Moved here from parent dir

misc-scripts/xref_mapping/sql/populate_metadata.sql

+# Populate the appropriate tables in an xref metadata database
+
+
+################################################################################
+# SPECIES
+
+INSERT INTO species (taxonomy_id, name, aliases) VALUES (9606,  'homo_sapiens',            'human,hsapiens,homosapiens');
+INSERT INTO species (taxonomy_id, name, aliases) VALUES (10090, 'mus_musculus',            'mouse,mmusculus,musmusculus');
+INSERT INTO species (taxonomy_id, name, aliases) VALUES (10116, 'rattus_norvegicus',       'rat,rnovegicus,rattusnorvegicus');
+INSERT INTO species (taxonomy_id, name, aliases) VALUES (31033, 'fugu_rubripes',           'pufferfish,fugu,frubripes,fugurubripes');
+INSERT INTO species (taxonomy_id, name, aliases) VALUES (7165,  'anopheles_gambiae',       'mosquito,anopheles,agambiae,anophelesgambiae');
+INSERT INTO species (taxonomy_id, name, aliases) VALUES (7227,  'drosophila_melanogaster', 'drosophila,dmelongaster,drosophilamelanogaster' );
+INSERT INTO species (taxonomy_id, name, aliases) VALUES (6239,  'caenorhabditis_elegans',  'elegans,celegans,caenorhabditiselegans');
+INSERT INTO species (taxonomy_id, name, aliases) VALUES (6238,  'caenorhabditis_briggsae', 'briggsae,cbriggsae,caenorhabditisbriggsae');
+INSERT INTO species (taxonomy_id, name, aliases) VALUES (7955,  'danio_rerio',             'zebrafish,danio,drerio,daniorerio' );
+INSERT INTO species (taxonomy_id, name, aliases) VALUES (9598,  'pan_troglodytes',         'chimp,chimpanzee,ptroglodytes,pantroglodytes');
+INSERT INTO species (taxonomy_id, name, aliases) VALUES (9031,  'gallus_gallus',           'chicken,chick,ggallus,gallusgallus' );
+INSERT INTO species (taxonomy_id, name, aliases) VALUES (99883, 'tetraodon_nigroviridis',  'tetraodon,tnigroviridis,tetraodonnigroviridis');
+INSERT INTO species (taxonomy_id, name, aliases) VALUES (9913,   'bos_taurus',             'cow,btaurus,bostaurus');
+INSERT INTO species (taxonomy_id, name, aliases) VALUES (9615,   'canis_familaris',        'dog,doggy,cfamiliaris,canisfamiliaris');
+
+################################################################################
+# SOURCES - types of data we can read
+
+# "High level" sources that we will also download from (via source_url)
+
+INSERT INTO source VALUES (1, "UniProtSwissProt", 1, 'Y',1);
+INSERT INTO source VALUES (2, "UniProtSPTrEMBL", 1, 'Y',1);
+INSERT INTO source VALUES (3, "RefSeq", 1, 'Y',1);
+
+# Other sources - used to create dependent xrefs, but not to upload from
+
+INSERT INTO source VALUES (1010, 'EMBL', 1, 'N', 2);
+INSERT INTO source VALUES (1020, 'MIM', 1, 'N', 2);
+INSERT INTO source VALUES (1030, 'PDB', 1, 'N', 2);
+INSERT INTO source VALUES (1040, 'protein_id', 1, 'N', 2);
+INSERT INTO source VALUES (1050, 'PUBMED', 1, 'N', 2);
+INSERT INTO source VALUES (1060, 'MEDLINE', 1, 'N', 2);
+INSERT INTO source VALUES (1100, 'LocusLink', 1, 'N', 2);
+
+INSERT INTO source VALUES (1070, 'GO', 1, 'Y',2);
+INSERT INTO source VALUES (1080, 'MarkerSymbol', 1, 'Y',2);
+INSERT INTO source VALUES (1090, 'HUGO', 1, 'Y',2);
+
+
+################################################################################
+# Files to fetch data from
+
+# --------------------------------------------------------------------------------
+# UniProt (SwissProt & SPTrEMBL)
+
+# Note currently no UniProt data for fugu, anopheles, c.briggsae or chicken.
+
+
+###HUMAN
+##       uniprot
+INSERT INTO source_url (source_id, species_id, url, checksum, file_modified_date, upload_date, parser) VALUES (1, 1,'ftp://ftp.ebi.ac.uk/pub/databases/SPproteomes/swissprot_files/proteomes/9606.SPC', '', now(), now(), "UniProtParser");
+
+##       refseq
+INSERT INTO source_url (source_id, species_id, url, checksum, file_modified_date, upload_date, parser) VALUES (3, 1,'ftp://ftp.ncbi.nih.gov/refseq/H_sapiens/mRNA_Prot/human.protein.gpff.gz', '', now(), now(), "RefSeqGPFFParser");
+
+##       refseq
+INSERT INTO source_url (source_id, species_id, url, checksum, file_modified_date, upload_date, parser) VALUES (3, 1,'ftp://ftp.ncbi.nih.gov/refseq/H_sapiens/mRNA_Prot/human.rna.fna.gz', '', now(), now(), "RefSeqParser");
+
+##       GO
+INSERT INTO source_url (source_id, species_id, url, checksum, file_modified_date, upload_date, parser) VALUES (1070, 1,'ftp://ftp.ebi.ac.uk/pub/databases/GO/goa/HUMAN/gene_association.goa_human.gz', '', now(), now(), "GOParser");
+
+##       HUGO
+INSERT INTO source_url (source_id, species_id, url, checksum, file_modified_date, upload_date, parser) VALUES (1090, 1,'http://www.gene.ucl.ac.uk/public-files/nomen/ens4.txt http://www.gene.ucl.ac.uk/public-files/nomen/ens1.txt', '', now(), now(), "HUGOParser");
+
+
+
+
+
+###MOUSE
+##      uniprot
+INSERT INTO source_url (source_id, species_id, url, checksum, file_modified_date, upload_date, parser) VALUES (1, 2, 'ftp://ftp.ebi.ac.uk/pub/databases/SPproteomes/swissprot_files/proteomes/10090.SPC', '', now(), now(), "UniProtParser");
+
+##      refseq
+INSERT INTO source_url (source_id, species_id, url, checksum, file_modified_date, upload_date, parser) VALUES (3, 2,'ftp://ftp.ncbi.nih.gov/refseq/M_musculus/mRNA_Prot/mouse.protein.gpff.gz', '', now(), now(), "RefSeqGPFFParser");
+
+##      refseq
+INSERT INTO source_url (source_id, species_id, url, checksum, file_modified_date, upload_date, parser) VALUES (3, 2,'ftp://ftp.ncbi.nih.gov/refseq/M_musculus/mRNA_Prot/mouse.rna.fna.gz', '', now(), now(), "RefSeqParser");
+
+##      mgd (MGI -- MarkerSymbol)
+INSERT INTO source_url (source_id, species_id, url, checksum, file_modified_date, upload_date, parser) VALUES (1080, 2,'ftp://ftp.informatics.jax.org/pub/reports/MRK_SwissProt_TrEMBL.rpt', '', now(), now(), "MGDParser");
+
+##      GO 
+INSERT INTO source_url (source_id, species_id, url, checksum, file_modified_date, upload_date, parser) VALUES (1070, 2,'ftp://ftp.ebi.ac.uk/pub/databases/GO/goa/MOUSE/gene_association.goa_mouse.gz', '', now(), now(), "GOParser");
+
+################################################################################
+

misc-scripts/xref_mapping/sql/table.sql

+# Schema for internal-external database mappings (xrefs)
+
+
+################################################################################
+#
+# General external annotation.
+
+CREATE TABLE xref (
+
+  xref_id                     int unsigned not null auto_increment,
+  accession                   varchar(255) not null,
+  label                       varchar(255),
+  description                 varchar(255),
+  source_id                   int unsigned not null,
+  species_id                  int unsigned not null,
+
+  PRIMARY KEY (xref_id),
+  UNIQUE acession_idx(accession,source_id)
+
+);
+
+################################################################################
+
+CREATE TABLE primary_xref (
+
+  xref_id                     int unsigned not null,
+  sequence                    mediumtext,
+  sequence_type               enum('dna','peptide'),
+  status                      enum('experimental','predicted'),
+  source_id                   int unsigned not null,
+
+  PRIMARY KEY (xref_id)
+
+);
+
+################################################################################
+
+CREATE TABLE dependent_xref (
+
+  master_xref_id              int unsigned not null,
+  dependent_xref_id           int unsigned not null,
+  linkage_annotation          varchar(255),
+  source_id                   int unsigned not null,
+
+  KEY master_idx(master_xref_id),
+  KEY dependent_idx(dependent_xref_id)
+
+);
+
+################################################################################
+
+CREATE TABLE synonym (
+
+  xref_id                     int unsigned not null,
+  synonym_xref_id             int unsigned not null,
+  source_id                   int unsigned not null,
+
+  KEY xref_idx(xref_id)
+
+);
+
+################################################################################
+
+CREATE TABLE source (
+
+  source_id                   int unsigned not null auto_increment,
+  name                        varchar(255) not null,
+  release                     varchar(255),
+  download                    enum('Y', 'N') default 'Y',
+  ordered                     int unsigned not null, 
+   
+  PRIMARY KEY (source_id),
+  KEY name_idx(name) 
+
+);
+
+################################################################################
+
+CREATE TABLE source_url (
+
+  source_url_id               int unsigned not null auto_increment,
+  source_id                   int unsigned not null,
+  species_id                  int unsigned not null,
+  url                         varchar(255),
+  checksum                    varchar(255),
+  file_modified_date          datetime,
+  upload_date                 datetime,
+  parser                      varchar(255),
+
+  PRIMARY KEY (source_url_id),
+  KEY source_idx(source_id)
+
+);
+
+################################################################################
+
+CREATE TABLE direct_xref (
+
+  general_xref_id             int unsigned not null,
+  ensembl_stable_id           varchar(255),
+  type                        enum('gene','transcript','translation'),
+  linkage_xref                varchar(255),
+
+  KEY primary_idx(general_xref_id),
+  KEY ensembl_idx(ensembl_stable_id)
+
+);
+
+################################################################################
+
+CREATE TABLE species (
+
+  species_id                  int unsigned not null auto_increment,
+  taxonomy_id                 int unsigned not null,
+  name                        varchar(255) not null,
+  aliases                     varchar(255),
+
+  PRIMARY KEY(species_id),
+  KEY taxonomy_idx(taxonomy_id),
+  KEY name_idx(name)
+
+);
+
+################################################################################