# BioMirror/MeowGenes.pm

=head1 NAME

BioMirror/MeowGenes.pm

=head1 DESCRIPTION

A collection of BioMirror::Data packages for building 
MeowGenes databank.  Includes

 BioMirror::TAIR -- for TAIR Arabidopsis databank
 BioMirror::CElegans -- AceDB for C. elegans databank
 BioMirror::MGD -- for MGD Mouse databank
 BioMirror::LocusLink -- for NCBI LocusLink Human databank
 See also BioMirror::FlybaseGenes

mar02 - changed 'meowgenes' to 'eugenes' path

=cut

package BioMirror::euGenes;
 @ISA = qw( BioMirror::Data ); 

sub new {
	my $class= shift;
	my $self = $class->SUPER::new( 
		srsdb => 'MEOW',
		name => 'euGenes',
		
		dosummary => 1, # add to summary table
   	source => {
   		home => 'IUBio',
   		comment => 'Eukaryote Genes Summary Databank',
   		url => 'ftp://iubio.bio.indiana.edu/biomirror/eugenes/',
   		web => 'http://iubio.bio.indiana.edu/eugenes/',
			},
		mirror_dir =>	'$zpath/eugenes/', 
		@_ );
	return $self;
}

package BioMirror::TAIR;
 @ISA = qw( BioMirror::Data ); 
## for now don't srsindex

sub new {
	my $class= shift;
	my $self = $class->SUPER::new( 
		srsdb => 'TAIR',
		name => 'TAIR Arabidopsis Genome',
		makeflags => $BioMirror::Data::kFromArchive, 
			## |$BioMirror::Data::kDoSrsIndex -- not here

		dosummary => 0, 
   	source => {
   		comment => 'Arabidopsis locus data from TAIR',
   		url => 'ftp://tairpub:tairpub@ftp.arabidopsis.org/home/tair/Locus/',
   		web => 'http://www.arabidopsis.org/',
			get_patt => '(locus.tab$|locus_on_meinke.tab$|README)',
     	recursive => 'false', 
			},
			
		mirror_dir =>	'$zpath/eugenes/arabidopsis/', 
		expanded_dir =>	'', # '$dpath/arabidopsis/',
		data => [], # [ '(\.tab)', ],
		@_ );
	return $self;
}

 
package BioMirror::CElegans;
 @ISA = qw( BioMirror::Data ); 
## for now don't srsindex

sub new {
	my $class= shift;
	my $self = $class->SUPER::new( 
		srsdb => 'CELEGANS',
		name => 'C elegans AceDB',
		makeflags => $BioMirror::Data::kFromArchive, 
			## |$BioMirror::Data::kDoSrsIndex -- not here

		dosummary => 0, 
   	source => {
   		comment => 'C. elegans WormBase data from Sanger',
		web => 'http://wormbase.sanger.ac.uk/',
		##web => 'http://www.sanger.ac.uk/Projects/C_elegans/',
   		url => 'ftp://ftp.sanger.ac.uk/pub/wormbase/current_release/',
     	get_patt => '(^database.WS.*.tar.Z|wspec|CHROMOSOME|README|INSTALL)',
	exclude_patt => '(pictures|additional_files|core)',
     	recursive => 'true', 
			},
			
        source0 => {
   		comment => 'C. elegans AceDB data from NCBI',
   		url => 'ftp://ncbi.nlm.nih.gov/repository/acedb/celegans/',
     	get_patt => '(^database.WS.*.tar.Z|README|INSTALL)',
     	#? add pictures/ ?
     	recursive => 'false', 
			},
			
		mirror_dir =>	'$zpath/eugenes/celegans/', 
		expanded_dir =>	'', ## '$dpath/eugenes/celegans/',
		data => [], ## [ '(database)', ],
		@_ );
	return $self;
}


package BioMirror::MGD;
 @ISA = qw( BioMirror::Data ); 
## for now don't srsindex

sub new {
	my $class= shift;
	my $self = $class->SUPER::new( 
		srsdb => 'MGD',
		name => 'MGD Mouse Genome',
		makeflags => $BioMirror::Data::kFromArchive, 
			## |$BioMirror::Data::kDoSrsIndex -- not here

		dosummary => 0, 
   	source => {
   		comment => 'Mouse Genome data from MGD',
   		url => 'ftp://ftp.informatics.jax.org/pub/informatics/reports/',
   		web => 'http://www.informatics.jax.org/',
   		## get only useful ones, most others are redundant w/ these
     	get_patt => '(GDB_Accession|MRK_Dump2|MRK_MLC|MRK_Sequence|MRK_SwissProt|index.html|tab-delimited.html)',
     	recursive => 'false', 
			},
			
