rbbt-util 5.14.33 → 5.14.34

data/lib/rbbt/tsv/util.rb

@@ -171,7 +171,7 @@ module TSV
     str = "" 
     str << preamble.strip << "\n" if preamble and not preamble.empty?
     if fields
-      str << "#" << (key_field || "ID") << sep << fields * sep << "\n"
+      str << "#" << (key_field || "ID").to_s << sep << (fields * sep) << "\n" 
     end
 
     str

data/lib/rbbt/tsv.rb

@@ -124,7 +124,8 @@ module TSV
 
   def self.parse(stream, data, options = {})
 
-    parser = TSV::Parser.new stream, options
+    parser = Misc.process_options options, :parser
+    parser = TSV::Parser.new stream, options if parser.nil?
 
     # dump with tchmgr
     if defined? TokyoCabinet and TokyoCabinet::HDB === data and parser.straight and

data/lib/rbbt/util/R/model.rb

@@ -16,7 +16,7 @@ module R
 
   self.model_dir = Rbbt.var.R.models
   class Model
-    R_METHOD = :debug
+    R_METHOD = :eval
 
     attr_accessor :name, :formula
     def initialize(name, formula, options = {})

data/lib/rbbt/util/log.rb

@@ -235,8 +235,8 @@ module Log
 
   def self.tsv(tsv)
     STDERR.puts Log.color :magenta, "TSV log: " << Log.last_caller(caller)
-    iif tsv
-    ppp tsv.summary
+    puts Log.color(:blue, "=> "<< Misc.fingerprint(tsv), true) 
+    puts Log.color(:cyan, "=> " << tsv.summary)
   end
 
   def self.stack(stack)

data/lib/rbbt/util/misc/bgzf.rb

@@ -1,3 +1,5 @@
+require 'rbbt/persist'
+require 'set'
 
 module Bgzf
   attr_accessor :data_offset, :compressed_stream, :block_cache_size

data/lib/rbbt/util/misc/inspect.rb

@@ -161,7 +161,7 @@ module Misc
 
       end
 
-      str << "_" << hash2md5(v.info) if defined? Annotated and Annotated === v and not AssociationItem === v
+      str << "_" << hash2md5(v.info) if defined? Annotated and Annotated === v and not (defined? AssociationItem and AssociationItem === v)
     end
     hash.unnamed = unnamed if hash.respond_to? :unnamed
 

data/lib/rbbt-util.rb

@@ -1,13 +1,17 @@
 require 'rbbt'
+require 'rbbt/util/misc'
 require 'rbbt/util/open'
-require 'rbbt/util/cmd'
-require 'rbbt/util/tmpfile'
-require 'rbbt/util/filecache'
-require 'rbbt/tsv'
+Open.cachedir        = Rbbt.var.cache["open-remote"].find :user
+
 require 'rbbt/persist'
-require 'rbbt/util/misc'
+Persist.cachedir = Rbbt.var.cache.persistence
 
+require 'rbbt/util/filecache'
 FileCache.cachedir   = Rbbt.var.cache.filecache.find :user
-Open.cachedir        = Rbbt.var.cache["open-remote"].find :user
+
+require 'rbbt/util/tmpfile'
 TmpFile.tmpdir       = Rbbt.tmp.find :user
-Persist.cachedir = Rbbt.var.cache.persistence
+
+require 'rbbt/util/cmd'
+require 'rbbt/tsv'
+

data/lib/rbbt.rb

@@ -1,5 +1,4 @@
 require 'rbbt/resource'
-
 module Rbbt
   extend Resource
   pkgdir = 'rbbt'

data/share/rbbt_commands/app/start

@@ -3,7 +3,7 @@
 require 'rbbt-util'
 require 'rbbt/util/simpleopt'
 
-options = SOPT.get "-e--environment*:-p--port*:-s--server*:-f--finder-R--Rserve_session*"
+options = SOPT.get "-e--environment*:-p--port*:-s--server*:-f--finder:-R--Rserve_session*"
 options[:Port] ||= options[:port]
 
 

data/share/rbbt_commands/tsv/change_id

@@ -11,7 +11,7 @@ $ rbbt command command [options] -f file
 
 Use - to read from STDIN. The `field` is the tsv column to change.
 the `target` is the target format, as appearing in the `identifiers`
-file or in `Organism.identifiers("Hsa")`.
+file or in `Organism.identifiers(Organism.default_code("Hsa"))`.
 
