfo}]];
+                push @{$nameHash{lc $alias}}, [ $trackHash{$track},
+                                                @{$nameinfo}[2..$#{$nameinfo}]];
             }
         }
     }
 }
-$/ = $OLDSEP;
 
 # clear out old data
 $root->seek(0, SEEK_SET);
 $root->truncate(0);
-my @lazyFiles = glob(catfile($nameDir, "lazy-*"));
-if (@lazyFiles) {
+if( my @lazyFiles = glob( catfile( $nameDir, "lazy-*" )) ) {
     unlink @lazyFiles
       or die "couldn't unlink name files: $!";
 }
 
-my $trie = LazyPatricia::create(\%nameHash);
+my $trie = LazyPatricia::create( \%nameHash );
 $trie->[0] = \@tracksWithNames;
 
 my ($total, $thisChunk) =
-  LazyPatricia::partition($trie, "", $thresh, \&partitionCallback);
+  LazyPatricia::partition( $trie, "", $thresh, sub {
+      my ($subtreeRoot, $prefix, $thisChunk, $total) = @_;
+      # output subtree
+      writeJSON( catfile($nameDir, "lazy-$prefix.json"),
+                 $subtreeRoot,
+                 {pretty => 0} );
+      printf STDERR ( "subtree for %15s   has %10d  in chunk, %10d  total\n",
+                      $prefix, $thisChunk, $total )
+          if $verbose;
+  });
 
 print STDERR "$total total names, with $thisChunk in the root chunk\n"
   if $verbose;
 
 # write the root
-$root->print(JSON::to_json($trie, {pretty => 0}));
-$root->close()
-  or die "couldn't close $rootFile: $!";
+$root->print( JSON::to_json($trie, {pretty => 0}) );
+
+exit;
+
+############
+
+sub writeJSON {
+    my ( $file, $contents, $opts ) = @_;
+    open my $f, '>', $file or die "$! writing $file";
+    print $f JSON::to_json( $contents, $opts || {} );
+}
 
 
 =head1 AUTHOR

diff --git a/bin/prepare-refseqs.pl b/bin/prepare-refseqs.pl
@@ -204,7 +204,7 @@ =head1 OPTIONS
     exit;
 }
 
-JsonGenerator::modifyJsonFile( catfile( $outDir, 'refSeqs.json' ),
+JsonGenerator::modifyJsonFile( catfile( $outDir, 'seq', 'refSeqs.json' ),
                               sub {
                                   #add new ref seqs while keeping the order
                                   #of the existing ref seqs

diff --git a/bin/ucsc-to-json.pl b/bin/ucsc-to-json.pl
@@ -304,7 +304,6 @@ =head1 EXAMPLE
                 last;
             }
         }
-        # next unless defined($refSeqs{$curChrom});
         my $jsonRow = $converter->($row, \%fields, $type);
         $track->addSorted($jsonRow);
         if (defined $nameCol) {

diff --git a/bin/wig-to-json.pl b/bin/wig-to-json.pl
@@ -128,14 +128,14 @@ =head1 OPTIONS
 pod2usage( -verbose => 2 ) if $help;
 pod2usage( 'Must provide a --wig argument.' ) unless defined $path;
 
-my @refSeqs = @{ JSON::from_json(do{ open my $f, '<', "$outdir/refSeqs.json"; local $/; <$f>}) || [] }
+my $gdb = GenomeDB->new( $outdir );
+
+my @refSeqs = @{ $gdb->refSeqs }
    or die "Run prepare-refseqs.pl first to supply information about your reference sequences.\n";
 
 $trackLabel = basename( $path ) unless defined $trackLabel;
 my $urlTemplate = "tracks/$trackLabel/{refseq}/trackData.json";
 
-my $gdb = GenomeDB->new( $outdir );
-
 my %style = (
     "key"            => defined($key) ? $key : $trackLabel,
     "urlTemplate"    => $urlTemplate,

diff --git a/index.html b/index.html
@@ -53,7 +53,7 @@
            var dataRoot = queryParams.data || 'data';
            var b = new Browser({
                                    containerID: "GenomeBrowser",
-                                   refSeqs: {url: dataRoot + "/refSeqs.json"},
+                                   refSeqs: {url: dataRoot + "/seq/refSeqs.json"},
                                    tracklists: [
                                        {url: dataRoot + "/trackList.json"}
                                    ],

diff --git a/lib/GenomeDB.pm b/lib/GenomeDB.pm
@@ -171,6 +171,59 @@ sub trackDir {
     return $result;
 }
 
