- Timestamp:
- 05/15/10 21:10:07 (2 years ago)
- Files:
-
- 1 modified
Legend:
- Unmodified
- Added
- Removed
-
trunk/dbbuilder/src/org/bridgedb/extract/Ensembl_ETL_Device_v18_local.pl
r352 r353 41 41 use HashSpeciesList; 42 42 use lib '/home/socr/c/users2/apico/src/ensembl/modules'; 43 use lib '/home/socr/c/users2/apico/src/ensembl-compara/modules';44 use lib '/home/socr/c/users2/apico/src/ensembl-variation/modules';43 #use lib '/home/socr/c/users2/apico/src/ensembl-compara/modules'; 44 #use lib '/home/socr/c/users2/apico/src/ensembl-variation/modules'; 45 45 use lib '/home/socr/c/users2/apico/src/ensembl-functgenomics/modules'; 46 46 use lib '/home/socr/c/users2/apico/bioperl-live'; 47 47 use Bio::EnsEMBL::Registry; 48 48 use Bio::EnsEMBL::DBSQL::DBConnection; 49 use Bio::EnsEMBL::Compara::DBSQL::DBAdaptor;49 #use Bio::EnsEMBL::Compara::DBSQL::DBAdaptor; 50 50 use Bio::EnsEMBL::DBSQL::GOTermAdaptor; 51 51 use Bio::EnsEMBL::OntologyTerm; … … 390 390 391 391 $registry->load_registry_from_db( 392 -host => 'localhost', 392 -host => 'mysql-dev.cgl.ucsf.edu', 393 -port => '13308', 393 394 -user => 'genmapp', 394 395 -pass => 'fun4genmapp', … … 421 422 my $dbname = $dbas[0]->dbc->dbname(); # e.g., core_mus_musculus_42_36c 422 423 my @split_dbname = split(/_/, $dbname); 423 if ($split_dbname[ 1] == "collection"){ # shift array elements for "collections"424 if ($split_dbname[2] == "collection"){ # shift array elements for "collections" 424 425 splice(@split_dbname,1,1); 425 426 } 426 my $build = $split_dbname[ 3]; # e.g., 42427 my $build_nums = $build.$split_dbname[ 4]; # e.g., 42_36c427 my $build = $split_dbname[4]; # e.g., 42 428 my $build_nums = $build.$split_dbname[5]; # e.g., 42_36c 428 429 429 430 ############################################################################ … … 1174 1175 my $sqlUser = "genmapp"; # username 1175 1176 my $sqlPass = "fun4genmapp"; # password 1176 my $sqlDBLog = "ensembl_etl_device_log"; # permanent log table 1177 my $sqlDB = "current_".$genus_species; # name of species database, based on user-selected species 1178 my $sqlCSDB = "current_CS_".$genus_species; # name of species database, based on user-selected species 1179 my $sqlConnection ="dbi:mysql:$sqlDBLog"; # connection statement used in DBI->connect() 1177 my $sqlDBLog = "genmapp_ensembl_etl_device_log"; # permanent log table 1178 my $sqlHost = "mysql-dev.cgl.ucsf.edu"; # plato host 1179 my $sqlPort = "13308"; # plato port 1180 my $sqlDB = "genmapp_current_".$genus_species; # name of species database, based on user-selected species 1181 my $sqlCSDB = "genmapp_current_CS_".$genus_species; # name of species database, based on user-selected species 1182 my $sqlConnection ="dbi:mysql:$sqlDBLog:$sqlHost:$sqlPort"; # connection statement used in DBI->connect() 1180 1183 1181 1184 $dbh = DBI->connect($sqlConnection, $sqlUser, $sqlPass) … … 1445 1448 $dateArg = $year.$mon.$mday; 1446 1449 } 1447 my $current_std = " current_".$genus_species;1448 my $new_std = $twoLetterSpecies."_Std_".$dateArg;1449 my $current_cs = " current_CS_".$genus_species;1450 my $new_cs = $twoLetterSpecies."_CS_".$dateArg;1450 my $current_std = "genmapp_current_".$genus_species; 1451 my $new_std = "genmapp_".$twoLetterSpecies."_Std_".$dateArg; 1452 my $current_cs = "genmapp_current_CS_".$genus_species; 1453 my $new_cs = "genmapp_".$twoLetterSpecies."_CS_".$dateArg; 1451 1454 my $sqlPass = "fun4genmapp"; 1452 1455 my $bkupDir = "/home/socr/c/users2/apico/EnsemblAPI/DB_Copies/"; 1453 my $mysqldump_cmd = "mysqldump - u genmapp --password=".