 -f--file* File to process
 -f--field* Field to change
@@ -26,7 +26,7 @@ SOPT.require options, :file, :field, :target
 
 require 'rbbt/sources/organism'
 
-identifiers = options[:identifiers] || Organism.identifiers("Hsa").find
+identifiers = options[:identifiers] || Organism.identifiers(Organism.default_code("Hsa")).find
 
 begin
   index = TSV.index(identifiers, :persist => true, :target => target, :merge => true)

data/test/rbbt/association/test_database.rb

@@ -0,0 +1,61 @@
+require File.expand_path(File.dirname(__FILE__) + '/../../test_helper')
+require 'rbbt-util'
+require 'rbbt/association'
+require 'rbbt/association'
+require 'rbbt/association/database'
+
+class TestAssociationDatabase < Test::Unit::TestCase
+   
+  EFFECT =<<-END
+#: :sep=" "#:type=:double
+#SG TG Effect directed?
+MDM2 TP53 inhibition false
+TP53 NFKB1|GLI1 activation|activation true|true
+  END
+
+  EFFECT_OPTIONS = {
+    :source => "SG=~Associated Gene Name",
+    :target => "TG=~Associated Gene Name=>Ensembl Gene ID",
+    :persist => false,
+    :identifiers => datafile_test('identifiers'),
+  }
+
+  EFFECT_TSV = TSV.open EFFECT, EFFECT_OPTIONS.dup
+  def test_tsv
+    assert_equal %w(NFKB1 GLI1), EFFECT_TSV["TP53"]["TG"]
+  end
+
+  EFFECT_DATABASE = Association.database(EFFECT_TSV, EFFECT_OPTIONS.dup)
+  
+  def test_open_new_tsv
+    tsv = Association.database EFFECT_TSV, :fields => ["Effect"], :target => "TG=~Associated Gene Name=>Ensembl Gene ID"
+    assert_equal ["Ensembl Gene ID", "Effect"], tsv.fields
+    assert_equal "SG", tsv.key_field
+    assert tsv.include? "MDM2"
+  end
+
+  def test_open_new_tsv_reverse
+    tsv = Association.database EFFECT_TSV, :fields => ["Effect"], :target => "SG=~Associated Gene Name=>Ensembl Gene ID", :zipped => true
+    assert_equal ["Ensembl Gene ID", "Effect"], tsv.fields
+    assert_equal "TG", tsv.key_field
+    assert tsv.include? "NFKB1"
+  end
+
+  def test_database_translate
+    database = Association.database(EFFECT_TSV, EFFECT_OPTIONS.merge({:source => "SG=~Associated Gene Name=>Ensembl Gene ID", :target => "TG"}))
+    assert_equal %w(inhibition), database["ENSG00000135679"]["Effect"]
+    assert_equal %w(activation activation), database["ENSG00000141510"]["Effect"]
+  end
+
+  def test_database
+    assert_equal %w(inhibition), EFFECT_DATABASE["MDM2"]["Effect"]
+    assert_equal %w(activation activation), EFFECT_DATABASE["TP53"]["Effect"]
+  end
+
+  def test_index_list
+    file = datafile_test('gene_ages')
+    tsv = Association.database(file)
+    assert_equal [["Bilateria"], ["Euteleostomi"], ["Duplicate"]], tsv["ENSG00000000003"]
+  end
+
+end

data/test/rbbt/association/test_index.rb

@@ -1,37 +1,82 @@
 require File.expand_path(File.dirname(__FILE__) + '/../../test_helper')
-require 'rbbt/workflow'
+require 'rbbt/util/misc'
 require 'rbbt/association'
 require 'rbbt/association/index'
 