		mirror_dir =>	'$zpath/eugenes/mouse/', 
		expanded_dir =>	'', # '$dpath/eugenes/mouse/',
		data => [], # [ '(\.rpt)', ],
		@_ );
	return $self;
}


package BioMirror::ZFIN;
@ISA = qw( BioMirror::Data ); 

sub new {
	my $class= shift;
	my $self = $class->SUPER::new( 
		srsdb => 'ZFIN',
		name => 'ZFIN Zebrafish Genome',
		makeflags => $BioMirror::Data::kFromArchive, 
			## |$BioMirror::Data::kDoSrsIndex -- not here

		dosummary => 0, 
   	source => {
   		comment => 'Zebrafish Genome data from ZFIN',
   		url => 'ftp://zfish.uoregon.edu/transfer/MEOW/',
   		web => 'http://zfish.uoregon.edu/ZFIN/',
			# get_patt => '(*)',# get all
     	# recursive => 'false', 
			},
			
		mirror_dir =>	'$zpath/eugenes/fish/', 
		expanded_dir =>	'', # '$dpath/eugenes/fish/',
		data => [], # [ '(\.txt)', ],
		doc => [], # [ 'README', ],
		@_ );
	return $self;
}




package BioMirror::LocusLink;
 @ISA = qw( BioMirror::Data ); 

sub new {
	my $class= shift;
	my $self = $class->SUPER::new( 
		srsdb => 'LOCUSLINK',
		name => 'LocusLink Human Genome',
		makeflags => $BioMirror::Data::kFromArchive|$BioMirror::Data::kDoSrsIndex,

		dosummary => 0, 
		mirror_dir =>	'$zpath/eugenes/human/',
		expanded_dir => '$dpath/locuslink/',
   	source => {
   		comment => 'LocusLink data from NCBI',
   		url => 'ftp://ncbi.nlm.nih.gov/refseq/LocusLink/',
   		web => 'http://www.ncbi.nlm.nih.gov/LocusLink/',
			get_patt => '(LL_tmpl.gz$|README)',
			exclude_patt => '(^Old)',
		  recursive => 'false',
			},

		data => [ 'LL_tmpl' ],
		doc => [ 'README' ],
		@_ );
	return $self;
}


sub FIXME_getRelease { 
			# Release 25.00, June 30, 1998 
	my $self= shift;
	my $rel= undef;
	my $df= BioMirror::replaceVars( $self->expanded_dir . ${$self->data}[0]);
	my $buf= $self->readChunk($df);
	if ($buf =~ m/Release ([0-9\.]+), ([0-9A-z\, ]+)/) {
		$rel= "$1, $2";
		}
	return $rel;
}


package BioMirror::SGD;
 @ISA = qw( BioMirror::Data ); 
## for now don't srsindex

sub new {
	my $class= shift;
	my $self = $class->SUPER::new( 
		srsdb => 'SGD',
		name => 'SGD Yeast Genome',
		makeflags => $BioMirror::Data::kFromArchive, 
			## |$BioMirror::Data::kDoSrsIndex -- not here

		dosummary => 0, 
   	source => {
   		comment => 'Saccharomyces genome data from SGD',
   		url => 'ftp://genome-ftp.stanford.edu/pub/yeast/',
   		web => 'http://genome-www.stanford.edu/Saccharomyces/', 
			get_patt => '(gene_registry/registry.genenames.tab|gene_registry/READ_ME'
				.'|ORF_Descriptions/orf_descriptions.tab|ORF_Descriptions/README'
				.'|ORF_Locations/ORF_chr.*\.txt|ORF_Locations/README)',
		  recursive => 'true',
   		## ftp://genome-ftp.stanford.edu/pub/yeast/gene_registry/registry.genenames.tab
 			## ftp://genome-ftp.stanford.edu/pub/yeast/tables/ORF_Descriptions/orf_descriptions.tab
			## ftp://genome-ftp.stanford.edu/pub/yeast/tables/ORF_Locations/ORF_chr.*\.txt
			},
			
		mirror_dir =>	'$zpath/eugenes/yeast/', 
		expanded_dir =>	'', # '$dpath/eugenes/yeast/',
		data => [], 
		@_ );
	return $self;
}


1;