+=head2 refSeqs
+
+Returns a arrayref of hashrefs defining the reference sequences, as:
+
+    [ {
+        name         => 'ctgB',
+        seqDir       => 'seq/ctgB',
+
+        start        => 0
+        end          => 66,
+        length       => 66,
+
+        seqChunkSize => 20000,
+      },
+      ...
+    ]
+
+=cut
+
+sub refSeqs {
+    shift->{rootStore}->get( 'seq/refSeqs.json', [] );
+}
+
+
+=head2 trackList
+
+Return an arrayref of track definition hashrefs similar to:
+
+    [
+        {
+          config => {
+                      autocomplete => "all",
+                      compress => 0,
+                      feature => ["remark"],
+                      key => "Example Features",
+                      style => { className => "feature2" },
+                      track => "ExampleFeatures",
+                      urlTemplate => "tracks/ExampleFeatures/{refseq}/trackData.json",
+                    },
+          key    => "Example Features",
+          label  => "ExampleFeatures",
+          type   => "FeatureTrack",
+        },
+        ...
+    ]
+
+=cut
+
+sub trackList {
+    shift->{rootStore}->get( 'trackList.json', { tracks => [] } )->{tracks}
+}
+
+
 1;
 
 =head1 AUTHOR

diff --git a/lib/ImageTrackRenderer.pm b/lib/ImageTrackRenderer.pm
@@ -91,7 +91,7 @@ sub new {
         'datadir'       => "data",
         'trackdir'      => "tracks",
         'tiledir'       => undef, #< ignored for backcompat
-        'refseqsfile'   => 'refSeqs.json',
+        'refseqsfile'   => undef,
         'trackinfofile' => 'trackList.json',
         'zooms'         =>
             [ 1, 2, 5, 10, 20, 50, 100, 200, 500, 1000, 2000, 5000, 10000, 20000, 50000, 100000 ],
@@ -139,7 +139,7 @@ then adds the track to the data/trackList.json file.
 
 sub render {
     my ($self) = @_;
-    my @refSeqs = $self->refseqs
+    my @refSeqs = @{ $self->_genomedb->refSeqs }
         or die "No reference sequences defined";
 
     foreach my $seqInfo (@refSeqs) {
@@ -327,23 +327,13 @@ sub tilefilepath {
     $self->_file( $self->datadir, $self->trackdir, $self->tilefile( @_ ));
 }
 
-sub refseqspath {
-    my ( $self ) = @_;
-    $self->_file( $self->datadir, $self->refseqsfile );
-}
-sub refseqs {
-    my ($self) = @_;
-    my $contents = eval { open my $f, '<',$self->refseqspath; local $/; <$f> };
-    return @{ JSON::from_json( $contents || '[]' ) };
-}
-
 ######### read-only accessors
 
 sub link        { shift->{link}        }
 sub datadir     { shift->{datadir}     }
 sub tracklabel  { shift->{tracklabel}  }
 sub key         { shift->{key}         }
-sub refseqsfile { shift->{refseqsfile} }
+sub refseqsfile { undef                } #< only for backcompat
 sub trackdir    { shift->{trackdir}    }
 sub tilewidth   { shift->{tilewidth}   }
 sub zooms       { shift->{zooms}       }

diff --git a/...se_pairing/mahoney_formatted/refSeqs.json → ...airing/mahoney_formatted/seq/refSeqs.json b/...se_pairing/mahoney_formatted/refSeqs.json → ...airing/mahoney_formatted/seq/refSeqs.json
diff --git a/...ata/volvox_formatted_refseqs/refSeqs.json → ...volvox_formatted_refseqs/seq/refSeqs.json b/...ata/volvox_formatted_refseqs/refSeqs.json → ...volvox_formatted_refseqs/seq/refSeqs.json