rbbt-util 5.6.10 → 5.6.11

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: b271b167bce0dcfd0cdd1c72e569633f410f02a0
-  data.tar.gz: a0ded4de2b73712c66255e3ecaf2ee18198d4178
+  metadata.gz: 1c30ab5dc60391c384494c4607cc74c765d507b5
+  data.tar.gz: 8a599300be751940d1feb9b11a14fc83be187722
 SHA512:
-  metadata.gz: 09962fd13e045bcf3f97b1655d41e1262e9216c454c7af0c3b207ec61f640ffac382e84dbaa995c45b25efd85bf32a56b2b3068ff75146b6125e54ac668c89ac
-  data.tar.gz: 174808a701eba99667b447e8bfee0cc5bd0fbd4f29047b53de0358d8dd9ec75130340c63ed1861f6e9a63c018cbe4b48ca38ac74a50e19139501f8f72396b797
+  metadata.gz: 05b7d25435ee522bb08c7b0d8ad5355d56d1b1d91816c191bf1a4d831ae2fb829af9b0201ae84f7544fd51f37abb72395f4cd1d88bdc432ac0688c6e6fd27de7
+  data.tar.gz: 185cfbaa226f9512d56a42c76a9fbcc2c8b24de266c7ed4a91e7d1f6055608c4a5b7e6f087ac3090a673015af29c5d93e4b19d292a04fdf00c48ab5f8b4de4cc

data/lib/rbbt/tsv/attach.rb

@@ -31,7 +31,8 @@ module TSV
 
     done = false
     Open.write(output) do |os|
-      os.puts TSV.header_lines(key_field, fields, options)
+      options.delete :sep if options[:sep] == "\t"
+      os.puts TSV.header_lines(key_field, fields, options) 
 
       while line
         key, *parts = line.sub("\n",'').split(sep, -1)

data/lib/rbbt/tsv.rb

@@ -171,33 +171,38 @@ module TSV
     end
 
     line_num = 1
-    while not line.nil? 
-      begin
-        progress_monitor.tick(stream.pos) if progress_monitor 
-
-        raise Parser::SKIP_LINE if line.empty?
-
-        line = Misc.fixutf8(line)
-        line = parser.process line
-        parts = parser.chop_line line
-        key, values = parser.get_values parts
-        values = parser.cast_values values if parser.cast?
-        parser.add_to_data data, key, values
-        line = stream.gets
-        line_num += 1
-        raise Parser::END_PARSING if head and line_num > head.to_i
-      rescue Parser::SKIP_LINE
+    begin
+      while not line.nil? 
         begin
+          progress_monitor.tick(stream.pos) if progress_monitor 
+
+          raise Parser::SKIP_LINE if line.empty?
+
+          line = Misc.fixutf8(line)
+          line = parser.process line
+          parts = parser.chop_line line
+          key, values = parser.get_values parts
+          values = parser.cast_values values if parser.cast?
+          parser.add_to_data data, key, values
           line = stream.gets
-          next
+          line_num += 1
+          raise Parser::END_PARSING if head and line_num > head.to_i
+        rescue Parser::SKIP_LINE
+          begin
+            line = stream.gets
+            next
+          rescue IOError
+            break
+          end
+        rescue Parser::END_PARSING
+          break
         rescue IOError
+          Log.exception $!
           break
         end
-      rescue Parser::END_PARSING
-        break
-      rescue IOError
-        break
       end
+    ensure
+      stream.close unless stream.closed?
     end
 
     parser.setup data

data/lib/rbbt/util/misc.rb

@@ -278,7 +278,8 @@ module Misc
         obj.each do |k,v|
           new << k.to_s << '=>' << fingerprint(v) << ' '
         end
-        new << "}"
+        new[-1] =  "}"
+        new
       end
     else
       obj.to_s

data/lib/rbbt/util/semaphore.rb

@@ -64,6 +64,7 @@ void post_semaphore(char* name){
   def self.fork_each_on_semaphore(elems, size, file = nil)
     with_semaphore(size, file) do |file|
       begin
+
         pids = elems.collect do |elem| 
           Process.fork do 
             begin
@@ -78,7 +79,13 @@ void post_semaphore(char* name){
             end
           end
         end
-        pids.each do |pid| Process.waitpid pid end
+
+        while pids.any?
+          pid = Process.wait -1
+          pids.delete pid
+        end
+        #pids.each do |pid| Process.waitpid pid end
+        
       rescue Exception
         Log.error "Killing children: #{pids.sort * ", " }"
         pids.each do |pid| begin Process.kill("INT", pid); rescue; end; end

data/test/rbbt/association/test_index.rb

@@ -3,9 +3,8 @@ require 'rbbt/workflow'
 require 'rbbt/association'
 require 'rbbt/association/index'
 