$sqlPass;1454 my $mysql_cmd = "mysql - u genmapp --password=".$sqlPass;1456 my $mysqldump_cmd = "mysqldump --host=mysql-dev.cgl.ucsf.edu --port=13308 -u genmapp --password=".$sqlPass; 1457 my $mysql_cmd = "mysql --host=mysql-dev.cgl.ucsf.edu --port=13308 -u genmapp --password=".$sqlPass; 1455 1458 1456 1459 ## BACKUP and COPY STD TABLES? … … 1703 1706 if (!${$seen{GeneOntology}{$dbe_primary_id}}++){ 1704 1707 # Get GO term annotations using $go_adaptor 1705 my $acc = $dbe_primary_id; 1706 $acc =~ s/\'//g; # strip single quotes to use as variable 1707 my $term = $go_adaptor->fetch_by_accession($acc); 1708 my $name = mysql_quotes($term->name()); #e.g., plasma membrane 1709 my $namespace = mysql_quotes($term->namespace()); # e.g., cellular component 1708 my $name = ""; 1709 my $namespace = ""; 1710 if ($go_adaptor){ 1711 my $acc = $dbe_primary_id; 1712 $acc =~ s/\'//g; # strip single quotes to use as variable 1713 my $term = $go_adaptor->fetch_by_accession($acc); 1714 my $name = mysql_quotes($term->name()); #e.g., plasma membrane 1715 my $namespace = mysql_quotes($term->namespace()); # e.g., cellular component 1716 } 1710 1717 $$GeneTables{GeneOntology}{$count.$dot.$subcount{GeneOntology}} = [$dbe_primary_id, $name, $namespace]; 1711 1718 $$Ensembl_GeneTables{GeneOntology}{$count.$dot.$subcount{GeneOntology}} = [$gene_stable_id, $dbe_primary_id]; … … 1715 1722 elsif ($dbe_dbname =~ /^\'goslim/){ 1716 1723 $ADMIN_Xrefs{$dbe_dbname}[10] = "\'Y\'"; # collected 1717 # Get GO term annotations using $go_adaptor1718 1724 my $acc = $dbe_primary_id; 1719 1725 $acc =~ s/\'//g; # strip single quotes to use as variable 1720 my $term = $go_adaptor->fetch_by_accession($acc); 1721 my $name = mysql_quotes($term->name()); #e.g., plasma membrane 1722 my $namespace = mysql_quotes($term->namespace()); # e.g., cellular component 1723 if ($namespace =~ /\'biological_process\'/){ 1726 # Get GO term annotations using $go_adaptor 1727 if ($go_adaptor){ 1728 my $term = $go_adaptor->fetch_by_accession($acc); 1729 my $name = mysql_quotes($term->name()); #e.g., plasma membrane 1730 my $namespace = mysql_quotes($term->namespace()); # e.g., cellular component 1731 if ($namespace =~ /\'biological_process\'/){ 1724 1732 $$GeneTables{GOslimBP}{$count.$dot.$subcount{GOslimBP}} = [$dbe_primary_id, $name]; 1725 1733 $$Ensembl_GeneTables{GOslimBP}{$count.$dot.$subcount{GOslimBP}} = [$gene_stable_id, $dbe_primary_id]; 1726 1734 ++$subcount{GOslimBP}; 1727 } elsif ($namespace =~ /\'cellular_component\'/){1735 } elsif ($namespace =~ /\'cellular_component\'/){ 1728 1736 $$GeneTables{GOslimCC}{$count.$dot.$subcount{GOslimCC}} = [$dbe_primary_id, $name]; 1729 1737 $$Ensembl_GeneTables{GOslimCC}{$count.$dot.$subcount{GOslimCC}} = [$gene_stable_id, $dbe_primary_id]; 1730 1738 ++$subcount{GOslimCC}; 1731 } elsif ($namespace =~ /\'molecular_function\'/){1739 } elsif ($namespace =~ /\'molecular_function\'/){ 1732 1740 $$GeneTables{GOslimMF}{$count.$dot.$subcount{GOslimMF}} = [$dbe_primary_id, $name]; 1733 1741 $$Ensembl_GeneTables{GOslimMF}{$count.$dot.$subcount{GOslimMF}} = [$gene_stable_id, $dbe_primary_id]; 1734 1742 ++$subcount{GOslimMF}; 1743 } else { 1744 #garbage? 1745 } 1735 1746 } else { 1736 # garbage?1747 # no go 1737 1748 } 1738 1749 } … … 2092 2103 ## Solution #2 only works after API method has been added 2093 2104 ## See NathJohnson_API_patch.txt 2105 print $gene_stable_id; 2106 print $gene; 2094 2107 my $probe_features = $probe_adaptor->fetch_all_by_linked_transcript_Gene($gene); 2095 2108
