bel_parser 1.0.0.alpha.27-java

data/lib/bel/translator/plugins/bel_script.rb

@@ -0,0 +1,38 @@
+module BEL::Translator::Plugins
+  # BELScript plugs a translator for BEL Script into bel.rb.
+  # Supports multiple BEL specifications by using the
+  # {BELParser::Language::ExpressionValidator expression validator}.
+  module BelScript
+    ID          = :bel_script
+    NAME        = 'BEL Script Translator'
+    DESCRIPTION = 'A translator that reads/writes BEL nanopubs to BEL Script.'
+    MEDIA_TYPES = %i(application/bel)
+    EXTENSIONS  = %i(bel)
+
+    def self.create_translator(options = {})
+      require 'bel'
+      require_relative 'bel_script/translator'
+      BelScriptTranslator.new
+    end
+
+    def self.id
+      ID
+    end
+
+    def self.name
+      NAME
+    end
+
+    def self.description
+      DESCRIPTION
+    end
+
+    def self.media_types
+      MEDIA_TYPES
+    end
+
+    def self.file_extensions
+      EXTENSIONS
+    end
+  end
+end

data/lib/bel/translator/plugins/bel_script/bel_citation_serialization.rb

@@ -0,0 +1,125 @@
+require_relative 'nanopub_serialization.rb'
+
+# BEL Script nanopub serialization that groups nanopub by citation scoped to
+# individual statement groups (i.e. BEL Script's +SET STATEMENT_GROUP+ and
+# +UNSET STATEMENT_GROUP+).
+#
+# @example Citation serialization for a group of nanopub
+#   SET STATEMENT_GROUP = 12857727
+#   SET Citation = {"PubMed", "Journal...", "12857727", "2003-08-11", "", ""}
+#   SET Support = "USF1 and USF2 bound the IGF2R promoter in vitro, ..."
+#   SET CellLine = "MCF 10A"
+#   SET TextLocation = Abstract
+#   complex(p(HGNC:USF1),g(HGNC:IGF2R))
+#
+#   complex(p(HGNC:USF2),g(HGNC:IGF2R))
+#
+#   tscript(p(HGNC:USF2)) directlyIncreases r(HGNC:IGF2R)
+#
+#   tscript(p(HGNC:USF1)) causesNoChange r(HGNC:IGF2R)
+#
+#   SET Support = "c-Myc was present on the CDK4 promoter to the ..."
+#   complex(p(HGNC:MYC),g(HGNC:CDK4))
+#   UNSET STATEMENT_GROUP
+module BEL::Translator::Plugins::BelScript::BelCitationSerialization
+  include BEL::Translator::Plugins::BelScript::NanopubSerialization
+
+  # Serialize the {BEL::Nanopub::Nanopub nanopub} to a BEL Script string.
+  #
+  # Includes +SET AnnotationName+ and +UNSET AnnotationName+ where needed in
+  # order to remove duplicating annotations.
+  #
+  # @param  [BEL::Nanopub::Nanopub] nanopub the nanopub to serialize
+  # @return [String]               the BEL Script string
+  def to_bel(nanopub)
+    bel = ''
+
+    citation     = citation_value(nanopub)
+    support = support_value(nanopub)
+    annotations  = annotation_values(nanopub)
+
+    current_annotations            = {}.merge(annotations)
+    current_annotations[:Citation] = citation if citation
+    current_annotations[:Support]  = support if support
+
+    if !nanopub.citation.id || nanopub.citation.id.empty?
+      citation_id = quote('')
+    else
+      citation_id = quote_if_needed(nanopub.citation.id)
+    end
+
+    # Reset cumulative annotations if new citation.
+    if cumulative_citation == nil
+      bel << %Q{SET STATEMENT_GROUP = #{citation_id}\n}
+      cumulative_annotations.clear
+    elsif nanopub.citation != cumulative_citation
+      bel << %Q{UNSET STATEMENT_GROUP\n}
+      bel << "\n\n"
+      bel << %Q{SET STATEMENT_GROUP = #{citation_id}\n}
+      cumulative_annotations.clear
+    end
+
+    # Hang on to the last citation.
+    self.cumulative_citation = nanopub.citation
+
+    # UNSET unused annotations from previous nanopub.
+    (cumulative_annotations.keys - current_annotations.keys).each do |unset_key|
+      bel << "UNSET #{unset_key}\n"
+      cumulative_annotations.delete(unset_key)
+    end
+
+    # Remove annotation if key/value was SET by a previous nanopub.
+    Hash[
+      cumulative_annotations.to_a & current_annotations.to_a
+    ].each do |same_k, _|
+      current_annotations.delete(same_k)
+    end
+
+    # Retain the current nanopub's annotation in cumulative set.
+    cumulative_annotations.merge!(current_annotations)
+
+    # SET Citation
+    citation = current_annotations.delete(:Citation)
+    if citation
+      bel << "SET Citation = {#{citation}}\n"
+    end
+
+    # SET Support
+    support = current_annotations.delete(:Support)
+    if support
+      bel << %Q{SET Support = "#{support}"\n}
+    end
+
+    # SET new or modified annotations
+    current_annotations.sort.each do |(name, value)|
+      bel << "SET #{name} = #{value}\n"
+    end
+
+    # Assert BEL statement
+    bel << "#{nanopub.bel_statement}\n"
+
+    # Separate nanopub by new line.
+    bel << "\n"
+
+    bel
+  end
+
+  private
+
+  # The cumulative citation that is active for the current nanopub. This is
+  # tracked in order to decide when to begin a new statement group.
+  attr_accessor :cumulative_citation
+
+  # Returns the cumulative +Hash+ of annotations. This *state* is used to keep
+  # track of the active, scoped annotations as nanopub is serialized.
+  def cumulative_annotations
+    @cumulative_annotations ||= {}
+  end
+
+  # Return BEL Script syntax that should completes the BEL Script document.
+  # For Citation serialization we will always end with the unset of a
+  # statement group.
+  def epilogue
+    %Q{UNSET STATEMENT_GROUP\n}
+  end
+end

