rbbt-util 5.19.18 → 5.19.19

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 64ccfc0af909dbc006460786c31cd03ecb5a09d4
-  data.tar.gz: 221e3508c946239991fbb3fa8e4f9a4649cc954a
+  metadata.gz: a1c106d7f3ec1535abec1883fd465cee9c60ce1f
+  data.tar.gz: 31d9d1fc96bf50f6fa6c9f5b566d82fb10bc9f06
 SHA512:
-  metadata.gz: 585970e6c3d50f7edd9b9188e5e45a7938b821bd0d6ed69b03b08b8fd6ee893cdff167d7804005ea96c7e5f5f3dea68445e4755d6be8724b3a8d5e03e31e94e3
-  data.tar.gz: 3f3d63792bdd57aee6b3e7b863c7af2ed65e77706edcc55b1fda8b185dc0207bfd2e193b197d67f91bbf5afc7c8fbf0bc384654f4aeb2c510f5fde11d49e0ce5
+  metadata.gz: b7f0d24616ceed78fa2671f40ce3e0a9221b6739a0a65f09783dece72d74b661ca54f480b930059bda643c446f1a28e9c81b9deba534fe09ae23dd0a8f8955f8
+  data.tar.gz: 0802e21fe07e9f3fedfe244c3456106edfb74733cfbad4e092faea47e1734faa212605169222e7bccadc68d05c7f9a3794912af4c294ff462d1a246bf8db8f7b

data/lib/rbbt/association/index.rb

@@ -29,7 +29,7 @@ module Association
 
       key_field = [source_field, target_field, undirected ? "undirected" : nil].compact * "~"
 
-      TSV.setup(data, :key_field => key_field, :fields => fields[1..-1], :type => :list, :serializer => :list)
+      TSV.setup(data, :key_field => key_field, :fields => fields[1..-1], :type => :list, :serializer => :list, :namespace => database.namespace)
 
       data.key_field = key_field
       data.fields = fields[1..-1]
@@ -69,6 +69,8 @@ module Association
 
             annotations.each do |target, info|
               next if target.nil? or target.empty?
+              source.gsub!('~','-..-')
+              target.gsub!('~','-..-')
               key = [source, target] * "~"
 
               if data[key].nil? or info.nil?

data/lib/rbbt/association/item.rb

@@ -83,7 +83,8 @@ module AssociationItem
   end
   property :value => :array2single do
     index = index(database)
-    value = (reverse ? index.reverse : index).chunked_values_at self
+    #value = (reverse ? index.reverse : index).chunked_values_at self
+    value = index.chunked_values_at self
     value.collect{|v| NamedArray.setup(v, index.fields)}
   end
 

data/lib/rbbt/knowledge_base/entity.rb

@@ -23,6 +23,8 @@ class KnowledgeBase
     IndiferentHash.setup entity_options if entity_options and not IndiferentHash === entity_options
     options = entity_options[type.to_s] || entity_options[Entity.formats[type.to_s].to_s] || {}
     options[:format] = @format[type] if @format.include? :type
+    namespace = self.namespace
+    namespace = db_namespace(database_name) if namespace.nil? and database_name
     options = {:organism => namespace}.merge(options)
     if database_name  
       database = get_database(database_name)
@@ -67,12 +69,17 @@ class KnowledgeBase
     get_database(name).identifier_files.dup
   end
 
+  def db_namespace(name)
+    get_database(name).namespace
+  end
+
   def source_index(name)
     Persist.memory("Source index #{name}: KB directory #{dir}") do
       identifier_files = identifier_files(name)
       identifier_files.concat Entity.identifier_files(source(name)) if defined? Entity
       identifier_files.uniq!
       identifier_files.collect!{|f| f.annotate(f.gsub(/\bNAMESPACE\b/, namespace))} if namespace
+      identifier_files.collect!{|f| f.annotate(f.gsub(/\bNAMESPACE\b/, db_namespace(name)))} if not namespace and db_namespace(name)
       identifier_files.reject!{|f| f.match(/\bNAMESPACE\b/)}
       TSV.translation_index identifier_files, source(name), nil, :persist => true
     end
@@ -83,7 +90,8 @@ class KnowledgeBase
       identifier_files = identifier_files(name)
       identifier_files.concat Entity.identifier_files(target(name)) if defined? Entity
       identifier_files.uniq!
-      identifier_files.collect!{|f| f.annotate(f.gsub(/\bNAMESPACE\b/, namespace))} if namespace
+      identifier_files.collect!{|f| f.annotate(f.gsub(/\bNAMESPACE\b/, namespace))} if self.namespace
+      identifier_files.collect!{|f| f.annotate(f.gsub(/\bNAMESPACE\b/, db_namespace(name)))} if namespace.nil? and db_namespace(name)
       identifier_files.reject!{|f| f.match(/\bNAMESPACE\b/)}
       TSV.translation_index identifier_files, target(name), nil, :persist => true
     end

