From 95d7d298ae49c96555289e33caba44e95a77b8f9 Mon Sep 17 00:00:00 2001
From: Kevin Howe <klh@ebi.ac.uk>
Date: Tue, 16 Dec 2014 15:08:50 +0000
Subject: [PATCH] Added UniProt MULTI-invertebrate sources, and cleaned up
 WormBase sources

---
 misc-scripts/xref_mapping/xref_config.ini | 229 +++++++++++++++++++---
 1 file changed, 203 insertions(+), 26 deletions(-)

diff --git a/misc-scripts/xref_mapping/xref_config.ini b/misc-scripts/xref_mapping/xref_config.ini
index 504e9696c1..7f2f854f27 100644
--- a/misc-scripts/xref_mapping/xref_config.ini
+++ b/misc-scripts/xref_mapping/xref_config.ini
@@ -3719,6 +3719,17 @@ prio_descr      =
 parser          = UniProtAltParser
 release_uri     =
 
+[source Uniprot/SPTREMBL::MULTI-invertebrate]
+name            = Uniprot/SPTREMBL
+download        = Y
+order           = 20
+priority        = 1
+prio_descr      =
+parser          = UniProtAltParser
+dependent_on    = MIM
+release_uri     = ftp://ftp.ebi.ac.uk/pub/databases/uniprot/current_release/knowledgebase/taxonomic_divisions/reldate.txt
+data_uri        = ftp://ftp.ebi.ac.uk/pub/databases/uniprot/current_release/knowledgebase/taxonomic_divisions/uniprot_trembl_invertebrates.dat.gz
+
 [source Uniprot/SPTREMBL::MULTI]
 # Used by aedes_aegypti, anolis_carolinensis, anopheles_gambiae, acyrthosiphon_pisum, apis_mellifera, bos_taurus, caenorhabditis_elegans, canis_familiaris, cavia_porcellus, ciona_intestinalis, ciona_savignyi, danio_rerio, dasypus_novemcinctus, echinops_telfairi, erinaceus_europaeus, equus_caballus, felis_catus, ficedula_albicollis, gallus_gallus, gasterosteus_aculeatus, homo_sapiens, ixodes_scapularis, loxodonta_africana, macaca_mulatta, monodelphis_domestica, mus_musculus, myotis_lucifugus, ochotona_princeps, ornithorhynchus_anatinus, oryctolagus_cuniculus, oryzias_latipes, pan_troglodytes,pongo_abelii, rattus_norvegicus, saccharomyces_cerevisiae, ictidomys_tridecemlineatus, sus_scrofa, taeniopygia_guttata, takifugu_rubripes, tupaia_belangeri, xenopus_tropicalis, phaeodactylum_tricornutum , thalassiosira_pseudonana, lepisosteus_oculatus
 name            = Uniprot/SPTREMBL
@@ -3767,6 +3778,18 @@ release_uri     = ftp://ftp.ebi.ac.uk/pub/databases/uniprot/knowledgebase/reldat
 data_uri        = ftp://ftp.ebi.ac.uk/pub/databases/uniprot/knowledgebase/uniprot_sprot.dat.gz
 
 
+[source Uniprot/SWISSPROT::MULTI-invertebrate]
+name            = Uniprot/SWISSPROT
+download        = Y
+order           = 20
+priority        = 2
+prio_descr      = sequence_mapped
+parser          = UniProtAltParser
+dependent_on    = MIM
+release_uri     = ftp://ftp.ebi.ac.uk/pub/databases/uniprot/current_release/knowledgebase/taxonomic_divisions/reldate.txt
+data_uri        = ftp://ftp.ebi.ac.uk/pub/databases/uniprot/current_release/knowledgebase/taxonomic_divisions/uniprot_sprot_invertebrates.dat.gz
+
+
 #
 # Direct Uniprot mapping.
 # Note these have to be done LAST so that we get all
@@ -4560,7 +4583,7 @@ parser          =
 release_uri     =
 data_uri        =
 
-[source wormbase::caenorhabditis_elegans]
+[source wormbase::celegans]
 name            = wormbase_all
 download        = Y
 order           = 50
@@ -4568,10 +4591,80 @@ priority        = 1
 prio_descr      =
 parser          = WormbaseDirectParser
 release_uri     =
-data_uri        = ftp://ftp.sanger.ac.uk/pub2/wormbase/releases/WS230/species/c_elegans/c_elegans.WS230.xrefs.txt.gz
+data_uri        = ftp://ftp.ebi.ac.uk/pub/databases/wormbase/releases/WS245/species/c_elegans/PRJNA13758/c_elegans.PRJNA13758.WS245.xrefs.txt.gz
 
