bioroebe 0.10.80

data/lib/bioroebe/codons/show_this_codon_table.rb

@@ -0,0 +1,573 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# === Bioroebe::ShowThisCodonTable
+#
+# This class will, on the commandline, show a particular codon table,
+# e. g. the codon table used by eukaryotes, or by bacteria and so
+# forth.
+#
+# It is able to respond to input such as 1, 2, 3 etc... if such a
+# codon table exists.
+#
+# The commandline option --rna (or --RNA) can be used to have the output
+# default to RNA rather than DNA. By default, however had, DNA is the
+# output for this class. 
+#
+# Invocation examples:
+#
+#   Bioroebe::ShowThisCodonTable.new(ARGV)
+#   Bioroebe::ShowThisCodonTable.new(3)
+#
+# =========================================================================== #
+# require 'bioroebe/codons/show_this_codon_table.rb'
+# Bioroebe::ShowThisCodonTable.new(ARGV)
+# =========================================================================== #
+require 'bioroebe/base/commandline_application/commandline_application.rb'
+
+module Bioroebe
+
+class ShowThisCodonTable < ::Bioroebe::CommandlineApplication
+
+  require 'bioroebe/codons/codon_table.rb'
+  require 'bioroebe/codons/codon_tables.rb'
+
+  # ========================================================================= #
+  # === LPAD
+  # ========================================================================= #
+  LPAD   = '  '     # Left Padding.
+
+  # ========================================================================= #
+  # === MPAD
+  # ========================================================================= #
+  MPAD   = '      ' # 6 space characters.
+
+  # ========================================================================= #
+  # === CYAN
+  # ========================================================================= #
+  CYAN   = Colours::CYAN
+  CYAN_U = CYAN+'U'
+  CYAN_A = CYAN+'A'
+  CYAN_G = CYAN+'G'
+  CYAN_C = CYAN+'C'
+
+  # ========================================================================= #
+  # === CodonTable::ARRAY_ALL_CODONS                                (array tag)
+  #
+  # This Array will be used to batch-define the necessary methods.
+  # ========================================================================= #
+  ARRAY_ALL_CODONS = %w(
+
+    TTT? TTC? TTA? TTG?
+    TCT? TCC? TCA? TCG?
+    TAT? TAC? TAA? TAG?
+    TGT? TGC? TGA? TGG?
+
+    CTT? CCT? CAT? CGT?
+    CTC? CCC? CAC? CGC?
+    CTA? CCA? CAA? CGA?
+    CTG? CCG? CAG? CGG?
+
+    ATT? ACT? AAT? AGT?
+    ATC? ACC? AAC? AGC?
+    ATA? ACA? AAA? AGA?
+    ATG? ACG? AAG? AGG?
+
+    GTT? GCT? GAT? GGT?
+    GTC? GCC? GAC? GGC?
+    GTA? GCA? GAA? GGA?
+    GTG? GCG? GAG? GGG?
+
+  )
+
+  # ========================================================================= #
+  # === STOP
+  #
+  # The stop codon.
+  # ========================================================================= #
+  STOP = Colours::RED+'STOP'+Colours::GREEN
+
+  # ========================================================================= #
+  # === DEFAULT_CODON_TABLE_TO_USE
+  # ========================================================================= #
+  DEFAULT_CODON_TABLE_TO_USE = :human
+
+  # ========================================================================= #
+  # === initialize
+  # ========================================================================= #
+  def initialize(
+      commandline_arguments = ARGV,
+      run_already           = true
+    )
+    reset
+    set_commandline_arguments(
+      commandline_arguments
+    )
+    run if run_already
+  end
+
+  # ========================================================================= #
+  # === reset
+  # ========================================================================= #
+  def reset
+    super()
+    @internal_hash = {}
+    # ======================================================================= #
+    # === :title
+    # ======================================================================= #
+    @internal_hash[:title] = nil
+    # ======================================================================= #
+    # === @is_available
+    #
+    # This instance variable keeps track as to whether the given coden
+    # table is available or whether it is not, based on a number. For
+    # some reasons, probably historic reasons, some numbers are missing.
+    #
+    # By default this variable is true, but it will be set to false if
+    # the input can not be correlated to an existing codon table.
+    # ======================================================================= #
+    @is_available = false
+    # ======================================================================= #
+    # === @show_rna_or_dna
+    #
+    # This variable can be either :dna or :rna.
+    #
+    # The default is :dna. If you wish to set this to :rna on the commandline
+    # then you can use "codontable --rna".
+    # ======================================================================= #
+    @show_rna_or_dna = :dna
+    # ======================================================================= #
+    # === @autogenerate_codon_related_methods
+    # ======================================================================= #
+    @autogenerate_codon_related_methods = true
+  end
+
+  # ========================================================================= #
+  # === set_title
+  # ========================================================================= #
+  def set_title(
+      i = nil
+    )
+    if i and i.is_a?(String)
+      i = i.dup if i.frozen?
+      i << ' Codon Table' unless i.end_with?(' Codon Table')
+    end
+    # ======================================================================= #
+    # === :title
+    #
+    # @title will keep track of the title of the specific codon table.
+    # That is, every codon table has a specific name such as "Standard
+    # Eukaryotic Table".
+    # ======================================================================= #
+    @internal_hash[:title] = i
+  end
+
+  # ========================================================================= #
+  # === green
+  # ========================================================================= #
+  def green
+    Colours::GREEN
+  end; alias rev green # rev tag
+
+  # ========================================================================= #
+  # === stop?
+  # ========================================================================= #
+  def stop?
+    STOP
+  end
+
+  # ========================================================================= #
+  # === cyana
+  # ========================================================================= #
+  def cyana
+    cyan+'A'+rev
+  end; alias cyan_a cyana # === cyan_a
+       alias cyan_A cyana # === cyan_A
+
+  # ========================================================================= #
+  # === cyang
+  # ========================================================================= #
+  def cyang
+    cyan+'G'+rev
+  end; alias cyan_g cyang # === cyan_g
+       alias cyan_G cyang # === cyan_G
+
+  # ========================================================================= #
+  # === cyan
+  # ========================================================================= #
+  def cyan
+    CYAN # This will work, as CYAN has been defined in this file.
+  end
+
+  # ========================================================================= #
+  # === consider_using_a_specific_codon_table
+  # ========================================================================= #
+  def consider_using_a_specific_codon_table(i = first_argument?)
+    set_use_this_codon_table(i)
+  end
+
+  # ========================================================================= #
+  # === is_this_codon_table_available?
+  # ========================================================================= #
+  def is_this_codon_table_available?(i)
+    target = directory_to_the_codon_tables?+"#{i}.yml"
+    return File.exist?(target)
+  end
+
+  # ========================================================================= #
+  # === lpad
+  # ========================================================================= #
+  def lpad(i)
+    if i.include? 'STOP'
+      i+'   '
+    elsif i.include? 'START'
+      i+'     '
+    else
+      i+'      '
+    end
+  end
+
+  # ========================================================================= #
+  # === lpad?
+  # ========================================================================= #
+  def lpad?
+    LPAD
+  end
+
+  # ========================================================================= #
+  # === use_this_codon_table?
+  # ========================================================================= #
+  def use_this_codon_table?
+    ::Bioroebe.codon_table?
+  end; alias use_which_codon_table? use_this_codon_table? # === use_which_codon_table
+       alias codon_table?           use_this_codon_table? # === codon_table?
+
+  # ========================================================================= #
+  # === dataset?
+  # ========================================================================= #
+  def dataset?
+    ::Bioroebe.codon_table_dataset?
+  end
+
+  # ========================================================================= #
+  # === return_dataset_from_codon_table
+  #
+  # This method will return the dataset from the codon table.
+  # ========================================================================= #
+  def return_dataset_from_codon_table(
+      this_codon_table = use_which_codon_table?
+    )
+    if this_codon_table.is_a?(String) and (this_codon_table =~ /^\d+$/)
+      this_codon_table = this_codon_table.to_i
+      set_codon_table(this_codon_table)
+    end
+    ::Bioroebe.codon_table_dataset?
+  end
+
+  # ========================================================================= #
+  # === set_codon_table
+  # ========================================================================= #
+  def set_codon_table(i)
+    ::Bioroebe.set_codon_table(i)
+  end
+
+  # ========================================================================= #
+  # === title?
+  # ========================================================================= #
+  def title?
+    @internal_hash[:title]
+  end
+
+  # ========================================================================= #
+  # === lpad_on
+  #
+  # This method combines several calls such as this:
+  #
+  #   lpad(TTC?)+lpad(TCC?)+lpad(TAC?)+lpad(TGC?)
+  #
+  # It was too long, so it was changed to.
+  #
+  #   lpad_on(TTT?, TCT?, TAT?, TGT?)
+  #
+  # ========================================================================= #
+  def lpad_on(*i)
+    i.map {|entry| lpad(entry) }.join
+  end
+
+  # ========================================================================= #
+  # === determine_all_codon_methods
+  #
+  # Dynamically define the proper methods.
+  # ========================================================================= #
+  def determine_all_codon_methods(
+      dataset = dataset?
+    )
+    start_codons = dataset['START']
+    if start_codons.include? '|'
+      start_codons = start_codons.split('|').map(&:strip)
+    end
+    if @autogenerate_codon_related_methods
+      dataset = dataset?
+      ARRAY_ALL_CODONS.each {|entry|
+        entry = entry.to_sym
+        # =================================================================== #
+        # We will only define these methods once.
+        # =================================================================== #
+        unless self.respond_to? entry
+          # ================================================================= #
+          # We define methods on this class, such as .GGG? and similar.
+          # ================================================================= #
+          self.class.instance_eval {
+            define_method(entry) {
+              name_of_the_codon = __method__.to_s.chop
+              if dataset.has_key? name_of_the_codon
+                result = dataset.fetch(name_of_the_codon)
+                result = stop? if result == '*'
+                if start_codons.include? name_of_the_codon
+                  result << 'START'
+                end
+              else
+                result = 'UNKNOWN (from '+name_of_the_codon+')'
+              end
+              return result
+            }
+          }
+        end
+      }
+      @autogenerate_codon_related_methods = false
+    end
+  end
+
+  # ========================================================================= #
+  # === colourized_line
+  # ========================================================================= #
+  def colourized_line(i)
+    result = i.dup.gsub(/\*/, stop?).
+                   gsub(/START/, '*').
+                   squeeze('*').
+                   gsub(/\*/,tomato('*')+rev)
+    e result
+    i
+  end
+
+  # ========================================================================= #
+  # === menu                                                       (menu tag)
+  # ========================================================================= #
+  def menu(
+      i = @commandline_arguments
+    )
+    if i.is_a? Array
+      i.each {|entry| menu(entry) }
+    else
+      case i # case tag
+      # ===================================================================== #
+      # === codontable --rna
+      #
+      # This variant will display the table as "RNA".
+      #
+      # Invocation examples:
+      #
+      #   ctable --rna
+      #   ctable 1 --rna
+      #
+      # ===================================================================== #
+      when /^-?-?rna$/i
+        @show_rna_or_dna = :rna
+        @commandline_arguments.reject! {|entry| entry.to_s =~ /^-?-?rna$/i }
+      else
+        if i.to_s =~ /^(\d{1,2})-(\d{1,2})$/
+          # ================================================================= #
+          #   ctable 1-3
+          # ================================================================= #
+          $1.to_i.upto($2.to_i) {|n|
+            consider_using_a_specific_codon_table(n)
+            determine_dataset
+            determine_all_codon_methods
+            display_codon_table
+          }
+          exit
+        elsif i.to_s =~ /^\d{1,2}$/
+          # ================================================================= #
+          # Need to convert into a real number.
+          #
+          # Usage Examples:
+          #
+          #   ctable 1
+          #   ctable 3
+          #   ctable 1-2
+          #
+          # ================================================================= #
+          require 'bioroebe/codons/codon_tables.rb'
+          @commandline_arguments[0] = i.to_i
+          set_title(
+            Bioroebe::CodonTables.title_for_entry_number?(i.to_i)
+          )
+        end
+      end
+    end
+  end
+
+  # ========================================================================= #
+  # === display_title
+  #
+  # This method will show the title of the codon table in question.
+  # ========================================================================= #
+  def display_title
+    title = title?
+    if title
+      erev title+' (codon table number: '+
+           steelblue(codon_table?.to_s)+rev+'):'
+    end
+  end
+
+  # ========================================================================= #
+  # === determine_dataset
+  #
+  # This method will determine the correct codon-dataset. It has to
+  # make sure that the codon table actually exists, though.
+  # ========================================================================= #
+  def determine_dataset
+    # ======================================================================= #
+    # We must set the correct title to use, if no argument was provided.
+    # ======================================================================= #
+    if title?.nil? and first_argument?.nil?
+      set_title("Standard (Eukaryote) Codon Table")
+      set_codon_table(1)
+    end
+    this_codon_table = use_which_codon_table?
+    if this_codon_table.is_a?(String) and (this_codon_table =~ /^\d+$/)
+      title = ::Bioroebe::CodonTables.definitions?[this_codon_table.to_i]
+      set_title(title)  
+    end
+  end
+
+  # ========================================================================= #
+  # === notify_the_user_that_no_such_codon_table_could_be_found
+  # ========================================================================= #
+  def notify_the_user_that_no_such_codon_table_could_be_found
+    erev 'No codon table matching this input ('+
+         steelblue(first_argument?.to_s)+
+         rev+') could be found.'
+  end
+
+  # ========================================================================= #
+  # === top_row_header
+  # ========================================================================= #
+  def top_row_header
+    e cyan+N+LPAD+'      '+cyanu+MPAD+cyanc+MPAD+cyana+MPAD+cyang
+  end
+
+  # ========================================================================= #
+  # === cyanu
+  #
+  # This will either show a U or a T.
+  # ========================================================================= #
+  def cyanu
+    case @show_rna_or_dna
+    # ======================================================================= #
+    # === :rna
+    # ======================================================================= #
+    when :rna
+      "#{cyan}U#{rev}"
+    # ======================================================================= #
+    # === :dna
+    # ======================================================================= #
+    when :dna
+      "#{cyan}T#{rev}"
+    end
+  end; alias cyan_U cyanu # === cyan_U
+       alias cyan_u cyanu # === cyan_u
+       alias cyan_T cyanu # === cyan_T
+
+  # ========================================================================= #
+  # === cyanc
+  # ========================================================================= #
+  def cyanc
+    cyan+'C'+rev
+  end; alias cyan_c cyanc # === cyan_c
+       alias cyan_C cyanc # === cyan_C
+
+  # ========================================================================= #
+  # === set_use_this_codon_table
+  #
+  # This will initially hold the codon table internally.
+  # ========================================================================= #
+  def set_use_this_codon_table(
+      i = DEFAULT_CODON_TABLE_TO_USE
+    )
+    i = i.first if i.is_a? Array
+    case i
+    when 'human'
+      i = 1
+    when nil
+      i = DEFAULT_CODON_TABLE_TO_USE
+    end
+    if is_this_codon_table_available?(i)
+      @is_available = true
+      ::Bioroebe.set_use_this_codon_table(i)
+    end
+  end
+
+  # ========================================================================= #
+  # === left_pad
+  # ========================================================================= #
+  def left_pad
+    print '  ' 
+  end
+
+  # ========================================================================= #
+  # === display_codon_table
+  #
+  # This method will finally show the codon table, in DNA form.
+  # ========================================================================= #
+  def display_codon_table(
+      cliner_width_to_use = 41
+    )
+    left_pad; display_title
+    left_pad; top_row_header
+    left_pad; ccliner(cliner_width_to_use)
+    left_pad; colourized_line lpad?+cyan_U+' |   '+lpad_on(TTT?, TCT?, TAT?, TGT?)+'| '+cyan_T
+    left_pad; colourized_line lpad?+cyan_U+' |   '+lpad_on(TTC?, TCC?, TAC?, TGC?)+'| '+cyan_C
+    left_pad; colourized_line lpad?+cyan_U+' |   '+lpad_on(TTA?, TCA?, TAA?, TGA?)+'| '+cyan_A
+    left_pad; colourized_line lpad?+cyan_U+' |   '+lpad_on(TTG?, TCG?, TAG?, TGG?)+'| '+cyan_G
+    left_pad; ccliner(cliner_width_to_use)
+    left_pad; colourized_line lpad?+cyan_C+' |   '+lpad_on(CTT?, CCT?, CAT?, CGT?)+'| '+cyan_T
+    left_pad; colourized_line lpad?+cyan_C+' |   '+lpad_on(CTC?, CCC?, CAC?, CGC?)+'| '+cyan_C
+    left_pad; colourized_line lpad?+cyan_C+' |   '+lpad_on(CTA?, CCA?, CAA?, CGA?)+'| '+cyan_A
+    left_pad; colourized_line lpad?+cyan_C+' |   '+lpad_on(CTG?, CCG?, CAG?, CGG?)+'| '+cyan_G
+    left_pad; ccliner(cliner_width_to_use)
+    left_pad; colourized_line lpad?+cyan_A+' |   '+lpad_on(ATT?, ACT?, AAT?, AGT?)+'| '+cyan_T
+    left_pad; colourized_line lpad?+cyan_A+' |   '+lpad_on(ATC?, ACC?, AAC?, AGC?)+'| '+cyan_C
+    left_pad; colourized_line lpad?+cyan_A+' |   '+lpad_on(ATA?, ACA?, AAA?, AGA?)+'| '+cyan_A
+    left_pad; colourized_line lpad?+cyan_A+' |   '+lpad_on(ATG?, ACG?, AAG?, AGG?)+'| '+cyan_G
+    left_pad; ccliner(cliner_width_to_use)
+    left_pad; colourized_line lpad?+cyan_G+' |   '+lpad_on(GTT?, GCT?, GAT?, GGT?)+'| '+cyan_T
+    left_pad; colourized_line lpad?+cyan_G+' |   '+lpad_on(GTC?, GCC?, GAC?, GGC?)+'| '+cyan_C
+    left_pad; colourized_line lpad?+cyan_G+' |   '+lpad_on(GTA?, GCA?, GAA?, GGA?)+'| '+cyan_A
+    left_pad; colourized_line lpad?+cyan_G+' |   '+lpad_on(GTG?, GCG?, GAG?, GGG?)+'| '+cyan_G
+    left_pad; ccliner(cliner_width_to_use)
+    e
+  end; private :display_codon_table
+
+  # ========================================================================= #
+  # === run
+  # ========================================================================= #
+  def run
+    menu
+    consider_using_a_specific_codon_table
+    determine_dataset
+    determine_all_codon_methods
+    if @is_available
+      display_codon_table
+    else
+      notify_the_user_that_no_such_codon_table_could_be_found
+    end
+  end
+
+end; end
+
+if __FILE__ == $PROGRAM_NAME
+  Bioroebe::ShowThisCodonTable.new(ARGV)
+end # ctable 3
+    # ctable 9

data/lib/bioroebe/codons/start_codons.rb

@@ -0,0 +1,105 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# require 'bioroebe/codons/start_codons.rb'
+# =========================================================================== #
+module Bioroebe
+
+  # ========================================================================= #
+  # === @start_codon
+  #
+  # The default start codon will be ATG, in DNA. This refers to the
+  # human/vertebrate start codon.
+  #
+  # You can query it via the method Bioroebe.start_codon?
+  # ========================================================================= #
+  @start_codon = 'ATG' # ← This should always be a String.
+
+  # ========================================================================= #
+  # === @start_codons
+  # ========================================================================= #
+  @start_codons = [] # Whereas this here is an Array.
+
+  # ========================================================================= #
+  # === Bioroebe.start_codon?
+  # ========================================================================= #
+  def self.start_codon?
+    @start_codon
+  end; self.instance_eval { alias start_codon? start_codon? } # === Bioroebe.start_codon?
+
+  # ========================================================================= #
+  # === Bioroebe.set_start_codon
+  #
+  # This will designate the most important start codon.
+  # ========================================================================= #
+  def self.set_start_codon(i = 'ATG')
+    @start_codon = i
+  end
+
+  # ========================================================================= #
+  # === Bioroebe.set_start_codons
+  #
+  # This method can be used to designate the start-codons.
+  #
+  # It will be called automatically from code within codon_table.rb
+  # as well, so that we always have the corect start-codons ready
+  # for use.
+  # ========================================================================= #
+  def self.set_start_codons(
+      i = @start_codon
+    )
+    if i.is_a? String and i.include? '|'
+      i = i.split('|').map(&:strip)
+    end
+    i = [i].flatten # Will always be an Array.
+    @start_codons = i
+  end
+
+  # ========================================================================= #
+  # === Initialize the start codons.
+  # ========================================================================= #
+  set_start_codons # Set the default start codons here. (Initializer)
+
+  # ========================================================================= #
+  # === Bioroebe.start_codons?
+  # ========================================================================= #
+  def self.start_codons?
+    @start_codons
+  end
+
+  # ========================================================================= #
+  # === Bioroebe.is_start_codon?
+  #
+  # Query-method to determine whether the given input is a start codon.
+  #
+  # The method will return true if the given input is a start codon
+  # in the currently selected codon table; otherwise this method
+  # will return false.
+  #
+  # This way the method can be used to determine whether something
+  # is a start codon.
+  #
+  # Note that the method does not do any additional checks, so,
+  # for instance, if you have RNA and work with that, then you
+  # may have to convert it to DNA before calling this method
+  # here. See the usage examples below for that.
+  #
+  # Usage examples:
+  #
+  #   Bioroebe.is_a_start_codon? 'AUG' # => false
+  #   Bioroebe.is_a_start_codon? 'ATG' # => true
+  #
+  # ========================================================================= #
+  def self.is_start_codon?(i)
+    @start_codon == i.upcase
+  end; self.instance_eval { alias is_a_start_codon? is_start_codon? } # === Bioroebe.is_a_start_codon?
+
+end
+
+if __FILE__ == $PROGRAM_NAME
+  alias e puts
+  # pp Bioroebe.is_a_start_codon? 'AUG' # => false
+  e 'The start codon in use is: '+
+     Bioroebe.start_codon?
+end

data/lib/bioroebe/colours/colour_schemes/README.md

@@ -0,0 +1,10 @@
+This directory will contain common, popular colours schemes for
+aminoacids and DNA/RNA.
+
+Most of these schemes have been created elsewhere first; the
+Bioroebe project will merely make use of them.
+
+The idea is that we should be able to quickly "style" different
+sequences in different colours - be it in a HTML file or a cgi
+file or some other web-application file, but also on the
+commandline.