data/lib/rbbt/knowledge_base/query.rb

@@ -60,7 +60,9 @@ class KnowledgeBase
 
   def parents(name, entity)
     entity = identify_target(name, entity)
-    setup(name, _parents(name, entity), true)
+    matches = _parents(name, entity)
+    matches.each{|m| m.replace(m.partition("~").reverse*"") } unless undirected(name)
+    setup(name, matches, true)
   end
 
   def _neighbours(name, entity)

data/lib/rbbt/knowledge_base/registry.rb

@@ -9,7 +9,7 @@ class KnowledgeBase
       Log.debug("Registering #{ name } from code block")
       @registry[name] = [block, options]
     else
-      Log.debug("Registering #{ name }: #{ Misc.fingerprint file }")
+      Log.debug("Registering #{ name }: #{ Misc.fingerprint file } #{Misc.fingerprint options}")
       @registry[name] = [file, options]
     end
   end
@@ -40,20 +40,26 @@ class KnowledgeBase
 
   def get_index(name, options = {})
     name = name.to_s
-    options[:organism] ||= options[:namespace] ||= self.namespace
     @indices[[name, options]] ||= 
       begin 
-        fp = Misc.hash2md5(options)
-        key = name.to_s + "_" + fp
+        if options.empty?
+          key = name.to_s
+        else
+          fp = Misc.hash2md5(options)
+          key = name.to_s + "_" + fp
+        end
 
         Persist.memory("Index:" << [key, dir] * "@") do
           options = options.dup
+
+          options[:organism] ||= options[:namespace] ||= self.namespace unless self.namespace.nil?
+
           persist_dir = dir
           persist_file = persist_dir[key].find
           file, registered_options = registry[name]
 
           options = Misc.add_defaults options, registered_options if registered_options and registered_options.any?
-          options = Misc.add_defaults options, :persist_file => persist_file, :persist_dir => persist_dir, :namespace => namespace, :format => format, :persist => true
+          options = Misc.add_defaults options, :persist_file => persist_file, :persist_dir => persist_dir, :format => format, :persist => true
 
           if entity_options
             options[:entity_options] ||= {}
@@ -75,7 +81,7 @@ class KnowledgeBase
                     Association.index(file, options, persist_options.dup)
                   end
 
-          index.namespace = self.namespace
+          index.namespace = self.namespace unless self.namespace
 
           index
         end
@@ -84,19 +90,30 @@ class KnowledgeBase
 
   def get_database(name, options = {})
     name = name.to_s
-    options[:organism] ||= options[:namespace] ||= self.namespace
+
     @databases[[name, options]] ||= 
       begin 
         fp = Misc.fingerprint([name,options])
-        key = name.to_s + "_" + Misc.digest(fp) + '.database'
+
+        if options.empty?
+          key = name.to_s
+        else
+          fp = Misc.hash2md5(options)
+          key = name.to_s + "_" + fp
+        end
+
+        key += '.database'
         Persist.memory("Database:" << [key, dir] * "@") do
           options = options.dup
+
+          options[:organism] ||= options[:namespace] ||= self.namespace unless self.namespace.nil?
+
           persist_dir = dir
           persist_file = persist_dir[key].find
           file, registered_options = registry[name]
 
           options = Misc.add_defaults options, registered_options if registered_options and registered_options.any?
-          options = Misc.add_defaults options, :persist_file => persist_file, :namespace => namespace, :format => format, :persist => true
+          options = Misc.add_defaults options, :persist_file => persist_file, :format => format, :persist => true
 
           if entity_options
             options[:entity_options] ||= {}
@@ -118,7 +135,7 @@ class KnowledgeBase
                     Association.open(file, options, persist_options)
                   end
 
-          database.namespace = self.namespace
+          database.namespace = self.namespace if self.namespace
 
           database
         end
@@ -129,4 +146,34 @@ class KnowledgeBase
     get_index(name).fields
   end
 
+  def produce(name, *rest,&block)
+    register(name, *rest, &block)
+    get_index(name)
+  end
+
+  def info(name)
+
+    source = self.source(name)
+    target = self.target(name)
+    source_type = self.source_type(name)
+    target_type = self.target_type(name)
+    fields = self.fields(name)
+    source_entity_options = self.entity_options_for source_type, name
+    target_entity_options = self.entity_options_for target_type, name
+    undirected = self.undirected(name) == 'undirected'
+
+    info = {
+      :source => source,
+      :target => target,
+      :source_type => source_type,
+      :target_type => target_type,
+      :source_entity_options => source_entity_options,
+      :target_entity_options => target_entity_options,
+      :fields => fields,
+      :undirected => undirected,
+    }
+
+    info
+  end
+
 end

data/lib/rbbt/knowledge_base.rb

@@ -25,6 +25,10 @@ class KnowledgeBase
     @databases = {}
   end
 
+  def self.load(dir)
+    KnowledgeBase.new dir
+  end
+
   def setup(name, matches, reverse = false)
     AssociationItem.setup matches, self, name, reverse
   end

