rbbt-sources 3.0.8 → 3.0.9

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 89ddc36cd6b2f23142423880dcd72b8203fcee94
-  data.tar.gz: e8726fccf6a3bd5f332557fa302f551e55ae76dd
+  metadata.gz: 7da001935348263d932ee50674a7988587cac6a9
+  data.tar.gz: cd1fd8a320c3d8ada473b0faf6ca6d8960f7646d
 SHA512:
-  metadata.gz: bf0802f382a8a120d14a603846f42dbcef825c18cccdb9511150d7d3eac4e2b6ab79954438d9b40ba88173e70e74210245ae50bf6abaa234593d1cee80773323
-  data.tar.gz: 77877befdf71b2c4d929ab5ab543a54c012cffbbcf2e1f8a6deec127515986cda06415bb7c44a94a73aaceab241cf81d492e04d47c08622040dcbca15264178f
+  metadata.gz: 2d43e22521331578c7d0e078c4b666e18b8af4666de5e389b4c99d98a80f924b59257b0838f2785f04aa6d4e92ce10d2f29faac22dcd8dceb8dc33d9619bc2fb
+  data.tar.gz: 779b4e1866adb97fe2b4ee1f6b61728087aa8a5212ed789b89d1bf9cd6a57970c74b499c01b3aa8993613a123189833ad37a1411f78f04df1957340534cd2bcb

data/lib/rbbt/sources/biomart.rb

@@ -35,15 +35,15 @@ module BioMart
     if defined? Rbbt and Rbbt.etc.allowed_biomart_archives.exists?
       raise "Biomart archive #{ date } is not allowed in this installation" unless Rbbt.etc.allowed_biomart_archives.read.split("\n").include? date
     end
-    @archive = date
-    @archive_url = BIOMART_URL.sub(/http:\/\/biomart\./, 'http://' + date + '.archive.ensembl.')
-    Log.debug "Using Archive URL #{ @archive_url }"
+    Thread.current['archive'] = date
+    Thread.current['archive_url'] = BIOMART_URL.sub(/http:\/\/biomart\./, 'http://' + date + '.archive.ensembl.')
+    Log.debug "Using Archive URL #{ Thread.current['archive_url'] }"
   end
 
   def self.unset_archive
     Log.debug "Restoring current version URL #{BIOMART_URL}"
-    @archive = nil
-    @archive_url = nil
+    Thread.current['archive'] = nil
+    Thread.current['archive_url'] = nil
   end
 
   def self.with_archive(data)
@@ -79,7 +79,7 @@ module BioMart
     query.sub!(/<!--MAIN-->/,"<Attribute name = \"#{main}\" />")
     query.sub!(/<!--ATTRIBUTES-->/, attrs.collect{|name| "<Attribute name = \"#{ name }\"/>"}.join("\n") )
 
-    url = @archive_url ? @archive_url + query.gsub(/\n/,' ') : BIOMART_URL + query.gsub(/\n/,' ')
+    url = Thread.current['archive_url'] ? Thread.current['archive_url'] + query.gsub(/\n/,' ') : BIOMART_URL + query.gsub(/\n/,' ')
 
     begin
       response = Open.read(url, open_options.dup)
@@ -191,8 +191,8 @@ module BioMart
   def self.tsv(database, main, attrs = nil, filters = nil, data = nil, open_options = {})
     attrs ||= []
 
-    if @archive_url 
-      attrs = attrs.reject{|attr| (MISSING_IN_ARCHIVE[@archive] || []).include? attr[1]}
+    if Thread.current['archive_url'] 
+      attrs = attrs.reject{|attr| (MISSING_IN_ARCHIVE[Thread.current['archive']] || []).include? attr[1]}
     end
 
 

data/share/Ensembl/release_dates

@@ -1,6 +1,5 @@
 #: :type=:single
 #Release	build
-
 release-75	feb2014
 release-74	dec2013
 release-73	sep2013

data/share/install/Organism/Hsa/Rakefile

@@ -99,5 +99,6 @@ $biomart_go_2009= [
   ["GO CC ID", 'go_cellular_component_id'],
 ]
 
-$namespace = File.basename(File.dirname(File.expand_path(__FILE__)))
+#$namespace = File.basename(File.dirname(File.expand_path(__FILE__)))
+Thread.current["namespace"] = File.basename(File.dirname(File.expand_path(__FILE__)))
 load File.join(File.dirname(__FILE__), '../organism_helpers.rb')

data/share/install/Organism/organism_helpers.rb

@@ -1,6 +1,7 @@
 require 'net/ftp'
 require 'rbbt/sources/ensembl_ftp'
 