-[source wormpep_id::caenorhabditis_elegans]
-# Used by caenorhabditis_elegans
+[source wormbase::cbriggsae]
+name            = wormbase_all
+download        = Y
+order           = 50
+priority        = 1
+prio_descr      =
+parser          = WormbaseDirectParser
+release_uri     =
+data_uri        = ftp://ftp.ebi.ac.uk/pub/databases/wormbase/releases/WS245/species/c_briggsae/PRJNA10731/c_briggsae.PRJNA10731.WS245.xrefs.txt.gz
+
+[source wormbase::cbrenneri]
+name            = wormbase_all
+download        = Y
+order           = 50
+priority        = 1
+prio_descr      =
+parser          = WormbaseDirectParser
+release_uri     =
+data_uri        = ftp://ftp.ebi.ac.uk/pub/databases/wormbase/releases/WS245/species/c_brenneri/PRJNA20035/c_brenneri.PRJNA20035.WS245.xrefs.txt.gz
+
+[source wormbase::cremanei]
+name            = wormbase_all
+download        = Y
+order           = 50
+priority        = 1
+prio_descr      =
+parser          = WormbaseDirectParser
+release_uri     =
+data_uri        = ftp://ftp.ebi.ac.uk/pub/databases/wormbase/releases/WS245/species/c_remanei/PRJNA53967/c_remanei.PRJNA53967.WS245.xrefs.txt.gz
+
+[source wormbase::cjaponica]
+name            = wormbase_all
+download        = Y
+order           = 50
+priority        = 1
+prio_descr      =
+parser          = WormbaseDirectParser
+release_uri     =
+data_uri        = ftp://ftp.ebi.ac.uk/pub/databases/wormbase/releases/WS245/species/c_japonica/PRJNA12591/c_japonica.PRJNA12591.WS245.xrefs.txt.gz
+
+[source wormbase::ppacificus]
+name            = wormbase_all
+download        = Y
+order           = 50
+priority        = 1
+prio_descr      =
+parser          = WormbaseDirectParser
+release_uri     =
+data_uri        = ftp://ftp.ebi.ac.uk/pub/databases/wormbase/releases/WS245/species/p_pacificus/PRJNA12644/p_pacificus.PRJNA12644.WS245.xrefs.txt.gz
+
+[source wormbase::bmalayi]
+name            = wormbase_all
+download        = Y
+order           = 50
+priority        = 1
+prio_descr      =
+parser          = WormbaseDirectParser
+release_uri     =
+data_uri        = ftp://ftp.ebi.ac.uk/pub/databases/wormbase/releases/WS245/species/b_malayi/PRJNA10729/b_malayi.PRJNA10729.WS245.xrefs.txt.gz
+
+[source wormbase::ovolvulus]
+name            = wormbase_all
+download        = Y
+order           = 50
+priority        = 1
+prio_descr      =
+parser          = WormbaseDirectParser
+release_uri     =
+data_uri        = ftp://ftp.ebi.ac.uk/pub/databases/wormbase/releases/WS245/species/o_volvulus/PRJEB513/o_volvulus.PRJEB513.WS245.xrefs.txt.gz
+
+[source wormpep_id::wormbase]
+# Used by wormbase core species
 name            = wormpep_id
 download        = N
 order           = 50
@@ -4581,8 +4674,8 @@ parser          = comes from WormbaseDirectParser
 release_uri     =
 data_uri        =
 
-[source wormbase_gene::caenorhabditis_elegans]
-# Used by caenorhabditis_elegans comes from WormbaseDatabaseStableIDParser
+[source wormbase_gene::wormbase]
+# Used by wormbase core species
 name            = wormbase_gene
 download        = N
 order           = 50
@@ -4592,8 +4685,8 @@ parser          = comes from WormbaseDirectParser
 release_uri     =
 data_uri        =
 
-[source wormbase_locus::caenorhabditis_elegans]
-# Used by caenorhabditis_elegans comes from WormbaseDatabaseStableIDParser
+[source wormbase_locus::wormbase]
+# Used by wormbase core species
 name            = wormbase_locus
 download        = N
 order           = 50
@@ -4604,7 +4697,7 @@ release_uri     =
 data_uri        =
 
 [source wormbase_gseqname::wormbase]
-# Used by caenorhabditis_elegans comes from WormbaseDirectParser
+# Used by wormbase core species
 name            = wormbase_gseqname
 download        = N
 order           = 50
@@ -4614,8 +4707,8 @@ parser          = comes from WormbaseDirectParser
 release_uri     =
 data_uri        =
 
-[source wormbase_transcript::caenorhabditis_elegans]
-# Used by caenorhabditis_elegans comes from WormbaseDatabaseStableIDParser
+[source wormbase_transcript::wormbase]
+# Used by wormbase core species
 name            = wormbase_transcript
 download        = N
 order           = 50
@@ -5519,21 +5612,6 @@ source          = RFAM::MULTI
 source		= miRBase::MULTI
 source 		= ArrayExpress::MULTI
 
