script for propogating external_db changes to all core databases

misc-scripts/external_db/external_dbs.txt

+100	AFFY_HG_U133	1	XREF
+110	AFFY_HG_U95	1	XREF
+120	AFFY_MG_U74	1	XREF
+130	AFFY_MG_U74v2	1	XREF
+140	AFFY_Mu11Ksub	1	XREF
+150	AFFY_RG_U34	1	XREF
+160	AFFY_RN_U34	1	XREF
+170	AFFY_RT_U34	1	XREF
+200	Anopheles_paper	1	KNOWNXREF
+210	Anopheles_symbol	1	KNOWN
+300	BRIGGSAE_HYBRID	1	KNOWNXREF
+400	Celera_Gene	1	PRED
+410	Celera_Pep	1	PRED
+420	Celera_Trans	1	PRED
+500	drosophila_gene_id	1	KNOWNXREF
+600	DROS_ORTH	1	ORTH
+700	EMBL	1	KNOWNXREF
+800	flybase_gene	1	KNOWNXREF
+810	flybase_symbol	1	KNOWNXREF
+820	flybase_transcript	1	KNOWNXREF
+900	GKB	1	XREF
+1000	GO	1	KNOWNXREF
+1100	HUGO	1	KNOWNXREF
+1200	Interpro	1	XREF
+1300	LocusLink	1	KNOWNXREF
+1400	MarkerSymbol	1	KNOWNXREF
+1500	MIM	1	KNOWNXREF
+1600	PDB	1	KNOWNXREF
+1700	protein_id	1	KNOWNXREF
+1800	RefSeq	1	KNOWN
+1900	Sanger_Hver1_2_1	1	XREF
+1910	Sanger_Mver1_1_1	1	XREF
+2000	SPTREMBL	1	KNOWN
+2100	Superfamily	1	KNOWNXREF
+2200	SWISSPROT	1	KNOWN
+2300	Vega_gene	1	XREF
+2310	Vega_transcript	1	XREF
+2320	Vega_translation	1	XREF
+2400	wormbase_gene	1	KNOWN
+2410	wormbase_transcript	1	KNOWN
+2420	wormpep_id	1	KNOWN
+2500	ZFIN	1	KNOWNXREF

misc-scripts/external_db/update_external_dbs.pl

+#
+# updates the external db tables on all of the core databases on a given host
+#
+
+
+use strict;
+
+use Getopt::Long;
+use DBI;
+use IO::File;
+
+my ( $host, $user, $pass, $port,@dbnames, $file, $release_num);
+
+GetOptions( "host=s", \$host,
+	    "user=s", \$user,
+	    "pass=s", \$pass,
+	    "port=i", \$port,
+	    "file=s", \$file,
+            "dbnames=s@", \@dbnames,
+	    "release_num=i", \$release_num
+	  );
+
+#both host and file are required
+usage() if(!$host || !$file);
+#release num XOR dbname are required
+usage() if(($release_num && @dbnames) || (!$release_num && !@dbnames));
+
+$port ||= 3306;
+
+my $dsn = "DBI:mysql:host=$host;port=$port";
+
+my $db = DBI->connect( $dsn, $user, $pass, {RaiseError => 1} );
+
+if($release_num) {
+  @dbnames = map {$_->[0] } @{ $db->selectall_arrayref( "show databases" ) };
+  
+  #
+  # filter out all non-core databases
+  #
+  @dbnames = grep {/^[a-zA-Z]+\_[a-zA-Z]+\_(core|est|estgene)\_${release_num}\_\d+$/} @dbnames;
+}
+
+
+#
+# make sure the user wishes to continue
+#
+print STDERR "The following databases will be external_db updated:\n  ";
+print join("\n  ", @dbnames);
+print "\ncontinue with update (yes/no)>  ";
+
+my $input = lc(<STDIN>);
+chomp($input);
+if($input ne 'yes') {
+  print "external_db conversion aborted\n";
+  exit();
+}
+
+#
+# read all of the new external_db entries from the file
+#
+my $fh = IO::File->new();
+$fh->open($file) or die("could not open input file $file");
+my @rows;
+while(<$fh>) {
+  my @a = split(/\t/);
+  push @rows, {'external_db_id' => $a[0],
+               'db_name' => $a[1],
+               'release' => $a[2],
+               'status'  => $a[3]};
+}
+$fh->close();
+
+foreach my $dbname (@dbnames) {
+  print STDERR "updating $dbname\n";
+  $db->do("use $dbname");
+  my $sth = $db->prepare('DELETE FROM external_db');
+  $sth->execute();
+  $sth->finish();
+
+  $sth = $db->prepare('INSERT INTO external_db (external_db_id, db_name,
+                                                release, status)
+                       VALUES (?,?,?,?)');
+  
+  foreach my $row (@rows) {
+    $sth->execute($row->{'external_db_id'}, 
+		  $row->{'db_name'},
+		  $row->{'release'},
+		  $row->{'status'});
+  }
+  
+  $sth->finish();
+}
+
+print STDERR "updates complete\n";
+
+
+sub usage {
+  print STDERR <<EOF
+
+             Usage: update_external_db options
+ Where options are: -host hostname 
+                    -user username 
+                    -pass password 
+                    -port port_of_server optional
+                    -release the release of the database to update used to 
+                             match database names.  e.g. 13
+                    -file the path of the file containing the insert statements
+                          of the entries of the external_db table
+                    -dbnames db1
+                          the names of the database to update. if not provided
+                          all of the core databases matching the release arg
+                          will be updated.  Either -dbnames or -release must
+                          be specified, but not both.  Multiple dbnames can
+                          be provided.
+
+ E.g.:
+
+  #update 2 databases
+  perl update_external_dbs.pl -host ecs1c -file external_dbs.txt -user ensadmin -pass secret -dbnames homo_sapiens_core_14_33 -dbnames mus_musculus_core_14_30
+
+  #update all core databases for release 14
+  perl update_external_dbs.pl -host ecs2d -file external_dbs.txt -user ensadmin -pass secret -release 14
+
+EOF
+;
+  exit;
+}