-#Workflow.require_workflow "Genomics"
-#require 'rbbt/entity/gene'
-
 class TestAssociationIndex < Test::Unit::TestCase
+   
+  EFFECT =<<-END
+#: :sep=" "#:type=:double
+#SG TG Effect directed?
+MDM2 TP53 inhibition false
+TP53 NFKB1|GLI1 activation|activation true|true
+  END
+
+  EFFECT_OPTIONS = {
+    :source => "SG=~Associated Gene Name",
+    :target => "TG=~Associated Gene Name=>Ensembl Gene ID",
+    :persist => false,
+    :identifiers => datafile_test('identifiers'),
+    :namespace => "Hsa"
+  }
+
+  EFFECT_TSV = TSV.open EFFECT, EFFECT_OPTIONS.dup
+
+  def test_index_no_persist_string
+    tsv = Association.index(EFFECT, EFFECT_OPTIONS.merge(:source => "TG", :target => "SG=~Associated Gene Name"), :persist => false)
+    tsv.unnamed = false
+    assert_equal "inhibition", tsv["TP53~MDM2"]["Effect"]
+  end
+
+  def test_index_no_persist_string_undirected
+    tsv = Association.index(EFFECT, EFFECT_OPTIONS.merge(:undirected => true, :source => "TG", :target => "SG=~Associated Gene Name"), :persist => false)
+    tsv.unnamed = false
+    assert_equal "inhibition", tsv["TP53~MDM2"]["Effect"]
+  end
 
-  def setup
+  def test_index_persist_string
+    tsv = Association.index(EFFECT, EFFECT_OPTIONS.merge(:source => "SG", :target => "TG=~Associated Gene Name"), :persist => true, :update => true)
+    tsv.unnamed = false
+    assert_equal "inhibition", tsv["MDM2~TP53"]["Effect"]
+    assert_equal %w(TP53~NFKB1 TP53~GLI1).sort, tsv.match("TP53").sort
   end
 
-  def teardown
+  def test_index_persist_reverse
+    tsv = Association.index(EFFECT, EFFECT_OPTIONS.merge(:source => "TG", :target => "SG=~Associated Gene Name"), :persist => true, :update => true).reverse
+    tsv.unnamed = false
+    assert_equal "inhibition", tsv["MDM2~TP53"]["Effect"]
+    assert_equal %w(MDM2~TP53), tsv.match("MDM2")
   end
 
+  def test_index_persist_undirected
+    tsv = Association.index(EFFECT, EFFECT_OPTIONS.merge(:undirected => true, :source => "TG=~Associated Gene Name", :target => "SG=~Associated Gene Name"), :persist => true)
+    tsv.unnamed = false
+    assert_equal "inhibition", tsv["MDM2~TP53"]["Effect"]
+    assert_equal "inhibition", tsv["TP53~MDM2"]["Effect"]
+    assert_equal %w(MDM2~TP53), tsv.match("MDM2")
+  end
 
-  def __test_subset
-    require 'rbbt/sources/pina'
-    require 'rbbt/sources/kegg'
-    tp53 = Gene.setup("TP53", "Associated Gene Name", "Hsa/jan2013").ensembl
-    index = Association.index(Pina.protein_protein,
-                              {:namespace => tp53.organism, 
-                                :target => "Interactor UniProt/SwissProt Accession=~UniProt/SwissProt Accession",
-                                :format => "Ensembl Gene ID",
-                                :undirected => true}, 
-                                {:dir => '/tmp/test_association3', :update => false})
-    genes = tp53.pina_interactors.ensembl
-    genes << tp53
+  def test_index_persist_directed_subset
+    tsv = Association.index(EFFECT, EFFECT_OPTIONS.merge(:undirected => true, :source => "TG=~Associated Gene Name", :target => "SG=~Associated Gene Name"), :persist => true)
+    tsv.unnamed = false
+    assert_equal %w(TP53~GLI1 TP53~MDM2).sort, tsv.subset(["TP53"], ["GLI1","MDM2"]).sort
+    assert_equal %w(MDM2~TP53).sort, tsv.subset(["MDM2"], :all).sort
+    assert_equal %w(GLI1~TP53).sort, tsv.subset(["GLI1"], :all).sort
+    assert_equal %w(TP53~GLI1).sort, tsv.subset(:all, ["GLI1"]).sort
+  end
 
-    Misc.benchmark(100) do
-      index.subset_entities("Ensembl Gene ID" => genes).length
-    end
+  def test_index_persist_undirected_subset
+    tsv = Association.index(EFFECT, EFFECT_OPTIONS.merge(:undirected => false, :source => "SG=~Associated Gene Name", :target => "TG=~Associated Gene Name"), :persist => true)
+    tsv.unnamed = false
+    assert_equal %w(TP53~GLI1 TP53~NFKB1).sort, tsv.subset(["TP53"], ["GLI1","MDM2", "NFKB1"]).sort
+  end
 
-    assert_equal 204, index.subset_entities("Ensembl Gene ID" => genes).select{|m| m.partition("~")[0] == tp53}.uniq.length
+  def test_index_flat
+    require 'rbbt/sources/tfacts'
+    file = TFacts.regulators
+    tsv = Association.index(file,  :type => :flat, :source => "Transcription Factor Associated Gene Name=~Associated Gene Name", :merge => true)
+    assert tsv.match("TP53").length > 10
   end
+
+
 end

data/test/rbbt/association/test_open.rb

@@ -0,0 +1,68 @@
+require File.expand_path(File.dirname(__FILE__) + '/../../test_helper')
+require 'rbbt/util/misc'
+require 'rbbt/association'
+require 'rbbt/association/open'
+
+class TestAssociationOpen < Test::Unit::TestCase
+   
+  EFFECT =<<-END
+#: :sep=" "#:type=:double
+#SG TG Effect directed?
+MDM2 TP53 inhibition false
+TP53 NFKB1|GLI1 activation|activation true|true
+  END
+
+  EFFECT_OPTIONS = {
+    :source => "SG=~Associated Gene Name",
+    :target => "TG=~Associated Gene Name=>Ensembl Gene ID",
+    :persist => false,
+    :identifiers => datafile_test('identifiers'),
+    :namespace => "Hsa"
+  }
+
+  EFFECT_TSV = TSV.open EFFECT, EFFECT_OPTIONS.dup
+
+  def test_open_no_persist
+    tsv = Association.open(EFFECT_TSV, EFFECT_OPTIONS, :persist => false)
+    assert_equal "ENSG00000141510", tsv["MDM2"]["Ensembl Gene ID"].first
+    assert_equal ["false"], tsv["MDM2"]["directed?"]
+  end
+
+  def test_open_no_persist_undirected
+    tsv = Association.open(EFFECT_TSV, :source => "SG=~Gene", :target => "TG=~Gene", :undirected => true, :persist => false)
+    assert_equal "TP53", tsv["MDM2"]["Gene"].first
+    assert tsv["TP53"]["Gene"].include? "MDM2"
+  end
+
+  def test_open_persist
+    tsv = Association.open(EFFECT_TSV, EFFECT_OPTIONS, :persist => true, :update => true)
+    assert_equal "ENSG00000141510", tsv["MDM2"]["Ensembl Gene ID"].first
+    assert_equal ["false"], tsv["MDM2"]["directed?"]
+  end
+
+  def test_open_no_persist_string
+    tsv = Association.open(EFFECT, EFFECT_OPTIONS.merge(:source => "TG", :target => "SG=~Associated Gene Name"), :persist => false)
+    tsv.include? "TP53"
+    assert_equal ["MDM2"], tsv["TP53"]["Associated Gene Name"]
+  end
+
+  def test_index_no_persist_string
+    tsv = Association.index(EFFECT, EFFECT_OPTIONS.merge(:source => "TG", :target => "SG=~Associated Gene Name"), :persist => false)
+    tsv.unnamed = false
+    assert_equal "inhibition", tsv["TP53~MDM2"]["Effect"]
+  end
+
+  def test_index_persist_string
+    tsv = Association.index(EFFECT, EFFECT_OPTIONS.merge(:source => "SG", :target => "TG=~Associated Gene Name"), :persist => true)
+    tsv.unnamed = false
+    assert_equal "inhibition", tsv["MDM2~TP53"]["Effect"]
+    assert_equal %w(TP53~GLI1 TP53~NFKB1), tsv.match("TP53").sort
+  end
+
+  def test_index_persist_reverse
+    tsv = Association.index(EFFECT, EFFECT_OPTIONS.merge(:source => "TG", :target => "SG=~Associated Gene Name"), :persist => true).reverse
+    tsv.unnamed = false
+    assert_equal "inhibition", tsv["MDM2~TP53"]["Effect"]
+    assert_equal %w(MDM2~TP53), tsv.match("MDM2")
+  end
+end

