rbbt-sources 3.0.24 → 3.0.25

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 907dd95d8535034e80e84c5df8194531bd86d1e3
-  data.tar.gz: 4a67f08de4f4b83da9967d009fd32f29d1f17958
+  metadata.gz: a527bda568cbbed6c8b2d1d97e9898fcd6b8b63c
+  data.tar.gz: 38dbe0d9ea4c5668f6368181c1b64a0790f2fb34
 SHA512:
-  metadata.gz: ca004897b4c92532dc14f4e2bafce4b90919c54598b27d428208b4d5115f4333cafc471596852e0de5cabb6d5ce226f34dbc61d779d38a5d06410a9dc9f346a9
-  data.tar.gz: 253068f65d0e9cd3733be2d2cac1865321d6ef6cd5272bf71a7ff5b11a4b740a91ed365ab11ceaf50d165489907b1d71aeecd8c06184b49241a828e635e975d0
+  metadata.gz: c8745c28b10e9b0312c06bbaf14ef42efa80467d3c64092a2440f426cd4139eae4151d93e9c471d66f6ecf2eeb26edd8d15c527369a14ac0908559f6c82f0a12
+  data.tar.gz: aecdcddd4dfd4e7474958aa6bb57ca8eaf06c346cd96e3c6515e8900224375641503976af8c60bb6949163ae53a79d6ede092512377218f0f3a028d556e02b5f

data/etc/biomart/missing_in_archive

@@ -1,5 +1,6 @@
 all:
     - refseq_mrna
+<aug2014:
     - external_gene_name~external_gene_id
 may2010:
     - agilent_wholegenome
@@ -33,5 +34,3 @@ aug2007:
     - hgnc
     - hgnc_id
     - illumina_humanwg_6_v3
-dec2014:
-    - external_gene_id

data/lib/rbbt/sources/biomart.rb

@@ -190,9 +190,18 @@ module BioMart
   def self.tsv(database, main, attrs = nil, filters = nil, data = nil, open_options = {})
     attrs ||= []
 
-    if Thread.current['archive'] 
-      missing = MISSING_IN_ARCHIVE[Thread.current['archive']] || []
-      missing += MISSING_IN_ARCHIVE['all'] || []
+    if current_archive = Thread.current['archive'] 
+      missing = MISSING_IN_ARCHIVE['all'] || []
+      missing += MISSING_IN_ARCHIVE[current_archive] || []
+      MISSING_IN_ARCHIVE.each do |k,v|
+        if k =~ /^<(.*)/ 
+          t = $1.strip
+          missing+=v if Organism.compare_archives(current_archive, t) == -1
+        elsif k=~ /^>(.*)/ 
+          t = $1.strip
+          missing+=v if Organism.compare_archives(current_archive, t) == 1
+        end
+      end
       attrs = attrs.uniq.reject{|attr| missing.include? attr[1]}
       changes = {}
       missing.select{|m| m.include? "~" }.each do |str|

data/lib/rbbt/sources/organism.rb

@@ -6,6 +6,21 @@ module Organism
   self.pkgdir = "rbbt"
   self.subdir = "share/organisms"
 
+  ARCHIVE_MONTH_INDEX = {}
+  %w(jan feb mar apr may jun jul aug sep oct nov dec).each_with_index{|d,i| ARCHIVE_MONTH_INDEX[d] = i }
+  def self.compare_archives(a1, a2)
+    m1,y1 = a1.match(/(...)(\d+)/).values_at 1, 2
+    m2,y2 = a2.match(/(...)(\d+)/).values_at 1, 2
+
+    y1 = y1.to_f
+    y2 = y2.to_f
+
+    ycmp = y1 <=> y2
+    return ycmp unless ycmp == 0
+
+    ARCHIVE_MONTH_INDEX[m1] <=> ARCHIVE_MONTH_INDEX[m2]
+  end
+
   def self.default_code(organism = "Hsa")
     organism.split("/").first << "/feb2014"
   end
@@ -199,7 +214,14 @@ module Organism
     @@gene_start_end ||= {}
     gene_start_end = @@gene_start_end[organism] ||= Organism.gene_positions(organism).tsv(:persist => true, :key_field => "Ensembl Gene ID", :fields => ["Gene Start", "Gene End"], :type => :list, :cast => :to_i, :unmamed => true)
 
-    ranges = genes.collect{|gene| start, eend = gene_start_end[gene]; (start..eend) }
+    ranges = genes.collect{|gene| 
+      start, eend = gene_start_end[gene]
+      if start.nil? or eend.nil?
+        Log.low "Gene #{ gene } does not have range" 
+        next 
+      end
+      (start..eend) 
+    }.compact
     Misc.total_length(ranges)
   end
 
@@ -220,7 +242,10 @@ module Organism
     exons = gene_exons.values_at(*genes).compact.flatten.uniq
 
     exon_ranges = exons.collect{|exon|
-      Log.low "Exon #{ exon } does not have range" and next if not exon_ranges.include? exon
+      if not exon_ranges.include? exon
+        Log.low "Exon #{ exon } does not have range" 
+        next 
+      end
       pos = exon_ranges[exon]
       (pos.first..pos.last)
     }.compact

data/lib/rbbt/sources/uniprot.rb

@@ -117,7 +117,7 @@ module UniProt
         eend = eend.to_i
         start, eend = eend, start if start > eend
       rescue
-        Log.warn("Error process Uniprot PDB line: #{line}")
+        Log.warn("Error process Uniprot PDB line: #{l}")
         next
       end
       pdb[id.downcase] = {:method => method, :resolution => resolution, :region => (start..eend), :chains => chains}

data/share/install/Organism/organism_helpers.rb

@@ -566,7 +566,7 @@ file 'transcript_sequence' => ["exons", "transcript_exons"] do |t|
   transcript_sequence = {}
   chr_transcript_ranges.each do |chr, transcript_ranges|
     begin
-      raise "LRG, GL, HG, NT, and HSCHR chromosomes not supported: #{chr}" if chr =~ /^(?:LRG_|GL0|HG|HSCHR|NT)/
+      raise "LRG, GL, HG, NT, KI, and HSCHR chromosomes not supported: #{chr}" if chr =~ /(?:LRG_|GL0|HG|HSCHR|NT|KI)/
       p = File.expand_path("./chromosome_#{chr}")
       Organism.root.annotate p
       p.sub!(%r{.*/organisms/},'share/organisms/')

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rbbt-sources
 version: !ruby/object:Gem::Version
-  version: 3.0.24
+  version: 3.0.25
 platform: ruby
 authors:
 - Miguel Vazquez
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-05-13 00:00:00.000000000 Z
+date: 2015-05-20 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rbbt-util