+#Thread.current['namespace'] = $namespace
 
 $biomart_ensembl_gene = ['Ensembl Gene ID', 'ensembl_gene_id']
 $biomart_ensembl_protein = ['Ensembl Protein ID', 'ensembl_peptide_id']
@@ -92,7 +93,7 @@ file 'ortholog_key' do |t|
 end
 
 file 'identifiers' do |t|
-  identifiers = BioMart.tsv($biomart_db, $biomart_ensembl_gene, $biomart_identifiers, [], nil, :namespace => $namespace)
+  identifiers = BioMart.tsv($biomart_db, $biomart_ensembl_gene, $biomart_identifiers, [], nil, :namespace => Thread.current['namespace'])
   identifiers.unnamed =  true
 
   $biomart_identifiers.each do |name, key, prefix|
@@ -150,7 +151,7 @@ end
 
 
 file 'protein_identifiers' do |t|
-  identifiers = BioMart.tsv($biomart_db, $biomart_ensembl_protein, $biomart_protein_identifiers, [], nil, :namespace => $namespace)
+  identifiers = BioMart.tsv($biomart_db, $biomart_ensembl_protein, $biomart_protein_identifiers, [], nil, :namespace => Thread.current['namespace'])
   $biomart_protein_identifiers.each do |name, key, prefix|
     if prefix
       identifiers.process name do |field, key, values| field.each{|v| v.replace "#{prefix}:#{v}"} end
@@ -161,7 +162,7 @@ file 'protein_identifiers' do |t|
 end
 
 file 'transcript_probes' do |t|
-  identifiers = BioMart.tsv($biomart_db, $biomart_ensembl_transcript, $biomart_probe_identifiers, [], nil, :namespace => $namespace)
+  identifiers = BioMart.tsv($biomart_db, $biomart_ensembl_transcript, $biomart_probe_identifiers, [], nil, :namespace => Thread.current['namespace'])
   $biomart_probe_identifiers.each do |name, key, prefix|
     if prefix
       identifiers.process name do |field, key, values| field.each{|v| v.replace "#{prefix}:#{v}"} end
@@ -172,13 +173,13 @@ file 'transcript_probes' do |t|
 end
 
 file 'gene_transcripts' do |t|
-  transcripts = BioMart.tsv($biomart_db, $biomart_ensembl_gene, $biomart_gene_transcript, [], nil, :type => :flat, :namespace => $namespace)
+  transcripts = BioMart.tsv($biomart_db, $biomart_ensembl_gene, $biomart_gene_transcript, [], nil, :type => :flat, :namespace => Thread.current['namespace'])
 
   Misc.sensiblewrite(t.name, transcripts.to_s)
 end
 
 file 'transcripts' => 'gene_positions' do |t|
-  transcripts = BioMart.tsv($biomart_db, $biomart_ensembl_transcript, $biomart_transcript, [], nil, :type => :list, :namespace => $namespace)
+  transcripts = BioMart.tsv($biomart_db, $biomart_ensembl_transcript, $biomart_transcript, [], nil, :type => :list, :namespace => Thread.current['namespace'])
   transcripts.attach TSV.open('gene_positions'), :fields => ["Chromosome Name"]
 
   Misc.sensiblewrite(t.name, transcripts.to_s)
@@ -191,7 +192,7 @@ file 'gene_positions' do |t|
 end
 
 file 'gene_sequence' do |t|
-  sequences = BioMart.tsv($biomart_db, $biomart_ensembl_gene, $biomart_gene_sequence, [], nil, :type => :flat, :namespace => $namespace)
+  sequences = BioMart.tsv($biomart_db, $biomart_ensembl_gene, $biomart_gene_sequence, [], nil, :type => :flat, :namespace => Thread.current['namespace'])
 
   Misc.sensiblewrite(t.name) do |f|
     f.puts "#: :type=:single"
@@ -208,20 +209,20 @@ file 'gene_sequence' do |t|
 end
 
 file 'exons' => 'gene_positions' do |t|
-  exons = BioMart.tsv($biomart_db, $biomart_ensembl_exon, $biomart_exons, [], nil, :merge => false, :type => :list, :namespace => $namespace)
+  exons = BioMart.tsv($biomart_db, $biomart_ensembl_exon, $biomart_exons, [], nil, :merge => false, :type => :list, :namespace => Thread.current['namespace'])
   exons.attach TSV.open('gene_positions'), :fields => ["Chromosome Name"]
 