-gem 'rbbt-sources'
-Workflow.require_workflow "Genomics"
-require 'rbbt/entity/gene'
+#Workflow.require_workflow "Genomics"
+#require 'rbbt/entity/gene'
 
 class TestAssociationIndex < Test::Unit::TestCase
 
@@ -16,7 +15,7 @@ class TestAssociationIndex < Test::Unit::TestCase
   end
 
 
-  def test_subset
+  def __test_subset
     require 'rbbt/sources/pina'
     require 'rbbt/sources/kegg'
     tp53 = Gene.setup("TP53", "Associated Gene Name", "Hsa/jan2013").ensembl

data/test/rbbt/persist/test_tsv.rb

@@ -11,11 +11,8 @@ class TestPersistTSV < Test::Unit::TestCase
 
   def tsv_path
     Random.new
-    #require 'rbbt/sources/organism'
-    #Organism.identifiers("Hsa")
     require 'rbbt/workflow'
-    Workflow.require_workflow "Genomics"
-    require 'rbbt/entity/gene'
+    Workflow.require_workflow "Genomes1000"
     require 'rbbt/sources/genomes1000'
     CMD.cmd("grep -v '^\\.\t' | head -n #{MAX}", :in => Genomes1000.mutations.open, :pipe => true)
   end
@@ -26,7 +23,6 @@ class TestPersistTSV < Test::Unit::TestCase
       db= nil
       Misc.benchmark(1, "Build database with #{MAX - 2} entries") do
         db = TSV.open(file, :fields => [1], :persist => true, :persist_engine => engine, :persist_dir => tmp_file, :type => :single, :unnamed => true)
-        fff db
       end
       test = db.keys.sort{rand}[1..100000]
       Misc.benchmark(5, "Access #{test.length} random entries") do
@@ -40,32 +36,32 @@ class TestPersistTSV < Test::Unit::TestCase
     end
   end
 
-  def test_benchmark_tch
+  def _test_benchmark_tch
     engine = "HDB"
     run_bechmark(tsv_path, engine)
   end
 
-  def test_benchmark_tcb
+  def _test_benchmark_tcb
     engine = "BDB"
     run_bechmark(tsv_path, engine)
   end
 
-  def test_benchmark_kch
+  def _test_benchmark_kch
     engine = "kch"
     run_bechmark(tsv_path, engine)
   end
 
-  def test_benchmark_kcb
+  def _test_benchmark_kcb
     engine = "kct"
     run_bechmark(tsv_path, engine)
   end
 
-  def test_benchmark_cdb
+  def _test_benchmark_cdb
     engine = "CDB"
     run_bechmark(tsv_path, engine)
   end
 
-  def test_benchmark_leveldb
+  def _test_benchmark_leveldb
     engine = "LevelDB"
     run_bechmark(tsv_path, engine)
   end

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

@@ -5,8 +5,6 @@ require 'rbbt/annotations'
 require 'rbbt/util/tmpfile'
 require 'test/unit'
 
-require 'rbbt/sources/organism'
-
 module TestAnnotation
   extend Annotation
 
@@ -23,7 +21,7 @@ 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/test_annotations.rb

@@ -17,7 +17,6 @@ module AnnotatedString2
   include AnnotatedString
   self.annotation :annotation_str2
 end
-puts AnnotatedString.annotations.inspect
 
 class TestAnnotations < Test::Unit::TestCase
 

data/test/rbbt/test_association.rb

@@ -8,13 +8,6 @@ require 'rbbt/entity'
 require 'rbbt/association'
 require 'rbbt/knowledge_base'
 
-gem 'rbbt-sources'
-gem 'rbbt-phgx'
-
-Log.severity=0
-Workflow.require_workflow "Genomics"
-require 'rbbt/entity/gene'
-require 'rbbt/sources/pina'
 
 TEST_ASSOCIATIONS =<<-EOF
 #: :sep=" "#:namespace=Hsa/jan2013
@@ -42,6 +35,7 @@ class TestAssociations < Test::Unit::TestCase
 
   def test_simple_open
     database = Association.open(FAssocs, {}, :dir => DAssocs)
+    database.unnamed = false
     assert_equal ["C", "K"], database["c"]["Entity2"]
   end
  
@@ -57,7 +51,24 @@ class TestAssociations < Test::Unit::TestCase
     assert_equal ["kk", "c", "4", "PTEN"], database["K"].flatten
   end
 
-  def test_target_open
+
+  def test_gene_open
+    database = Association.open(FAssocs, {:source => "Gene=~Associated Gene Name", :target => "Entity3", :zipped => true}, :dir => DAssocs)
+    assert_equal ["aa"], database["TP53"].first
+  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'),
+                         { :source => "gene_affected=~Ensembl Gene ID=>Associated Gene Name", :target => "icgc_donor_id=~Sample", 
+                           :fields => ['consequence_type'],  
+                           :namespace => 'Hsa/jan2013',
+                           :merge => true, :header_hash=>''}, :persist => false)
+
+    assert_equal 9, assoc["SF3B1"]["Sample"].uniq.length
+  end
+
+  def __test_target_open
 
     database = Association.open(Pina.protein_protein,{ 
                                 :undirected => false, 
@@ -74,34 +85,16 @@ class TestAssociations < Test::Unit::TestCase
     assert database["Q13547"][0].include? "O15379"
   end
 
-
-  def test_gene_open
-    database = Association.open(FAssocs, {:source => "Gene=~Associated Gene Name", :target => "Entity3", :zipped => true}, :dir => DAssocs)
-    assert_equal ["aa"], database["TP53"].first
-  end
-
-  def test_gene_open_translate
+  def __test_gene_open_translate
     tp53 = Gene.setup("TP53", "Associated Gene Name", "Hsa/jan2013")
     database = Association.open(FAssocs, {:source => "Gene=~Associated Gene Name", :source_format => "Ensembl Gene ID", :target => "Entity3", :zipped => true}, :dir => DAssocs)
     assert_equal ["aa"], database[tp53.ensembl].first
   end
 
-  def test_gene_target_open_translate
+  def __test_gene_target_open_translate
     tp53 = Gene.setup("TP53", "Associated Gene Name", "Hsa/jan2013")
     database = Association.open(FAssocs, {:target => "Gene=~Associated Gene Name=>Ensembl Gene ID", :source => "Entity3", :zipped => true}, :dir => DAssocs)
     assert_equal [tp53.ensembl], database["aa"].first
   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'),
-                         { :source => "gene_affected=~Ensembl Gene ID=>Associated Gene Name", :target => "icgc_donor_id=~Sample", 
-                           :fields => ['consequence_type'],  
-                           :namespace => 'Hsa/jan2013',
-                           :merge => true, :header_hash=>''}, :persist => false)
-
-    assert_equal 9, assoc["SF3B1"]["Sample"].uniq.length
-
-  end
-
-
 end

data/test/rbbt/test_entity.rb

@@ -167,55 +167,12 @@ class TestEntity < Test::Unit::TestCase
     assert_equal string.length, string.annotation_list.length
   end
 
-  def __test_performance
-    require 'rbbt/entity/gene'
-    Misc.profile(:min_percent => 2) do
-      1000.times do
-        TestA.new "foo", "bar"
-      end
-    end
-
-    Misc.profile(:min_percent => 2) do
-      1000.times do
-        Gene.setup_positional("", "foo", "bar")
-      end
-    end
-
-    Misc.benchmark(100000) do
-      Gene.setup("", :foo => "foo", :bar => "bar")
-    end
-
-    Misc.benchmark(100000) do
-      TestA.new "foo", "bar"
-    end
-  end
-
-  def __test_clean_annotations
-    Workflow.require_workflow "StudyExplorer"
-
-    s = Study.setup("CLL")
-    mutations = s.cohort.metagenotype
+  def test_clean_annotations
 
-    mis = mutations.mutated_isoforms.compact.flatten
+    string = "test_string"
+    ReversableString.setup string
+    assert string.respond_to?(:reverse_text_single)
+    assert ! string.clean_annotations.respond_to?(:reverse_text_single)
 
-    Misc.profile(:min_percent => 1) do
-      mis.each{|m| m}
-    end
-
-
-    Misc.benchmark(10) do
-      mis.each{|m| m}
-    end
-
-    Misc.benchmark(10) do
-      mis.clean_annotations.each{|m| m}
-    end
-
-    m = mutations.first
-
-    assert_equal m.split(":")[1], m.position.to_s
-    assert_raise NoMethodError do 
-      m.clean_annotations.position
-    end
   end
 end

data/test/rbbt/test_knowledge_base.rb

@@ -3,17 +3,15 @@ $LOAD_PATH.unshift(File.dirname(__FILE__))
 require 'test/unit'
 require 'rbbt/knowledge_base'
 
-gem 'rbbt-sources'
-gem 'rbbt-phgx'
-gem 'rest-client'
-
 require 'rbbt/sources/pina'
 require 'rbbt/workflow'
-Workflow.require_workflow "Genomics"
-require 'genomics_kb'
 
 class TestKnowledgeBase < Test::Unit::TestCase
   def setup
+    if not defined? Genomics
+      Workflow.require_workflow "Genomics"
+      require 'genomics_kb'
+    end
     KnowledgeBase.knowledge_base_dir = Rbbt.tmp.knowledge_base_test.find
     @kb = Genomics.knowledge_base
   end

data/test/rbbt/tsv/test_accessor.rb

@@ -57,7 +57,6 @@ row2    A    B    Id3
 
     TmpFile.with_file(content) do |filename|
       tsv = TSV.open(filename, :sep => /\s+/)
-      puts tsv.to_s
       assert tsv.to_s =~ /row1\ta|aa|aaa/
       assert tsv.to_s =~ /:type=:double/
     end

data/test/rbbt/tsv/test_attach.rb

@@ -469,7 +469,6 @@ A    Id3
     tsv1 = tsv2 = index = nil
     TmpFile.with_file(content1) do |filename|
       tsv1 = TSV.open(File.open(filename), :flat, :fields => ["ValueA"], :sep => /\s+/)
-      puts tsv1
     end
 
     TmpFile.with_file(content2) do |filename|
@@ -477,8 +476,7 @@ A    Id3
     end
 
     res = tsv1.attach tsv2, :fields => ["OtherID"]
-    puts res
-
+    assert_equal "Id3", res["row2"]["OtherID"]
   end
 end
 

data/test/rbbt/tsv/test_change_id.rb

@@ -38,6 +38,6 @@ aa xx
 
   def test_change_id
     @f1.identifiers = @id
-    puts @f1.change_key "X"
+    assert @f1.change_key("X").include? "x"
   end
 end

data/test/rbbt/tsv/test_field_index.rb

@@ -4,7 +4,7 @@ require 'rbbt/tsv/field_index'
 
 class TestTSVFieldIndex < Test::Unit::TestCase
 
-  def test_zipped
+  def test_field_index
     content =<<-EOF
 #Id    ValueA    ValueB ValueC
 rowA    A|AA    B|BB  C|CC
@@ -13,7 +13,7 @@ rowa    a|aa    b|BB  C|CC
 
     TmpFile.with_file(content) do |filename|
       tsv = TSV.open(File.open(filename), :sep => /\s+/, :type => :double)
-      puts tsv.field_index("ValueA")
+      assert_equal ["rowA"], tsv.field_index("ValueA")["A"]
     end
   end
 end

data/test/rbbt/tsv/test_filter.rb

@@ -131,7 +131,6 @@ row2    A    B
       tsv.filter
       tsv.add_filter "field:ValueA", ["A"], Rbbt.tmp.test.Filter.fieldValueA.find(:user)
 
-      puts tsv
       elem = []
       tsv.through do |k,v| elem << k end
       assert_equal 1, elem.size

data/test/rbbt/util/test_R.rb

@@ -2,7 +2,7 @@ require File.join(File.expand_path(File.dirname(__FILE__)), '../..', 'test_helpe
 require 'rbbt/util/R'
 
 class TestR < Test::Unit::TestCase
-  def test_sum
+  def _test_sum
     assert_equal "6", R.run('cat(3+3)').read.split(/\n/).last
   end
 
@@ -16,9 +16,7 @@ data = data + 1
 
   def test_format_tsv
     tsv = TSV.setup({"a" => [1], "b" => [2]}, :type => :list, :key_field => "Letter", :fields => ["Number"])
-    puts tsv.transpose "Field"
-    tsv.unnamed = true
-    puts R.ruby2R tsv
+    assert tsv.transpose("Field").include? "Number"
   end
 end
 

data/test/rbbt/util/test_log.rb

@@ -3,7 +3,7 @@ require 'rbbt/util/log'
 
 class TestLog < Test::Unit::TestCase
   def test_color
-    puts Log.color(:green, "green")
+    assert Log.color(:green, "green")
   end
 end
 

data/test/rbbt/util/test_misc.rb

@@ -6,9 +6,6 @@ require 'rbbt/entity'
 
 class TestMisc < Test::Unit::TestCase
 
-  def test_humanize
-    assert_equal "mutation_enrichment", Misc.humanize("MutationEnrichment")
-  end
 
   def test_fixutf8
     string = "abc\xffdef"
@@ -84,13 +81,6 @@ class TestMisc < Test::Unit::TestCase
     assert_equal('d', Misc.string2hash("a=b#c=d#:h='j'")["c"])
     assert_equal('j', Misc.string2hash("a=b#c=d#:h='j'")[:h])
     assert_equal(:j, Misc.string2hash("a=b#c=d#:h=:j")[:h])
-
-    Misc.benchmark(1000) do
-      Misc.string2hash_old("a=b#c=d#:h=:j#test='1'#:num=1#:sep=/#/")
-    end
-    Misc.benchmark(1000) do
-      Misc.string2hash("a=b#c=d#:h=:j#test='1'#:num=1#:sep=/#/")
-    end
   end
   
   def test_named_array
@@ -227,6 +217,8 @@ class TestMisc < Test::Unit::TestCase
     assert_equal "Test string", Misc.humanize(str1)
     assert_equal "TEST string", Misc.humanize(str2)
     assert_equal "Test", Misc.humanize(str3)
+
+    assert_equal "mutation_enrichment", Misc.snake_case("MutationEnrichment")
   end
 
   def test_snake_case
@@ -241,7 +233,7 @@ class TestMisc < Test::Unit::TestCase
   end
 
   def test_fingerprint
-    puts Misc.fingerprint({:a => 1})
+    assert_equal '{a=>1}', Misc.fingerprint({:a => 1})
   end
 
   def test_tarize
@@ -266,7 +258,6 @@ class TestMisc < Test::Unit::TestCase
     t = 5
     stream = Misc.open_pipe do |sin|
       t.times do |i|
-        puts "Calculating line #{ i }"
         sleep 0.5
         sin.puts "LINE #{ i }"
       end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rbbt-util
 version: !ruby/object:Gem::Version
-  version: 5.6.10
+  version: 5.6.11
 platform: ruby
 authors:
 - Miguel Vazquez
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-02-25 00:00:00.000000000 Z
+date: 2014-02-26 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake
@@ -108,20 +108,6 @@ dependencies:
     - - ">="
       - !ruby/object:Gem::Version
         version: '0'
-- !ruby/object:Gem::Dependency
-  name: ZenTest
-  requirement: !ruby/object:Gem::Requirement
-    requirements:
-    - - '='
-      - !ruby/object:Gem::Version
-        version: '4.3'
-  type: :runtime
-  prerelease: false
-  version_requirements: !ruby/object:Gem::Requirement
-    requirements:
-    - - '='
-      - !ruby/object:Gem::Version
-        version: '4.3'
 description: Utilities for handling tsv files, caches, etc
 email: miguel.vazquez@cnio.es
 executables: