rbbt-util 5.28.9 → 5.29.0

data/share/rbbt_commands/workflow/prov

@@ -24,6 +24,7 @@ $ rbbt workflow prov <job-result>
 -i--inputs* List of inputs to print
 -if--info_fields* List of info fields to print
 -t--touch Update modification times to be consistent
+-e--expand_repeats Show all the dependency tree even if reapeated dependencies have already been seen before
 EOF
 
 SOPT.usage if options[:help]
@@ -130,6 +131,6 @@ if options[:plot]
   end
 
 else
-  puts Step.prov_report(step).strip
+  puts Step.prov_report(step, 0, nil, [], options[:expand_repeats]).strip
 end
 

data/test/rbbt/association/test_index.rb

@@ -73,21 +73,21 @@ TP53 NFKB1|GLI1 activation|activation true|true
 
   def test_index_flat
     require 'rbbt/sources/tfacts'
-    file = TFacts.regulators
+    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
 
   def test_index_flat_to_matrix
     require 'rbbt/sources/tfacts'
-    file = TFacts.regulators
+    file = TFactS.regulators
     tsv = Association.index(file,  :type => :flat, :source => "Transcription Factor Associated Gene Name=~Associated Gene Name", :merge => true)
     assert(tsv.to_matrix(false))
   end
 
   def test_filter_no_block
     require 'rbbt/sources/tfacts'
-    file = TFacts.regulators
+    file = TFactS.regulators
     tsv = Association.index(EFFECT, EFFECT_OPTIONS.merge(:undirected => false, :source => "SG=~Associated Gene Name", :target => "TG=~Associated Gene Name"), :persist => true)
     tsv.unnamed = false
     matches = tsv.filter :directed?
@@ -96,7 +96,7 @@ TP53 NFKB1|GLI1 activation|activation true|true
 
   def test_filter_no_block_value
     require 'rbbt/sources/tfacts'
-    file = TFacts.regulators
+    file = TFactS.regulators
     tsv = Association.index(EFFECT, EFFECT_OPTIONS.merge(:undirected => false, :source => "SG=~Associated Gene Name", :target => "TG=~Associated Gene Name"), :persist => true)
     tsv.unnamed = false
     matches = tsv.filter :Effect, "inhibition"
@@ -105,7 +105,7 @@ TP53 NFKB1|GLI1 activation|activation true|true
 
   def test_filter_block_value_field
     require 'rbbt/sources/tfacts'
-    file = TFacts.regulators
+    file = TFactS.regulators
     tsv = Association.index(EFFECT, EFFECT_OPTIONS.merge(:undirected => false, :source => "SG=~Associated Gene Name", :target => "TG=~Associated Gene Name"), :persist => true)
     tsv.unnamed = false
     matches = tsv.filter :Effect do |value|
@@ -116,7 +116,7 @@ TP53 NFKB1|GLI1 activation|activation true|true
 
   def test_filter_block_no_value_field
     require 'rbbt/sources/tfacts'
-    file = TFacts.regulators
+    file = TFactS.regulators
     tsv = Association.index(EFFECT, EFFECT_OPTIONS.merge(:undirected => false, :source => "SG=~Associated Gene Name", :target => "TG=~Associated Gene Name"), :persist => true)
     tsv.unnamed = false
     matches = tsv.filter do |key,values|

data/test/rbbt/knowledge_base/test_query.rb

@@ -22,7 +22,7 @@ TP53 NFKB1|GLI1 activation|activation true|true
     :namespace => "Hsa"
   }
 
-  EFFECT_TSV = TSV.open EFFECT, EFFECT_OPTIONS.dup 
+  EFFECT_TSV = TSV.open EFFECT, EFFECT_OPTIONS.dup
 
   KNOWLEDGE_BASE = KnowledgeBase.new Rbbt.tmp.test.kb_foo2, "Hsa"
   KNOWLEDGE_BASE.format = {"Gene" => "Ensembl Gene ID"}
@@ -34,11 +34,11 @@ TP53 NFKB1|GLI1 activation|activation true|true
       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 
+      assert_equal %w(Effect), KNOWLEDGE_BASE.subset(:effects, :all).info.first.keys
     end
   end
 