   Misc.sensiblewrite(t.name, exons.to_s)
 end
 
 file 'transcript_exons' do |t|
-  exons = BioMart.tsv($biomart_db, $biomart_ensembl_transcript, $biomart_transcript_exons, [], nil, :keep_empty => true, :namespace => $namespace)
+  exons = BioMart.tsv($biomart_db, $biomart_ensembl_transcript, $biomart_transcript_exons, [], nil, :keep_empty => true, :namespace => Thread.current['namespace'])
 
   Misc.sensiblewrite(t.name, exons.to_s)
 end
 
 file 'exon_phase' do |t|
-  exons = BioMart.tsv($biomart_db, $biomart_ensembl_exon, $biomart_exon_phase, [], nil, :keep_empty => true, :namespace => $namespace)
+  exons = BioMart.tsv($biomart_db, $biomart_ensembl_exon, $biomart_exon_phase, [], nil, :keep_empty => true, :namespace => Thread.current['namespace'])
 
   Misc.sensiblewrite(t.name, exons.to_s)
 end
@@ -263,11 +264,11 @@ $biomart_variation_id = ["SNP ID", "refsnp_id"]
 $biomart_variation_position = [["Chromosome Name", "chr_name"], ["Chromosome Start", "chrom_start"], ["Variant Alleles", "allele"]]
 
 file 'germline_variations' do |t|
-  BioMart.tsv($biomart_db_germline_variation, $biomart_variation_id, $biomart_variation_position, [], nil, :keep_empty => true, :type => :list, :filename => t.name, :namespace => $namespace)
+  BioMart.tsv($biomart_db_germline_variation, $biomart_variation_id, $biomart_variation_position, [], nil, :keep_empty => true, :type => :list, :filename => t.name, :namespace => Thread.current['namespace'])
 end
 
 file 'somatic_variations' do |t|
-  BioMart.tsv($biomart_db_somatic_variation, $biomart_variation_id, $biomart_variation_position, [], nil, :keep_empty => true, :type => :list, :filename => t.name, :namespace => $namespace)
+  BioMart.tsv($biomart_db_somatic_variation, $biomart_variation_id, $biomart_variation_position, [], nil, :keep_empty => true, :type => :list, :filename => t.name, :namespace => Thread.current['namespace'])
 end
 
 
@@ -275,7 +276,7 @@ end
 
 file 'gene_pmids' do |t|
   tsv =  Entrez.entrez2pubmed($taxs)
-  text = "#: :namespace=#{$namespace}\n"
+  text = "#: :namespace=#{Thread.current['namespace']}\n"
   text += "#Entrez Gene ID\tPMID"
   tsv.each do |gene, pmids|
     text << "\n" << gene << "\t" << pmids * "|"
@@ -322,7 +323,7 @@ file 'exon_offsets' => %w(exons transcript_exons gene_transcripts transcripts tr
   transcript_info  = TSV.open('transcripts', :list, :fields => ["Ensembl Protein ID"])
   transcript_exons = TSV.open('transcript_exons', :double, :fields => ["Ensembl Exon ID","Exon Rank in Transcript"])
 
-  string = "#: :namespace=#{$namespace}\n"
+  string = "#: :namespace=#{Thread.current['namespace']}\n"
   string += "#Ensembl Exon ID\tEnsembl Transcript ID\tOffset\n"
 
   exon_transcripts.unnamed = true
@@ -353,7 +354,7 @@ end
 
 file 'gene_go' do |t|
   if File.basename(FileUtils.pwd) =~ /^[a-z]{3}([0-9]{4})$/i and $1.to_i <= 2009
-    goterms = BioMart.tsv($biomart_db, $biomart_ensembl_gene, $biomart_go_2009, [], nil, :type => :double, :namespace => $namespace)
+    goterms = BioMart.tsv($biomart_db, $biomart_ensembl_gene, $biomart_go_2009, [], nil, :type => :double, :namespace => Thread.current['namespace'])
 
     goterms.each do |key, values|
       values.each do |list| list.uniq! end
@@ -371,7 +372,7 @@ file 'gene_go' do |t|
 
     Misc.sensiblewrite(t.name, goterms.slice(["GO ID", "GO Namespace"]).to_s)
   else
-    goterms = BioMart.tsv($biomart_db, $biomart_ensembl_gene, $biomart_go, [], nil, :type => :double, :namespace => $namespace)
+    goterms = BioMart.tsv($biomart_db, $biomart_ensembl_gene, $biomart_go, [], nil, :type => :double, :namespace => Thread.current['namespace'])
 
