From b3616580f12c6a9c07058ff300d7ea2ef452ffa4 Mon Sep 17 00:00:00 2001
From: Monika Komorowska <mk8@sanger.ac.uk>
Date: Tue, 7 Aug 2012 09:41:15 +0000
Subject: [PATCH] updated transcript_display_xref_sources and added
gene_display_xref_sources
---
.../XrefMapper/ornithorhynchus_anatinus.pm | 61 +++++++++++--------
.../XrefMapper/rattus_norvegicus.pm | 54 ++++++++++------
2 files changed, 68 insertions(+), 47 deletions(-)
diff --git a/misc-scripts/xref_mapping/XrefMapper/ornithorhynchus_anatinus.pm b/misc-scripts/xref_mapping/XrefMapper/ornithorhynchus_anatinus.pm
index 8cc0d18cfc..dba785e7e7 100644
--- a/misc-scripts/xref_mapping/XrefMapper/ornithorhynchus_anatinus.pm
+++ b/misc-scripts/xref_mapping/XrefMapper/ornithorhynchus_anatinus.pm
@@ -6,31 +6,22 @@ use vars '@ISA';
@ISA = qw{ XrefMapper::BasicMapper };
-# Same as in BasicMapper but Genoscope order reversed.
-sub transcript_display_xref_sources {
- my @list = qw(Platypus_olfactory_receptor
- Oxford_FGU_Oa_tscript
- Oxford_FGU_Oa_gene
- RFAM
- miRBase
- IMGT/GENE_DB
- HUGO
- SGD
- MGI
- flybase_symbol
- Anopheles_symbol
- Genoscope_annotated_gene
- Uniprot/SWISSPROT
- Uniprot/Varsplic
- RefSeq_peptide
- RefSeq_dna
- Uniprot/SPTREMBL
- EntrezGene);
+sub gene_display_xref_sources {
+
+ my $self = shift;
+
+ my @list = qw(Oxford_FGU_Oa_gene
+ RFAM
+ miRBase
+ Uniprot_genename
+ EntrezGene);
my %ignore;
- $ignore{"EntrezGene"} =(<<'IEG');
+ #don't use EntrezGene labels dependent on predicted RefSeqs
+
+$ignore{'EntrezGene'} =<<IEG;
SELECT DISTINCT ox.object_xref_id
FROM object_xref ox, dependent_xref dx,
xref xmas, xref xdep,
@@ -42,15 +33,31 @@ SELECT DISTINCT ox.object_xref_id
xdep.source_id = sdep.source_id AND
smas.name like "Refseq%predicted" AND
sdep.name like "EntrezGene" AND
- ox.ox_status = "DUMP_OUT"
+ ox.ox_status = "DUMP_OUT"
IEG
- $ignore{"Uniprot/SPTREMBL"} =(<<BIGN);
+ #don't use labels starting with LOC
+
+$ignore{'LOC_prefix'} =<<LOCP;
SELECT object_xref_id
- FROM object_xref JOIN xref USING(xref_id) JOIN source USING(source_id)
- WHERE ox_status = 'DUMP_OUT' AND name = 'Uniprot/SPTREMBL'
- AND priority_description = 'protein_evidence_gt_2'
-BIGN
+ FROM object_xref JOIN xref USING(xref_id) JOIN source USING(source_id)
+ WHERE ox_status = 'DUMP_OUT' AND label REGEXP '^LOC[[:digit:]]+'
+LOCP
+
+ return [\@list,\%ignore];
+}
+
+
+sub transcript_display_xref_sources {
+ my @list = qw(Platypus_olfactory_receptor
+ Oxford_FGU_Oa_tscript
+ RFAM
+ miRBase
+ Uniprot/SWISSPROT
+ Uniprot/Varsplic
+ );
+
+ my %ignore;
return [\@list,\%ignore];
}
diff --git a/misc-scripts/xref_mapping/XrefMapper/rattus_norvegicus.pm b/misc-scripts/xref_mapping/XrefMapper/rattus_norvegicus.pm
index 68e0e32da5..0ab3ade136 100644
--- a/misc-scripts/xref_mapping/XrefMapper/rattus_norvegicus.pm
+++ b/misc-scripts/xref_mapping/XrefMapper/rattus_norvegicus.pm
@@ -7,25 +7,20 @@ use vars '@ISA';
@ISA = qw{ XrefMapper::BasicMapper };
-sub transcript_display_xref_sources {
+sub gene_display_xref_sources {
my $self = shift;
-
+
my @list = qw(RFAM
- miRBase
- RGD
- MGI
- flybase_symbol
- Anopheles_symbol
- Genoscope_annotated_gene
- Uniprot/SWISSPROT
- Uniprot/Varsplic
- Uniprot/SPTREMBL
- EntrezGene);
-
+ miRBase
+ MGI
+ Uniprot_genename
+ EntrezGene);
my %ignore;
- $ignore{"EntrezGene"} =(<<'IEG');
+ #don't use EntrezGene labels dependent on predicted RefSeqs
+
+$ignore{'EntrezGene'} =<<IEG;
SELECT DISTINCT ox.object_xref_id
FROM object_xref ox, dependent_xref dx,
xref xmas, xref xdep,
@@ -37,15 +32,34 @@ SELECT DISTINCT ox.object_xref_id
xdep.source_id = sdep.source_id AND
smas.name like "Refseq%predicted" AND
sdep.name like "EntrezGene" AND
- ox.ox_status = "DUMP_OUT"
+ ox.ox_status = "DUMP_OUT"
IEG
- $ignore{"Uniprot/SPTREMBL"} =(<<BIGN);
+ #don't use labels starting with LOC
+
+$ignore{'LOC_prefix'} =<<LOCP;
SELECT object_xref_id
- FROM object_xref JOIN xref USING(xref_id) JOIN source USING(source_id)
- WHERE ox_status = 'DUMP_OUT' AND name = 'Uniprot/SPTREMBL'
- AND priority_description = 'protein_evidence_gt_2'
-BIGN
+ FROM object_xref JOIN xref USING(xref_id) JOIN source USING(source_id)
+ WHERE ox_status = 'DUMP_OUT' AND label REGEXP '^LOC[[:digit:]]+'
+LOCP
+
+ return [\@list,\%ignore];
+
+}
+
+
+sub transcript_display_xref_sources {
+ my $self = shift;
+
+ my @list = qw(RFAM
+ miRBase
+ RGD
+ Uniprot/SWISSPROT
+ Uniprot/Varsplic
+);
+
+
+ my %ignore;
return [\@list,\%ignore];
--
GitLab