rbbt-sources 3.0.12 → 3.0.13

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: fee4f38313f5e77ca9887dfcc3b9048609da7d70
-  data.tar.gz: 79061d834de12b2f2436d697f35956bef131f67f
+  metadata.gz: e02b0afc9999bf09316fdf670616018c6901ea4a
+  data.tar.gz: d2f69f47d90ee82bbe93a0ed296fc02d2039b12f
 SHA512:
-  metadata.gz: 4432dd46252b0f301a190671ea81a9c3fca32415ba03edcee7ae22dcc73251a578a18db2b7df8b154b87ca5f06862353c52429153d360b8b7484a16cb40f1b59
-  data.tar.gz: e46115accb053c9ab3a1df47cebafffe71730307005dd1a1408a33d9e74e2a0368619e021bc05fd3fadc8ea258cf96e54ed9e79c98ae4f6df7f5b6be7abdf3e6
+  metadata.gz: 529580fa92fe6641913436758a10d429d2be6343686abfa823c6e1078a18f691feba85b245e9256f1ea3cbaf604865edc0fb1cf5ba4635d8c74f6bc60f77c3e6
+  data.tar.gz: adb3c36787f15ac5910b7f3fcfb890b6e4a6200a59c49a3d7f3062970ab94ba6fc83d979ea79e5163626ff92ca69eb008e1530991cbcdaa372af5cc6bcd4eb0e

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rbbt-sources
 version: !ruby/object:Gem::Version
-  version: 3.0.12
+  version: 3.0.13
 platform: ruby
 authors:
 - Miguel Vazquez
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-08-02 00:00:00.000000000 Z
+date: 2014-08-13 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rbbt-util
@@ -92,7 +92,6 @@ files:
 - lib/rbbt/sources/COSTART.rb
 - lib/rbbt/sources/CTCAE.rb
 - lib/rbbt/sources/HPRD.rb
-- lib/rbbt/sources/InterPro.rb
 - lib/rbbt/sources/NCI.rb
 - lib/rbbt/sources/PSI_MI.rb
 - lib/rbbt/sources/STITCH.rb

data/lib/rbbt/sources/InterPro.rb

@@ -1,105 +0,0 @@
-require 'rbbt-util'
-require 'rbbt/sources/organism'
-
-module InterPro
-  extend Resource
-  self.subdir = "share/databases/InterPro"
-
-  InterPro.claim InterPro.source.protein2ipr, :url, "ftp://ftp.ebi.ac.uk/pub/databases/interpro/protein2ipr.dat.gz"
-
-  InterPro.claim InterPro.protein_domains, :proc do
-    organism = "Hsa"
-    uniprot_colum = TSV::Parser.new(Organism.protein_identifiers(organism).open).all_fields.index("UniProt/SwissProt Accession")
-    uniprots = CMD.cmd("grep -v  '^#'|cut -f #{uniprot_colum+1}", :in => Organism.protein_identifiers(organism).open).read.split("\n").collect{|l| l.split("|")}.flatten.uniq.reject{|l| l.empty?}
-   
-    tsv = nil
-    TmpFile.with_file(uniprots * "\n") do |tmpfile|
-        tsv = TSV.open(CMD.cmd("cut -f 1,2,5,6 | sort -u |grep -w -F -f #{ tmpfile }", :in => InterPro.source.protein2ipr.open, :pipe => true), :merge => true, :type => :double)
-    end
- 
-    tsv.key_field = "UniProt/SwissProt Accession"
-    tsv.fields = ["InterPro ID", "Domain Start AA", "Domain End AA"]
-    tsv.to_s
-  end
-
-  InterPro.claim InterPro.domain_names, :proc do
-    #tsv = InterPro.source.protein2ipr.tsv :key_field => 1, :fields => [2], :type => :single
-    tsv = TSV.open(CMD.cmd("cut -f 2,3 | sort -u", :in => InterPro.source.protein2ipr.open, :pipe => true), :merge => true, :type => :single)
- 
-    tsv.key_field = "InterPro ID"
-    tsv.fields = ["Domain Name"]
-    tsv.to_s
-  end
-
-  def self.name_index
-    @@name_index ||= InterPro.domain_names.tsv(:persist => true, :unnamed => true)
-  end
-
-  def self.gene_index
-    @@gene_index ||= InterPro.protein_domains.tsv(:persist => true, :key_field => "InterPro ID", :fields => ["UniProt/SwissProt Accession"], :type => :flat, :merge => true, :unnamed => true)
-  end
-
-  def self.domain_index
-    @@domain_index ||= InterPro.protein_domains.tsv(:persist => true, :unnamed => true, :key_field => "UniProt/SwissProt Accession", :fields => ["InterPro ID"], :merge => true)
-  end
-
-  def self.domain_position_index
-    @@domain_position_index ||= InterPro.protein_domains.tsv(:persist => true, :unnamed => true, :key_field => "UniProt/SwissProt Accession", :fields => ["InterPro ID", "Domain Start AA", "Domain End AA"], :type => :double, :merge => true)
-  end
-
-  def self.ens2uniprot(organism)
-    @@ens2uniprot_index ||= {}
-    @@ens2uniprot_index[organism] ||= Organism.protein_identifiers(organism).tsv(:persist => true, :unnamed => true, :fields => ["UniProt/SwissProt Accession"], :key_field => "Ensembl Protein ID", :type => :double, :merge => true)
-  end
-
-end
-
-if defined? Entity 
-  module InterProDomain
-    extend Entity
-    self.format = "InterPro ID"
-
-    self.annotation :organism
-    property :description => :array2single do
-      InterPro.name_index.values_at *self
-    end
-
-    property :name => :array2single do
-      InterPro.name_index.values_at *self
-    end
-
-    property :proteins => :array2single do
-      InterPro.gene_index.values_at(*self).
-        collect{|genes| genes = genes.uniq;  genes.organism = organism if genes.respond_to? :organism; genes }.tap{|o| Protein.setup(o, "UniProt/SwissProt Accession", organism)}
-    end
-
-    property :genes => :array2single do
-      InterPro.gene_index.values_at(*self).
-        collect{|genes| genes = [] if genes.nil?; genes = genes.uniq;  genes.organism = organism if genes.respond_to? :organism; genes }.tap{|o| Gene.setup(o, "UniProt/SwissProt Accession", organism)}
-    end
-  end
-
-  if defined? Protein and Entity === Protein
-    module Protein
-      property :interpro_domains => :array2single do
-        self.collect do |protein|
-          uniprot = (InterPro.ens2uniprot(protein.organism)[protein] || []).flatten
-          uniprot.empty? ? nil : 
-            InterPro.domain_index.values_at(*uniprot).compact.flatten.  each{|pth| pth.organism = organism if pth.respond_to? :organism }.uniq.tap{|o| InterProDomain.setup(o, organism)}
-        end
-      end
-
-      property :interpro_domain_positions => :array2single do
-        self.collect do |protein|
-          if protein.nil?
-            [].tap{|o| InterProDomain.setup(o, organism)}
-          else
-            uniprot = (InterPro.ens2uniprot(protein.organism)[protein] || []).flatten
-            uniprot.empty? ? nil : 
-              InterPro.domain_position_index.values_at(*uniprot).compact.flatten(1).tap{|o| InterProDomain.setup(o, organism)}
-          end
-        end
-      end
-    end
-  end
-end