data/test/rbbt/association/test_util.rb

@@ -0,0 +1,108 @@
+require File.expand_path(File.dirname(__FILE__) + '/../../test_helper')
+require 'rbbt/util/misc'
+require 'rbbt/association'
+require 'rbbt/association'
+require 'rbbt/association/util'
+
+class TestAssociationUtil < Test::Unit::TestCase
+
+  def test_specs
+    spec = Association.parse_field_specification "=~Associated Gene Name=>Ensembl Gene ID"
+    assert_equal [nil, "Associated Gene Name", "Ensembl Gene ID"], spec
+
+    spec = Association.parse_field_specification "TG=~Associated Gene Name=>Ensembl Gene ID"
+    assert_equal ["TG", "Associated Gene Name", "Ensembl Gene ID"], spec
+
+    spec = Association.parse_field_specification "TG=~Associated Gene Name"
+    assert_equal ["TG", "Associated Gene Name", nil], spec
+
+    spec = Association.parse_field_specification "Associated Gene Name=>Ensembl Gene ID"
+    assert_equal ["Associated Gene Name", nil, "Ensembl Gene ID"], spec
+  end
+
+  def test_normalize_specs
+    spec = Association.normalize_specs "TG=~Associated Gene Name=>Ensembl Gene ID", %w(SG TG Effect directed?)
+    assert_equal ["TG", "Associated Gene Name", "Ensembl Gene ID"], spec
+
+    spec = Association.normalize_specs "Ensembl Gene ID", %w(SG TG Effect directed?)
+    assert_equal ["Ensembl Gene ID", nil, nil], spec
+
+    spec = Association.normalize_specs "Ensembl Gene ID"
+    assert_equal ["Ensembl Gene ID", nil, nil], spec
+  end
+
+  def test_process_specs
+    spec = Association.extract_specs %w(SG TG Effect directed?), :target => "TG=~Associated Gene Name=>Ensembl Gene ID"
+    assert_equal ["TG", "Associated Gene Name", "Ensembl Gene ID"], spec[:target]
+    assert_equal ["SG", nil, nil], spec[:source]
+
+    spec = Association.extract_specs %w(SG TG Effect directed?)
+    assert_equal ["SG", nil, nil], spec[:source]
+    assert_equal ["TG", nil,nil], spec[:target]
+
+    spec = Association.extract_specs %w(SG TG Effect directed?), :source => "TG"
+    assert_equal ["TG", nil, nil], spec[:source]
+    assert_equal ["SG", nil,nil], spec[:target]
+
+    spec = Association.extract_specs %w(SG TG Effect directed?), :source => "SG"
+    assert_equal ["SG", nil, nil], spec[:source]
+    assert_equal ["TG", nil,nil], spec[:target]
+
+    spec = Association.extract_specs %w(SG TG Effect directed?), :target => "SG"
+    assert_equal ["TG", nil, nil], spec[:source]
+    assert_equal ["SG", nil,nil], spec[:target]
+  end
+
+  def test_headers
+    spec = Association.headers  %w(SG TG Effect directed?), %w(Effect)
+    assert_equal 0, spec[0]
+    assert_equal 1, spec[1][0]
+
+    spec = Association.headers  %w(SG TG Effect directed?), %w(Effect), :source => "SG"
+    assert_equal 0, spec[0]
+    assert_equal 1, spec[1][0]
+
+    spec = Association.headers  %w(SG TG Effect directed?), %w(Effect), :target => "TG"
+    assert_equal 0, spec[0]
+    assert_equal 1, spec[1][0]
+
+    spec = Association.headers  %w(SG TG Effect directed?), %w(Effect), :target => "TG", :source => "SG"
+    assert_equal 0, spec[0]
+    assert_equal 1, spec[1][0]
+
+    spec = Association.headers  %w(SG TG Effect directed?), %w(Effect), :source => "TG"
+    assert_equal 1, spec[0]
+    assert_equal 0, spec[1][0]
+
+    spec = Association.headers  %w(SG TG Effect directed?), %w(Effect), :target => "SG"
+    assert_equal 1, spec[0]
+    assert_equal 0, spec[1][0]
+
+    spec = Association.headers  %w(SG TG Effect directed?), %w(Effect), :target => "SG", :source => "TG"
+    assert_equal 1, spec[0]
+    assert_equal 0, spec[1][0]
+
+    spec = Association.headers  %w(SG TG Effect directed?), %w(Effect), :target => "SG", :source => "TG=~Associated Gene Name=>Ensembl Gene ID"
+    assert_equal 1, spec[0]
+    assert_equal 0, spec[1][0]
+
+    spec = Association.headers  %w(SG TG Effect directed?), %w(Effect), :target => "SG=~Associated Gene Name=>Ensembl Gene ID", :source => "TG=~Associated Gene Name=>Ensembl Gene ID"
+    assert_equal 1, spec[0]
+    assert_equal 0, spec[1][0]
+
+    spec = Association.headers  %w(SG TG Effect directed?), %w(Effect), :target => "SG=~Associated Gene Name", :source => "TG=~Associated Gene Name=>Ensembl Gene ID"
+    assert_equal 1, spec[0]
+    assert_equal 0, spec[1][0]
+
+    spec = Association.headers  %w(SG TG Effect directed?), %w(Effect), :target => "TG=~Associated Gene Name", :source => "SG=~Associated Gene Name", :target_format => "Ensembl Gene ID"
+    assert_equal 0, spec[0]
+    assert_equal "Ensembl Gene ID", spec[5]
+    assert_equal nil, spec[4]
+
+
+    spec = Association.headers  %w(SG TG Effect directed?), %w(Effect), :target => "TG=~Associated Gene Name", :source => "SG=~Associated Gene Name", :source_format => "Ensembl Gene ID"
+    assert_equal 0, spec[0]
+    assert_equal "Ensembl Gene ID", spec[4]
+    assert_equal nil, spec[5]
+  end
+end

data/test/rbbt/entity/test_identifiers.rb

@@ -0,0 +1,40 @@
+require File.join(File.expand_path(File.dirname(__FILE__)), '../..', 'test_helper.rb')
+require 'rbbt-util'
+require 'rbbt/entity'
+require 'rbbt/entity/identifiers'
+
+module Gene
+  extend Entity
+end
+
+require 'rbbt/sources/kegg'
+require 'rbbt/sources/organism'
+Gene.add_identifiers Organism.identifiers("NAMESPACE"), "Ensembl Gene ID", "Associated Gene Name"
+Gene.add_identifiers KEGG.identifiers
+Gene.add_identifiers Organism.identifiers("NAMESPACE"), "Ensembl Gene ID", "Associated Gene Name"
+Gene.add_identifiers KEGG.identifiers
+
+class TestEntityIdentifiers < Test::Unit::TestCase
+  def test_name
+    Gene.add_identifiers datafile_test('identifiers'), "Ensembl Gene ID", "Associated Gene Name"
+    assert_equal "TP53", Gene.setup("ENSG00000141510").name
+  end
+
+  def test_name_organism
+    assert_equal "TP53", Gene.setup("ENSG00000141510", "Ensembl Gene ID", "Hsa/feb2014").name
+  end
+
+  def test_identifier_files
+    assert Gene.identifier_files.any?
+  end
+
+  def test_Entity_identifier_files
+    assert Entity.identifier_files("Ensembl Gene ID").any?
+  end
+
+  def test_translate_kegg
+    assert_match "hsa", Gene.setup("ENSG00000141510", "Ensembl Gene ID", "Hsa/feb2014").to("KEGG Gene ID")
+    assert_match "TP53", Gene.setup("ENSG00000141510", "Ensembl Gene ID", "Hsa/feb2014").to("KEGG Gene ID").to(:name)
+  end
+end
+

data/test/rbbt/knowledge_base/test_query.rb

@@ -0,0 +1,45 @@
+require File.expand_path(File.dirname(__FILE__) + '../../../test_helper')
+require 'rbbt/util/tmpfile'
+require 'test/unit'
+require 'rbbt/knowledge_base'
+require 'rbbt/knowledge_base/query'
+
+class TestKnowledgeBaseQuery < Test::Unit::TestCase
+
+  EFFECT =StringIO.new <<-END
+#: :sep=" "#:type=:double
+#SG TG Effect
+MDM2 TP53 inhibition
+TP53 NFKB1|GLI1 activation|activation true|true
+  END
+
+  EFFECT_OPTIONS = {
+    :source => "SG=~Associated Gene Name",
+    :target => "TG=~Associated Gene Name=>Ensembl Gene ID",
+    :persist => false,
+    :identifiers => datafile_test('identifiers'),
+    :undirected => true,
+    :namespace => "Hsa"
+  }
+
+  EFFECT_TSV = TSV.open EFFECT, EFFECT_OPTIONS.dup 
+
+  KNOWLEDGE_BASE = KnowledgeBase.new '/tmp/kb.foo2', "Hsa"
+  KNOWLEDGE_BASE.format = {"Gene" => "Ensembl Gene ID"}
+
+  KNOWLEDGE_BASE.register :effects, EFFECT_TSV, EFFECT_OPTIONS.dup
+
+  def test_subset_all_persist
+    Misc.benchmark(1000) do
+      assert_equal 6, KNOWLEDGE_BASE.subset(:effects, :all).length
+      assert_equal 4, KNOWLEDGE_BASE.subset(:effects, :all).target_entity.uniq.length
+      assert_equal %w(Effect), KNOWLEDGE_BASE.subset(:effects, :all).info.first.keys 
+    end
+  end
+
+  def test_subset_all_persist_format
+    assert KNOWLEDGE_BASE.subset(:effects, :all).target_entity.reject{|e| e =~ /^ENS/}.empty?
+  end
+
+end
+

data/test/rbbt/knowledge_base/test_registry.rb

@@ -0,0 +1,52 @@
+require File.expand_path(File.dirname(__FILE__) + '../../../test_helper')
+require 'rbbt/util/tmpfile'
+require 'test/unit'
+require 'rbbt/knowledge_base'
+require 'rbbt/knowledge_base/registry'
+
+
+class TestKnowledgeBaseRegistry < Test::Unit::TestCase
+
+  EFFECT =StringIO.new <<-END
+#: :sep=" "#:type=:double
+#SG TG Effect
+MDM2 TP53 inhibition
+TP53 NFKB1|GLI1 activation|activation true|true
+  END
+
+  EFFECT_OPTIONS = {
+    :source => "SG=~Associated Gene Name",
+    :target => "TG=~Associated Gene Name=>Ensembl Gene ID",
+    :persist => false,
+    :identifiers => datafile_test('identifiers'),
+    :undirected => true,
+    :namespace => "Hsa"
+  }
+
+  EFFECT_TSV = TSV.open EFFECT, EFFECT_OPTIONS.dup 
+
+  KNOWLEDGE_BASE = KnowledgeBase.new '/tmp/kb.foo2'
+
+  KNOWLEDGE_BASE.register :effects, EFFECT_TSV, EFFECT_OPTIONS.dup
+
+  def test_database
+    assert_equal "Associated Gene Name", KNOWLEDGE_BASE.get_database(:effects, :source_format => "Associated Gene Name").key_field
+  end
+
+  def test_index
+    assert KNOWLEDGE_BASE.get_index(:effects, :source_format => "Associated Gene Name", :target_format => "Ensembl Gene ID", :persist => false).include? "MDM2~ENSG00000141510"
+  end
+
+  def test_index_persist
+    assert KNOWLEDGE_BASE.get_index(:effects, :source_format => "Associated Gene Name", :target_format => "Ensembl Gene ID", :persist => true).include? "MDM2~ENSG00000141510"
+  end
+
+
+  def test_index_flat
+    require 'rbbt/sources/tfacts'
+    file = TFacts.regulators
+    KNOWLEDGE_BASE.register :tfacts, file,  :type => :flat, :source => "Transcription Factor Associated Gene Name=~Associated Gene Name", :merge => true
+    assert KNOWLEDGE_BASE.subset(:tfacts, :source => ["TP53"], :target => :all).length > 10
+  end
+end
+

data/test/rbbt/test_association.rb

@@ -6,7 +6,9 @@ require 'rbbt/workflow'
 require 'rbbt/entity'
 
 require 'rbbt/association'
+require 'rbbt/association/open'
 require 'rbbt/knowledge_base'
+require 'rbbt/entity'
 
 
 TEST_ASSOCIATIONS =<<-EOF
@@ -57,9 +59,6 @@ class TestAssociations < Test::Unit::TestCase
     assert_equal ["aa"], database["TP53"].first
   end
 
-  def test_reciprocal
-  end
-
 
   def __test_ICGC
     assoc = Association.open(Open.open('ftp://data.dcc.icgc.org/current/Chronic_Lymphocytic_Leukemia-ISC_MICINN-ES/simple_somatic_mutation.CLLE-ES.tsv.gz'),
@@ -100,4 +99,5 @@ class TestAssociations < Test::Unit::TestCase
     assert_equal [tp53.ensembl], database["aa"].first
   end
 
+
 end