     Misc.sensiblewrite(t.name, goterms.to_s)
   end
@@ -425,19 +426,19 @@ end
 
 
 file 'gene_biotype' do |t|
-  biotype = BioMart.tsv($biomart_db, $biomart_ensembl_gene, $biomart_gene_biotype, [], nil, :type => :single, :namespace => $namespace)
+  biotype = BioMart.tsv($biomart_db, $biomart_ensembl_gene, $biomart_gene_biotype, [], nil, :type => :single, :namespace => Thread.current['namespace'])
 
   Misc.sensiblewrite(t.name, biotype.to_s)
 end
 
 file 'gene_pfam' do |t|
-  pfam = BioMart.tsv($biomart_db, $biomart_ensembl_gene, $biomart_pfam, [], nil, :type => :double, :namespace => $namespace)
+  pfam = BioMart.tsv($biomart_db, $biomart_ensembl_gene, $biomart_pfam, [], nil, :type => :double, :namespace => Thread.current['namespace'])
 
   Misc.sensiblewrite(t.name, pfam.to_s)
 end
 
 file 'chromosomes' do |t|
-  goterms = BioMart.tsv($biomart_db, ['Chromosome Name', "chromosome_name"] , [] , [], nil, :type => :double, :namespace => $namespace)
+  goterms = BioMart.tsv($biomart_db, ['Chromosome Name', "chromosome_name"] , [] , [], nil, :type => :double, :namespace => Thread.current['namespace'])
 
   Misc.sensiblewrite(t.name, goterms.to_s)
 end
@@ -483,7 +484,7 @@ rule /^chromosome_.*/ do |t|
 
   raise "Fasta file for chromosome not found: '#{ chr }' - #{ archive }, #{ release }" if file.nil?
 
-  Log.debug("Downloading chromosome sequence: #{ file }")
+  Log.debug("Downloading chromosome sequence: #{ file } - #{release} #{t.name}")
 
   Misc.lock t.name + '.rake' do
     TmpFile.with_file do |tmpfile|
@@ -497,13 +498,13 @@ end
 rule /^possible_ortholog_(.*)/ do |t|
   other = t.name.match(/ortholog_(.*)/)[1]
   other_key = Organism.ortholog_key(other).produce.read
-  BioMart.tsv($biomart_db, $biomart_ensembl_gene, [["Ortholog Ensembl Gene ID", "inter_paralog_" + other_key]], [], nil, :keep_empty => false, :type => :flat, :filename => t.name, :namespace => $namespace)
+  BioMart.tsv($biomart_db, $biomart_ensembl_gene, [["Ortholog Ensembl Gene ID", "inter_paralog_" + other_key]], [], nil, :keep_empty => false, :type => :flat, :filename => t.name, :namespace => Thread.current['namespace'])
 end
 
 rule /^ortholog_(.*)/ do |t|
   other = t.name.match(/ortholog_(.*)/)[1]
   other_key = Organism.ortholog_key(other).produce.read
-  BioMart.tsv($biomart_db, $biomart_ensembl_gene, [["Ortholog Ensembl Gene ID", other_key]], [], nil, :keep_empty => false, :type => :flat, :filename => t.name, :namespace => $namespace)
+  BioMart.tsv($biomart_db, $biomart_ensembl_gene, [["Ortholog Ensembl Gene ID", other_key]], [], nil, :keep_empty => false, :type => :flat, :filename => t.name, :namespace => Thread.current['namespace'])
 end
 
 rule /[a-z]{3}[0-9]{4}\/.*/i do |t|
@@ -513,15 +514,15 @@ rule /[a-z]{3}[0-9]{4}\/.*/i do |t|
   Misc.in_dir(archive) do
     BioMart.set_archive archive
     begin
-      old_namespace = $namespace
-      $namespace = $namespace + "/" << archive
+      old_namespace = Thread.current['namespace']
+      Thread.current['namespace'] = Thread.current['namespace'] + "/" << archive
       Rake::Task[task].invoke
     rescue
       Log.error "Error producing archived (#{archive}) version of #{task}: #{t.name}"
       Log.exception $!
       raise $!
     ensure
-      $namespace = old_namespace
+      Thread.current['namespace'] = old_namespace
     end
     BioMart.unset_archive 
   end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rbbt-sources
 version: !ruby/object:Gem::Version
-  version: 3.0.8
+  version: 3.0.9
 platform: ruby
 authors:
 - Miguel Vazquez
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-06-10 00:00:00.000000000 Z
+date: 2014-06-15 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rbbt-util