-  def test_subset_all_persist_format
+  def _test_subset_all_persist_format
     assert KNOWLEDGE_BASE.subset(:effects, :all).target_entity.reject{|e| e =~ /^ENS/}.empty?
   end
 

data/test/rbbt/knowledge_base/test_registry.rb

@@ -45,7 +45,7 @@ TP53 NFKB1|GLI1 activation|activation true|true
 
   def test_index_flat
     require 'rbbt/sources/tfacts'
-    file = TFacts.regulators
+    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

data/test/rbbt/persist/tsv/test_cdb.rb

@@ -4,19 +4,12 @@ require 'rbbt/annotations'
 require 'rbbt/util/tmpfile'
 require 'test/unit'
 
-module TestAnnotation
-  extend Annotation
-
-  self.annotation :test_annotation
-end
-
 class TestPersistTSVCDB < Test::Unit::TestCase
 
   if Persist.respond_to? :open_cbd
     def test_organism
       require 'rbbt/sources/organism'
       TmpFile.with_file nil do |tmp_file|
-        file = CMD.cmd("head -n 1000000", :in => Organism.identifiers("Hsa").open, :pipe => true)
         tsv = Organism.identifiers("Hsa").tsv(:key_field => "Associated Gene Name", :fields => ["Ensembl Gene ID"], :type => :single, :persist => true, :persist_engine => "CDB", :persist_dir => tmp_file)
         assert_equal "ENSG00000141510", tsv["TP53"]
       end

data/test/rbbt/persist/tsv/test_kyotocabinet.rb

@@ -5,15 +5,9 @@ require 'rbbt/annotations'
 require 'rbbt/util/tmpfile'
 require 'test/unit'
 
-module TestAnnotation
-  extend Annotation
-
-  self.annotation :test_annotation
-end
-
 class TestPersistTSVKC < Test::Unit::TestCase
   if Persist.respond_to? :open_kyotocabinet
-    def _test_organism_kch
+    def test_organism_kch
       require 'rbbt/sources/organism'
       TmpFile.with_file do |tmp_file|
         tsv = Organism.identifiers("Hsa").tsv :key_field => "Associated Gene Name", :fields => ["Ensembl Gene ID"], :type => :single, :persist => true, :persist_engine => "kch", :persist_dir => tmp_file
@@ -22,7 +16,7 @@ class TestPersistTSVKC < Test::Unit::TestCase
     end
 
 
-    def _test_organism_kct
+    def test_organism_kct
       require 'rbbt/sources/organism'
       TmpFile.with_file do |tmp_file|
         tsv = Organism.identifiers("Hsa").tsv :key_field => "Associated Gene Name", :fields => ["Ensembl Gene ID"], :type => :single, :persist => true, :persist_engine => "kct", :persist_dir => tmp_file

data/test/rbbt/persist/tsv/test_leveldb.rb

@@ -4,12 +4,6 @@ require 'rbbt/annotations'
 require 'rbbt/util/tmpfile'
 require 'test/unit'
 
-module TestAnnotation
-  extend Annotation
-
-  self.annotation :test_annotation
-end
-
 class TestPersistTSVLevelDB < Test::Unit::TestCase
 
   if Persist.respond_to? :open_leveldb

data/test/rbbt/persist/tsv/test_lmdb.rb

@@ -4,12 +4,6 @@ require 'rbbt/annotations'
 require 'rbbt/util/tmpfile'
 require 'test/unit'
 
-module TestAnnotation
-  extend Annotation
-
-  self.annotation :test_annotation
-end
-
 class TestPersistTSVLMDB < Test::Unit::TestCase
 
   def test_organism

data/test/rbbt/persist/tsv/test_tokyocabinet.rb

@@ -4,7 +4,8 @@ require 'rbbt/annotations'
 require 'rbbt/util/tmpfile'
 require 'test/unit'
 
-module TestAnnotation
+
+module TestAnnotationModule
   extend Annotation
 
   self.annotation :test_annotation
@@ -25,8 +26,8 @@ class TestPersistTSVTC < Test::Unit::TestCase
       entity1 = "Entity 1"
       entity2 = "Entity 2"
 
-      TestAnnotation.setup(entity1, :test_annotation => "1")
-      TestAnnotation.setup(entity2, :test_annotation => "2")
+      TestAnnotationModule.setup(entity1, :test_annotation => "1")
+      TestAnnotationModule.setup(entity2, :test_annotation => "2")
 
       annotations = [entity1, entity2]
 
@@ -56,9 +57,9 @@ class TestPersistTSVTC < Test::Unit::TestCase
       entity2 = "Entity 2"
       entity2bis = "Entity 2"
 
-      TestAnnotation.setup(entity1, :test_annotation => "1")
-      TestAnnotation.setup(entity2, :test_annotation => "2")
-      TestAnnotation.setup(entity2bis, :test_annotation => "2")
+      TestAnnotationModule.setup(entity1, :test_annotation => "1")
+      TestAnnotationModule.setup(entity2, :test_annotation => "2")
+      TestAnnotationModule.setup(entity2bis, :test_annotation => "2")
 
       annotations = [entity1, entity2, entity2bis]
 
@@ -105,8 +106,8 @@ class TestPersistTSVTC < Test::Unit::TestCase
       entity1 = "Entity 1"
       entity2 = "Entity 2"
 
-      TestAnnotation.setup(entity1, :test_annotation => "1")
-      TestAnnotation.setup(entity2, :test_annotation => "2")
+      TestAnnotationModule.setup(entity1, :test_annotation => "1")
+      TestAnnotationModule.setup(entity2, :test_annotation => "2")
 
       annotations = [entity1, entity2]
 
@@ -138,7 +139,7 @@ class TestPersistTSVTC < Test::Unit::TestCase
       entity2 = "Entity 2"
 
       annotations = [entity1, entity2]
-      TestAnnotation.setup(annotations, :test_annotation => "1")
+      TestAnnotationModule.setup(annotations, :test_annotation => "1")
       annotations.extend AnnotatedArray
 
       persisted_annotations = Persist.persist("Test", :annotations, :annotation_repo => repo) do
@@ -171,11 +172,11 @@ class TestPersistTSVTC < Test::Unit::TestCase
       entity2 = "Entity 2"
 
       annotations = [entity1, entity2]
-      TestAnnotation.setup(annotations, :test_annotation => "1")
+      TestAnnotationModule.setup(annotations, :test_annotation => "1")
       annotations.extend AnnotatedArray
 
       annotations_ary = [annotations]
-      TestAnnotation.setup(annotations_ary, :test_annotation => "1")
+      TestAnnotationModule.setup(annotations_ary, :test_annotation => "1")
       annotations_ary.extend AnnotatedArray
 
       persisted_annotations = Persist.persist("Test", :annotations, :annotation_repo => repo) do
@@ -210,9 +211,9 @@ class TestPersistTSVTC < Test::Unit::TestCase
       entity2 = "Entity 2"
       entity2bis = "Entity 2"
 
-      TestAnnotation.setup(entity1, :test_annotation => "1")
-      TestAnnotation.setup(entity2, :test_annotation => "2")
-      TestAnnotation.setup(entity2bis, :test_annotation => "2")
+      TestAnnotationModule.setup(entity1, :test_annotation => "1")
+      TestAnnotationModule.setup(entity2, :test_annotation => "2")
+      TestAnnotationModule.setup(entity2bis, :test_annotation => "2")
 
       annotations = [entity1, entity2, entity2bis]
 

data/test/rbbt/test_entity.rb

@@ -199,7 +199,6 @@ class TestEntity < Test::Unit::TestCase
 
 
     $processed_multiple = []
-    Log.severity = 0
 
     array = ReversableString.setup([string1, string2])
     assert_equal [string1, string2].collect{|s| s.chars}, array.multiple_annotation_list

data/test/rbbt/test_knowledge_base.rb

@@ -62,7 +62,7 @@ class TestKnowledgeBase < Test::Unit::TestCase
       kb = KnowledgeBase.new tmpdir, Organism.default_code("Hsa")
       kb.format = {"Gene" => "Ensembl Gene ID"}
 
-      kb.register :tfacts, TFacts.regulators, :source =>"=~Associated Gene Name"
+      kb.register :tfacts, TFactS.regulators, :source =>"=~Associated Gene Name"
 
       kb.get_index(:tfacts).reverse
     end
@@ -74,7 +74,7 @@ class TestKnowledgeBase < Test::Unit::TestCase
       kb = KnowledgeBase.new tmpdir, Organism.default_code("Hsa")
       kb.format = {"Gene" => "Ensembl Gene ID"}
 
-      kb.register :tfacts, TFacts.regulators, :source =>"=~Associated Gene Name"
+      kb.register :tfacts, TFactS.regulators, :source =>"=~Associated Gene Name"
 
       assert_equal "Ensembl Gene ID", kb.get_database(:tfacts).key_field
 
@@ -113,10 +113,9 @@ class TestKnowledgeBase < Test::Unit::TestCase
 
   def test_knowledge_base_reuse
     organism = Organism.default_code("Hsa")
-    Log.severity = 0
     TmpFile.with_file do |tmpdir|
       Path.setup(tmpdir)
-      Association.index(TFacts.regulators, :persist_file => tmpdir.tfacts, :format => {"Gene" => "Ensembl Gene ID"}, :namespace => Organism.default_code("Hsa"))
+      Association.index(TFactS.regulators, :persist_file => tmpdir.tfacts, :format => {"Gene" => "Ensembl Gene ID"}, :namespace => Organism.default_code("Hsa"))
 
       kb = KnowledgeBase.load(tmpdir)
       assert kb.identify_source('tfacts', "TP53") =~ /ENSG/

data/test/rbbt/test_persist.rb

@@ -4,12 +4,6 @@ require 'rbbt/annotations'
 require 'rbbt/util/tmpfile'
 require 'test/unit'
 
-module TestAnnotation
-  extend Annotation
-
-  self.annotation :test_annotation
-end
-
 class TestPersist < Test::Unit::TestCase
 
   def test_array_persist
@@ -78,4 +72,14 @@ class TestPersist < Test::Unit::TestCase
       stream.join
     end
   end
+
+  def test_newer
+    TmpFile.with_file("Test1") do |tmp1|
+      sleep 1
+      TmpFile.with_file("Test1") do |tmp2|
+        assert Persist.newer?(tmp1, tmp2)
+        assert ! Persist.newer?(tmp2, tmp1)
+      end
+    end
+  end
 end

data/test/rbbt/test_workflow.rb

@@ -156,6 +156,15 @@ for this dependency
     Open.read(file).reverse
   end
 
+  task :create_file => :text do |file|
+    Open.write(file('a'), "A")
+    Open.write(file('b'), "B")
+    "DONE"
+  end
+
+  dep_task :reverse_step_file, TestWF, :reverse_file do |jobname, options, dependencies|
+    dep = dependencies.flatten.first
+  end
 
 end
 
@@ -204,7 +213,6 @@ class TestWorkflow < Test::Unit::TestCase
 
   def test_update_on_input_dependency_update
     Open.repository_dirs << File.join(ENV["HOME"],".rbbt/tmp/test/workflow")
-    Log.severity = 0
     Misc.with_env "RBBT_UPDATE", "true" do
       send_input_dep_to_reverse_job = TestWF.job(:send_input_dep_to_reverse, nil, :name => "Miguel")
       send_input_dep_to_reverse_job.clean
@@ -376,21 +384,23 @@ class TestWorkflow < Test::Unit::TestCase
   def test_delete_dep
     job = TestWF.job(:t3).recursive_clean
     job.run
-    assert job.checks.select{|d| d.task_name.to_s == "t1" }.any?
-    job = TestWF.job(:t3)
-    job.step(:t1).clean
-    assert job.checks.select{|d| d.task_name.to_s == "t1" }.empty?
-    job = TestWF.job(:t3).recursive_clean
-    job.run
-    assert job.checks.select{|d| d.task_name.to_s == "t1" }.any?
-    job = TestWF.job(:t3)
-    sleep 1
-    Open.touch job.step(:t1).path
-    Misc.with_env "RBBT_UPDATE", "false" do
-      assert job.updated?
-    end
-    Misc.with_env "RBBT_UPDATE", "true" do
-      assert ! job.updated?
+    Misc.with_env "RBBT_UPDATE", 'true' do
+      assert job.checks.select{|d| d.task_name.to_s == "t1" }.any?
+      job = TestWF.job(:t3)
+      job.step(:t1).clean
+      assert job.checks.select{|d| d.task_name.to_s == "t1" }.empty?
+      job = TestWF.job(:t3).recursive_clean
+      job.run
+      assert job.checks.select{|d| d.task_name.to_s == "t1" }.any?
+      job = TestWF.job(:t3)
+      sleep 1
+      Open.touch job.step(:t1).path
+      Misc.with_env "RBBT_UPDATE", "false" do
+        assert job.updated?
+      end
+      Misc.with_env "RBBT_UPDATE", "true" do
+        assert ! job.updated?
+      end
     end
   end
 
@@ -439,4 +449,27 @@ class TestWorkflow < Test::Unit::TestCase
       end
     end
   end
+
+  def test_input_step_file_check
+    job = TestWF.job(:t3).recursive_clean
+    job.run
+    Misc.with_env "RBBT_UPDATE", 'true' do
+      assert job.checks.select{|d| d.task_name.to_s == "t1" }.any?
+      job = TestWF.job(:t3)
+      job.step(:t1).clean
+      assert job.checks.select{|d| d.task_name.to_s == "t1" }.empty?
+      job = TestWF.job(:t3).recursive_clean
+      job.run
+      assert job.checks.select{|d| d.task_name.to_s == "t1" }.any?
+      job = TestWF.job(:t3)
+      sleep 1
+      Open.touch job.step(:t1).path
+      Misc.with_env "RBBT_UPDATE", "false" do
+        assert job.updated?
+      end
+      Misc.with_env "RBBT_UPDATE", "true" do
+        assert ! job.updated?
+      end
+    end
+  end
 end

data/test/rbbt/tsv/test_accessor.rb

@@ -222,6 +222,17 @@ row2    aa|aa|AA|AA    b1|b2|B1|B2    Id1|Id1|Id2|Id2
     end
   end
 
+  def test_to_s_no_fields
+    file1 =<<-EOF
+row6 dd dd ee
+row1 a b c
+row2 A B C
+row3 1 2 3
+   EOF
+    tsv1 = TSV.open StringIO.new(file1), :sep => " "
+    assert tsv1.to_s(:sort, true).include?('dd')
+  end
+
   def test_to_s_unmerge_expand
     content =<<-EOF
 #Id    ValueA    ValueB    OtherID

data/test/rbbt/tsv/test_attach.rb

@@ -566,7 +566,7 @@ row2,CC
       tsv3.keys.each{|k| tsv3[k] = nil if tsv3[k] == ""}
     end
 
-    assert_equal tsv1.attach(tsv2, :complete => true).attach(tsv3, :complete => true)["row1"], [nil, "B", nil]
+    assert_equal [nil, "B", nil], tsv1.attach(tsv2, :complete => true).attach(tsv3, :complete => true)["row1"]
   end
 
   def test_attach_index_both_non_key
@@ -597,10 +597,10 @@ A    Id3
 
     tsv1 = tsv2 = nil
 
-    tsv1 = Rbbt.tmp.test.test1.data.tsv :double,  :sep => /\s+/
-    tsv2 = Rbbt.tmp.test.test2.data.tsv :double,  :sep => /\s+/
+    tsv1 = Rbbt.tmp.test.test1.data.produce(true).tsv :double,  :sep => /\s+/
+    tsv2 = Rbbt.tmp.test.test2.data.produce(true).tsv :double,  :sep => /\s+/
 
-    tsv2.identifiers = Rbbt.tmp.test.test2.identifiers.produce.find #.to_s
+    tsv2.identifiers = Rbbt.tmp.test.test2.identifiers.produce(true).produce.find #.to_s
 
     tsv1.attach tsv2, :fields => ["ValueE"] #, :persist_input => true
     Log.tsv tsv1
@@ -627,14 +627,92 @@ E    B
 
     tsv1 = tsv2 = nil
 
-    tsv1 = Rbbt.tmp.test.test1.data.tsv :double,  :sep => /\s+/
-    tsv2 = Rbbt.tmp.test.test2.data.tsv :double,  :sep => /\s+/
+    tsv1 = Rbbt.tmp.test.test1.data.produce(true).tsv :double,  :sep => /\s+/
+    tsv2 = Rbbt.tmp.test.test2.data.produce(true).tsv :double,  :sep => /\s+/
 
-    Log.severity = 0
     tsv1.attach tsv2, :fields => ["ValueE"] #, :persist_input => true
     Log.tsv tsv1
-    ppp tsv1
     
   end
+
+  def test_attach_complete
+    content1 =<<-EOF
+#: :sep=/\\s+/
+#Id    ValueA
+row1    a|aa|aaa
+row2    A
+    EOF
+
+    content2 =<<-EOF
+#: :sep=/\\s+/
+#Id    ValueB
+row1    b
+row3    C
+    EOF
+    Rbbt.claim Rbbt.tmp.test.test1.data, :string, content1
+    Rbbt.claim Rbbt.tmp.test.test2.data, :string, content2
+
+    tsv1 = tsv2 = nil
+
+    tsv1 = Rbbt.tmp.test.test1.data.produce(true).tsv :double,  :sep => /\s+/
+    tsv2 = Rbbt.tmp.test.test2.data.produce(true).tsv :double,  :sep => /\s+/
+
+    tsv1.attach tsv2, :complete => true
+    assert_equal [[], ["C"]], tsv1["row3"]
+
+    tsv1 = Rbbt.tmp.test.test1.data.produce(true).tsv :double,  :sep => /\s+/
+    tsv2 = Rbbt.tmp.test.test2.data.produce(true).tsv :double,  :sep => /\s+/
+
+    ppp tsv1.attach tsv2, :complete => ["AA"]
+    tsv1.attach tsv2, :complete => ["AA"]
+    assert_equal [["AA"], ["C"]], tsv1["row3"]
+  end
+
+  def test_attach_complete_identifiers
+    content1 =<<-EOF
+#: :sep=/\\s+/
+#Id    ValueA
+row1    a|aa|aaa
+row2    A
+    EOF
+
+    content2 =<<-EOF
+#: :sep=/\\s+/
+#Id2    ValueB
+ROW_1    b
+ROW_2    C
+    EOF
+
+    identifiers =<<-EOF
+#: :sep=/\\s+/
+#Id    Id2
+row1    ROW_1
+row2    ROW_2
+row3    ROW_3
+    EOF
+    Rbbt.claim Rbbt.tmp.test.test1.data, :string, content1
+    Rbbt.claim Rbbt.tmp.test.test2.data, :string, content2
+    Rbbt.claim Rbbt.tmp.test.identifiers.data, :string, identifiers
+
+    tsv1 = tsv2 = nil
+
+    tsv1 = Rbbt.tmp.test.test1.data.produce(true).tsv :double,  :sep => /\s+/
+    tsv2 = Rbbt.tmp.test.test2.data.produce(true).tsv :double,  :sep => /\s+/
+    ids = Rbbt.tmp.test.identifiers.data.produce(true).tsv :double,  :sep => /\s+/
+
+    tsv1.identifiers = ids
+
+    tsv1.attach tsv2
+    assert_equal [["A"], ["C"]], tsv1["row2"]
+
+    tsv1 = Rbbt.tmp.test.test1.data.produce(true).tsv :double,  :sep => /\s+/
+    tsv2 = Rbbt.tmp.test.test2.data.produce(true).tsv :double,  :sep => /\s+/
+    ids = Rbbt.tmp.test.identifiers.data.produce(true).tsv :double,  :sep => /\s+/
+
+    tsv1.identifiers = ids
+
+    tsv1.attach tsv2, :complete => true
+    assert_equal [["A"], ["C"]], tsv1["row2"]
+  end
 end