data/lib/bel/translator/plugins/bel_script/bel_discrete_serialization.rb

@@ -0,0 +1,109 @@
+require_relative 'nanopub_serialization'
+
+# BEL Script nanopub serialization that writes each nanopub with their full
+# set of annotations (i.e. includes all `SET` and necessary `UNSET` records).
+# This style is more readable because it groups all set annotations near the
+# BEL statement.
+#
+# @example Discrete serialization for a group of nanopub
+#   SET Citation = {"PubMed", "Journal...", "12857727", "2003-08-11", "", ""}
+#   SET Support = "USF1 and USF2 bound the IGF2R promoter in vitro, ..."
+#   SET CellLine = "MCF 10A"
+#   SET TextLocation = Abstract
+#   complex(p(HGNC:USF1),g(HGNC:IGF2R))
+#
+#   SET Citation = {"PubMed", "Journal...", "12857727", "2003-08-11", "", ""}
+#   SET Support = "USF1 and USF2 bound the IGF2R promoter in vitro, ..."
+#   SET CellLine = "MCF 10A"
+#   SET TextLocation = Abstract
+#   complex(p(HGNC:USF2),g(HGNC:IGF2R))
+#
+#   SET Citation = {"PubMed", "Journal...", "12857727", "2003-08-11", "", ""}
+#   SET Support = "USF1 and USF2 bound the IGF2R promoter in vitro, ..."
+#   SET CellLine = "MCF 10A"
+#   SET TextLocation = Abstract
+#   tscript(p(HGNC:USF2)) directlyIncreases r(HGNC:IGF2R)
+#
+#   SET Citation = {"PubMed", "Journal...", "12857727", "2003-08-11", "", ""}
+#   SET Support = "USF1 and USF2 bound the IGF2R promoter in vitro, ..."
+#   SET CellLine = "MCF 10A"
+#   SET TextLocation = Abstract
+#   tscript(p(HGNC:USF1)) causesNoChange r(HGNC:IGF2R)
+#
+#   SET Citation = {"PubMed", "Journal...", "12857727", "2003-08-11", "", ""}
+#   SET Support = "c-Myc was present on the CDK4 promoter to the ..."
+#   SET CellLine = "MCF 10A"
+#   SET TextLocation = Abstract
+#   complex(p(HGNC:MYC),g(HGNC:CDK4))
+#
+#   UNSET CellLine
+module BEL::Translator::Plugins::BelScript::BelDiscreteSerialization
+  include BEL::Translator::Plugins::BelScript::NanopubSerialization
+
+  # Serialize the {BEL::Nanopub::Nanopub nanopub} to a BEL Script string.
+  # Includes all necessary +SET AnnotationName+ and +UNSET AnnotationName+
+  # records around the BEL statement.
+  #
+  # @param  [BEL::Nanopub::Nanopub] nanopub the nanopub to serialize
+  # @return [String]               the BEL Script string
+  def to_bel(nanopub)
+    bel = ''
+
+    citation     = citation_value(nanopub)
+    support = support_value(nanopub)
+    annotations  = annotation_values(nanopub)
+
+    current_annotations            = {}.merge(annotations)
+    current_annotations[:Citation] = citation if citation
+    current_annotations[:Support]  = support if support
+
+    # UNSET unused annotations from previous nanopub.
+    (cumulative_annotations.keys - current_annotations.keys).each do |unset_key|
+      bel << "UNSET #{unset_key}\n"
+      cumulative_annotations.delete(unset_key)
+    end
+
+    # Retain the current nanopub's annotation in cumulative set.
+    cumulative_annotations.merge!(current_annotations)
+
+    # SET Citation
+    citation = current_annotations.delete(:Citation)
+    if citation
+      bel << "SET Citation = {#{citation}}\n"
+    end
+
+    # SET Support
+    support = current_annotations.delete(:Support)
+    if support
+      bel << %Q{SET Support = "#{support}"\n}
+    end
+
+    # SET new or modified annotations
+    current_annotations.sort.each do |(name, value)|
+      bel << "SET #{name} = #{value}\n"
+    end
+
+    # Assert BEL statement
+    bel << "#{nanopub.bel_statement}\n"
+
+    # Separate nanopub by new line.
+    bel << "\n"
+
+    bel
+  end
+
+  private
+
+  # Returns the cumulative +Hash+ of annotations. This *state* is used to keep
+  # track of the active, scoped annotations as nanopub is serialized.
+  def cumulative_annotations
+    @cumulative_annotations ||= {}
+  end
+
+  # Return BEL Script syntax that completes the BEL Script document.
+  # The empty string is returned since no ending syntax is necessary when
+  # serializing each nanopub discretely.
+  def epilogue
+    ""
+  end
+end

data/lib/bel/translator/plugins/bel_script/bel_top_down_serialization.rb

@@ -0,0 +1,100 @@
+require_relative 'nanopub_serialization.rb'
+
+# BEL Script nanopub serialization that writes nanopubs sequentially while
+# including only the necessary unsetting of annotations (i.e. BEL Script's
+# +UNSET AnnotationName+ syntax).
+#
+# @example Top-down serialization for a group of nanopubs
+#   SET Citation = {"PubMed", "Journal...", "12857727", "2003-08-11", "", ""}
+#   SET Support = "USF1 and USF2 bound the IGF2R promoter in vitro, ..."
+#   SET CellLine = "MCF 10A"
+#   SET TextLocation = Abstract
+#   complex(p(HGNC:USF1),g(HGNC:IGF2R))
+#
+#   complex(p(HGNC:USF2),g(HGNC:IGF2R))
+#
+#   tscript(p(HGNC:USF2)) directlyIncreases r(HGNC:IGF2R)
+#
+#   tscript(p(HGNC:USF1)) causesNoChange r(HGNC:IGF2R)
+#
+#   SET Support = "c-Myc was present on the CDK4 promoter to the ..."
+#   complex(p(HGNC:MYC),g(HGNC:CDK4))
+#
+#   UNSET CellLine
+module BEL::Translator::Plugins::BelScript::BelTopDownSerialization
+  include BEL::Translator::Plugins::BelScript::NanopubSerialization
+
+  # Serialize the {BEL::Nanopub::Nanopub nanopub} to a BEL Script string.
+  # Includes all necessary +SET AnnotationName+ and +UNSET AnnotationName+
+  # records within the scope of a citation's statement group.
+  #
+  # @param  [BEL::Nanopub::Nanopub] nanopub the nanopub to serialize
+  # @return [String]               the BEL Script string
+  def to_bel(nanopub)
+    bel = ''
+
+    citation     = citation_value(nanopub)
+    support = support_value(nanopub)
+    annotations  = annotation_values(nanopub)
+
+    current_annotations            = {}.merge(annotations)
+    current_annotations[:Citation] = citation if citation
+    current_annotations[:Support]  = support if support
+
+    # UNSET unused annotations from previous nanopub.
+    (cumulative_annotations.keys - current_annotations.keys).each do |unset_key|
+      bel << "UNSET #{unset_key}\n"
+      cumulative_annotations.delete(unset_key)
+    end
+
+    # Remove annotation if key/value was SET by a previous nanopub.
+    Hash[
+      cumulative_annotations.to_a & current_annotations.to_a
+    ].each do |same_k, _|
+      current_annotations.delete(same_k)
+    end
+
+    # Retain the current nanopub's annotation in cumulative set.
+    cumulative_annotations.merge!(current_annotations)
+
+    # SET Citation
+    citation = current_annotations.delete(:Citation)
+    if citation
+      bel << "SET Citation = {#{citation}}\n"
+    end
+
+    # SET Support
+    support = current_annotations.delete(:Support)
+    if support
+      bel << %Q{SET Support = "#{support}"\n}
+    end
+
+    # SET new or modified annotations
+    current_annotations.sort.each do |(name, value)|
+      bel << "SET #{name} = #{value}\n"
+    end
+
+    # Assert BEL statement
+    bel << "#{nanopub.bel_statement}\n"
+
+    # Separate nanopub by new line.
+    bel << "\n"
+
+    bel
+  end
+
+  private
+
+  # Returns the cumulative +Hash+ of annotations. This *state* is used to keep
+  # track of the active, scoped annotations as nanopub is serialized.
+  def cumulative_annotations
+    @cumulative_annotations ||= {}
+  end
+
+  # Return BEL Script syntax that completes the BEL Script document.
+  # The empty string is returned since no ending syntax is necessary when
+  # serializing in a top-down manner.
+  def epilogue
+    ""
+  end
+end

data/lib/bel/translator/plugins/bel_script/nanopub_serialization.rb

@@ -0,0 +1,79 @@
+require 'bel/quoting'
+
+# Serializing of common {BEL::Nanopub::Nanopub nanopub} components to BEL
+# Script syntax.
+#
+# @abstract
+module BEL::Translator::Plugins::BelScript::NanopubSerialization
+  include BEL::Quoting
+
+  # Serialize the {BEL::Nanopub::Nanopub nanopub} to a BEL Script string.
+  #
+  # @param  [BEL::Nanopub::Nanopub] nanopub the nanopub to serialize
+  # @return [String]               the BEL Script string
+  # @abstract Include and override {#to_bel} to implement serialization
+  #           {BEL::Nanopub::Nanopub nanopub} to BEL Script
+  def to_bel(nanopub)
+  end
+
+  # Return BEL Script syntax that completes the BEL Script document.
+  #
+  # @abstract
+  def epilogue
+    raise NotImplementedError.new("#{self.class}#epilogue")
+  end
+
+  protected
+
+  def citation_value(nanopub)
+    citation = nanopub.citation
+
+    return nil unless citation && citation.valid?
+
+    values = citation.to_a
+    values.map! { |v|
+      v ||= ""
+      if v.respond_to?(:each)
+        %Q{"#{v.join('|')}"}
+      else
+        %Q{"#{v}"}
+      end
+    }
+    values.join(', ')
+  end
+
+  def support_value(nanopub)
+    support = nanopub.support
+
+    return nil unless support && support.value
+
+    value = support.value
+    value.gsub!("\n", "")
+    value.gsub!('"', %Q{\\"})
+    value
+  end
+
+  def annotation_values(nanopub)
+    experiment_context = nanopub.experiment_context
+
+    return {} unless experiment_context
+
+    Hash[
+      experiment_context.
+        sort_by { |obj| obj[:name].to_sym }.
+        map { |obj|
+          name  = obj[:name].to_sym
+          value = obj[:value]
+
+          value_s =
+            if value.respond_to? :map
+              "{#{value.map { |v| quote(v) }.join(', ')}}"
+            else
+              quote(value)
+            end
+
+          [name, value_s]
+        }
+    ]
+  end
+end