-[species caenorhabditis_elegans]
-taxonomy_id     = 6239
-aliases         = caenorhabditiselegans, celegans, elegans
-source          = EntrezGene::MULTI
-source          = GO::caenorhabditis_elegans
-source          = goslim_goa::EG
-source          = RefSeq_dna::caenorhabditis_elegans
-source          = RefSeq_peptide::caenorhabditis_elegans
-source          = UniGene::caenorhabditis_elegans
-source          = Uniprot/SPTREMBL::MULTI
-source          = Uniprot/SWISSPROT::MULTI
-source          = UniParc::MULTI
-source          = wormbase::caenorhabditis_elegans
-source          = ArrayExpress::EG
-
 [species callithrix_jacchus]
 taxonomy_id     = 9483
 aliases         = Marmoset, marmoset, white-tufted-ear marmoset, callithrixjacchus, callithrix_jacchus
@@ -8447,3 +8525,102 @@ source          = RefSeq_peptide::MULTI-invertebrate
 source          = Uniprot/SPTREMBL::MULTI
 source          = Uniprot/SWISSPROT::MULTI
 source          = UniParc::MULTI
+
+[species caenorhabditis_elegans]
+taxonomy_id     = 6239
+aliases         = caenorhabditiselegans, celegans, elegans
+source          = EntrezGene::MULTI
+source          = GO::caenorhabditis_elegans
+source          = goslim_goa::EG
+source          = RefSeq_dna::caenorhabditis_elegans
+source          = RefSeq_peptide::caenorhabditis_elegans
+source          = UniGene::caenorhabditis_elegans
+source          = Uniprot/SPTREMBL::MULTI-invertebrate
+source          = Uniprot/SWISSPROT::MULTI-invertebrate
+source          = UniParc::MULTI
+source          = wormbase::celegans
+source          = ArrayExpress::EG
+
+[species caenorhabditis_briggsae]
+taxonomy_id     = 6238
+aliases         = caenorhabditis_briggae_prjna10731
+source          = EntrezGene::MULTI
+source          = GO::MULTI
+source          = RefSeq_dna::MULTI-invertebrate
+source          = RefSeq_peptide::MULTI-invertebrate
+source          = Uniprot/SPTREMBL::MULTI-invertebrate
+source          = Uniprot/SWISSPROT::MULTI-invertebrate
+source          = UniParc::MULTI
+source          = wormbase::cbriggsae
+
+[species caenorhabditis_remanei]
+taxonomy_id     = 31234
+aliases         = caenorhabditis_remanei_prjna53967
+source          = EntrezGene::MULTI
+source          = GO::MULTI
+source          = RefSeq_dna::MULTI-invertebrate
+source          = RefSeq_peptide::MULTI-invertebrate
+source          = Uniprot/SPTREMBL::MULTI-invertebrate
+source          = Uniprot/SWISSPROT::MULTI-invertebrate
+source          = UniParc::MULTI
+source          = wormbase::cremanei
+
+[species caenorhabditis_brenneri]
+taxonomy_id     = 135651
+aliases         = caenorhabditis_brenneri_prjna20035
+source          = EntrezGene::MULTI
+source          = GO::MULTI
+source          = RefSeq_dna::MULTI-invertebrate
+source          = RefSeq_peptide::MULTI-invertebrate
+source          = Uniprot/SPTREMBL::MULTI-invertebrate
+source          = Uniprot/SWISSPROT::MULTI-invertebrate
+source          = UniParc::MULTI
+source          = wormbase::cbrenneri
+
+[species caenorhabditis_japonica]
+taxonomy_id     = 281687
+aliases         = caenorhabditis_japonica_prjna12591
+source          = EntrezGene::MULTI
+source          = GO::MULTI
+source          = RefSeq_dna::MULTI-invertebrate
+source          = RefSeq_peptide::MULTI-invertebrate
+source          = Uniprot/SPTREMBL::MULTI-invertebrate
+source          = Uniprot/SWISSPROT::MULTI-invertebrate
+source          = UniParc::MULTI
+source          = wormbase::cjaponica
+
+[species brugia_malayi]
+taxonomy_id     = 6279
+aliases         = brugia_malayi_prjna10729
+source          = EntrezGene::MULTI
+source          = GO::MULTI
+source          = RefSeq_dna::MULTI-invertebrate
+source          = RefSeq_peptide::MULTI-invertebrate
+source          = Uniprot/SPTREMBL::MULTI-invertebrate
+source          = Uniprot/SWISSPROT::MULTI-invertebrate
+source          = UniParc::MULTI
+source          = wormbase::bmalayi
+
+[species onchocerca_volvulus]
+taxonomy_id     = 6282
+aliases         = onchocerca_volvulus_prjeb513
+source          = EntrezGene::MULTI
+source          = GO::MULTI
+source          = RefSeq_dna::MULTI-invertebrate
+source          = RefSeq_peptide::MULTI-invertebrate
+source          = Uniprot/SPTREMBL::MULTI-invertebrate
+source          = Uniprot/SWISSPROT::MULTI-invertebrate
+source          = UniParc::MULTI
+source          = wormbase::ovolvulus
+
+[species pristionchus_pacificus]
+taxonomy_id     = 54126
+aliases         = pristionchus_pacificus_prjna12644
+source          = EntrezGene::MULTI
+source          = GO::MULTI
+source          = RefSeq_dna::MULTI-invertebrate
+source          = RefSeq_peptide::MULTI-invertebrate
+source          = Uniprot/SPTREMBL::MULTI-invertebrate
+source          = Uniprot/SWISSPROT::MULTI-invertebrate
+source          = UniParc::MULTI
+source          = wormbase::ppacificus
-- 
GitLab