data/lib/rbbt/tsv/accessor.rb

@@ -633,15 +633,17 @@ module TSV
       break
     end
 
+    filename = Path === filename ? filename.find : (filename || "No filename")
+    filename + " [" + persistence_path + "]" if respond_to?(:persistence_path) and persistence_path
     with_unnamed do
       <<-EOF
-Filename = #{Path === filename ? filename.find : (filename || "No filename")}
+Filename = #{filename}
 Key field = #{key_field || "*No key field*"}
 Fields = #{fields ? Misc.fingerprint(fields) : "*No field info*"}
 Type = #{type}
 Serializer = #{serializer.inspect}
 Size = #{size}
-namespace = #{namespace}
+namespace = #{Misc.fingerprint namespace}
 identifiers = #{Misc.fingerprint identifiers}
 Example:
   - #{key} -- #{Misc.fingerprint values }

data/lib/rbbt/tsv/change_id.rb

@@ -8,6 +8,8 @@ module TSV
 
     identifiers, persist_input = Misc.process_options options, :identifiers, :persist_input
 
+    identifiers = Organism.identifiers(tsv.namespace) if identifiers.nil? and tsv.namespace
+
     if not tsv.fields.include? format
       new = {}
       tsv.each do |k,v|

data/lib/rbbt/tsv/util.rb

@@ -61,6 +61,7 @@ module TSV
       fields = nil
       tsv.tap{|e| e.unnamed =  true; fields = e.fields}.through do |gene, names|
         names.zip(fields).each do |list, format|
+          list = [list] unless Array === list
           list.delete_if do |name| name.empty? end
           next if list.empty?
           text << list.collect{|name| [name, format] * "\t"} * "\n" << "\n"
@@ -255,7 +256,7 @@ module TSV
     new = {}
     case type
     when :double
-      self
+      return self
     when :flat
       through do |k,v|
         new[k] = [v]

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

@@ -81,7 +81,11 @@ module R
                            args << "--RS-pidfile"
                            args << "'#{pid_file}'"
 
-                           bin_path = File.join(ENV["R_HOME"], "bin/Rserve")
+                           if ENV["R_HOME"]
+                             bin_path = File.join(ENV["R_HOME"], "bin/Rserve") 
+                           else
+                             bin_path = "Rserve"
+                           end
                            cmd = bin_path + " " + args*" "
                            $stdout.reopen File.new('/dev/null', 'w')
                            exec(ENV, cmd)

data/share/rbbt_commands/tsv/sort

@@ -14,6 +14,7 @@ Display summary information. Works with Tokyocabinet HDB and BDB as well.
 
 -hh--header_hash* Change the character used to mark the header line (defaults to #)
 -f--field* Field to sort by (name or number)
+-a--absolute Sort as absolute values
 -h--help Help
 EOF
 
@@ -23,19 +24,33 @@ file = ARGV.shift
 
 file = STDIN if file == '-' or file.nil?
 
-field = options[:field] || '2'
-if field =~ /^\d+$/
-  field_pos = field.to_i
+if options[:absolute]
+  tsv = TSV.open(file, options)
+  dumper = TSV::Dumper.new tsv.options
+  dumper.init
+  field = options[:field] || 1
+  keys = tsv.sort(field) do |a| a = a.first if Array === a; a.to_f.abs end
+  TSV.traverse keys, :type => :array, :into => dumper do |key|
+    values = tsv[key]
+    [key,values]
+  end
+  stream = dumper.stream
 else
-  parser = TSV.parse_header(file, options)
-  field_pos = parser.all_fields.index(field) + 1
-  saved_line = parser.first_line
-end
+  field = options[:field] || '2'
+
+  if field =~ /^\d+$/
+    field_pos = field.to_i
+  else
+    parser = TSV.parse_header(file, options)
+    field_pos = parser.all_fields.index(field) + 1
+    saved_line = parser.first_line
+  end
 
-rest = ARGV.inject("-t\"\t\" -g -k#{field_pos}"){|acc,e| acc << " '#{e}'" }
+  rest = ARGV.inject("-t\"\t\" -g -k#{field_pos}"){|acc,e| acc << " '#{e}'" }
 
-file_io = TSV.get_stream(file)
-stream = Misc.sort_stream(file_io, options[:header_hash] || "#", rest)
+  file_io = TSV.get_stream(file)
+  stream = Misc.sort_stream(file_io, options[:header_hash] || "#", rest)
+end
 
 begin
   while line = stream.gets

data/test/rbbt/test_knowledge_base.rb

@@ -111,5 +111,17 @@ class TestKnowledgeBase < Test::Unit::TestCase
       end
     end
   end
+
+  def test_knowledge_base_reuse
+    organism = Organism.default_code("Hsa")
+    Log.severity = 0
+    TmpFile.with_file(nil, false) do |tmpdir|
+      Path.setup(tmpdir)
+      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/
+    end
+  end
 end
 

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rbbt-util
 version: !ruby/object:Gem::Version
-  version: 5.19.18
+  version: 5.19.19
 platform: ruby
 authors:
 - Miguel Vazquez
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2016-02-27 00:00:00.000000000 Z
+date: 2016-03-14 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake