bioroebe 0.10.80 → 0.12.24

data/lib/bioroebe/utility_scripts/compacter/compacter.rb

@@ -0,0 +1,251 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# === Bioroebe::Compacter
+#
+# This class can be used to compact a file, in particular FASTA files.
+#
+# The first line will be removed if it starts with a ">" token.
+#
+# Usage example:
+#
+#   Biroebe::Compacter.new(ARGV)
+#
+# =========================================================================== #
+# require 'bioroebe/utility_scripts/compacter/compacter.rb'
+# =========================================================================== #
+require 'bioroebe/base/commandline_application/commandline_application.rb'
+
+module Bioroebe
+
+class Compacter < ::Bioroebe::CommandlineApplication # === Bioroebe::Compacter
+
+  require 'bioroebe/nucleotides/sanitize_nucleotide_sequence.rb'
+
+  # ========================================================================= #
+  # === initialize
+  # ========================================================================= #
+  def initialize(
+      commandline_arguments = nil,
+      run_already           = true,
+      &block
+    )
+    register_sigint
+    reset
+    set_commandline_arguments(
+      commandline_arguments
+    )
+    # ======================================================================= #
+    # === Handle blocks given to this class next
+    # ======================================================================= #
+    if block_given?
+      yielded = yield
+      case yielded
+      # ===================================================================== #
+      # === :do_not_ask_for_user_input
+      # ===================================================================== #
+      when :do_not_ask_for_user_input
+        @internal_hash[:ask_for_user_input] = false
+      end
+    end
+    run if run_already
+  end
+
+  # ========================================================================= #
+  # === reset                                                     (reset tag)
+  # ========================================================================= #
+  def reset
+    super()
+    infer_the_namespace
+    # ======================================================================= #
+    # === :work_on_these_files
+    # ======================================================================= #
+    @internal_hash[:work_on_these_files] = []
+    # ======================================================================= #
+    # === :ask_for_user_input
+    # ======================================================================= #
+    @internal_hash[:ask_for_user_input] = true
+    # ======================================================================= #
+    # === :perform_these_actions_on_each_file
+    #
+    # The following Array specifies which actions are to be performed onto
+    # the target file at hand. The user can modify this, to adjust the
+    # behaviour of this class.
+    # ======================================================================= #
+    @internal_hash[:perform_these_actions_on_each_file] = [
+      :strip_newlines,
+      :remove_numbers,
+      :upcase_the_sequence,
+      :eliminate_spaces
+    ]
+  end
+
+  # ========================================================================= #
+  # === menu                                                       (menu tag)
+  # ========================================================================= #
+  def menu(
+      i = commandline_arguments?
+    )
+    if i.is_a? Array
+      i.each {|entry| menu(entry) }
+    else
+      case i # case tag
+      # ===================================================================== #
+      # === --retain-newlines
+      #
+      # Usage example:
+      #
+      #   compacter SPRR4_protein.fasta --retain-newlines
+      #
+      # ===================================================================== #
+      when /-?-?retain(-|_)?newlines/
+        @internal_hash[:perform_these_actions_on_each_file].reject! {|entry|
+          entry == :strip_newlines
+        }
+      # ===================================================================== #
+      # === --help
+      # ===================================================================== #
+      when /help/
+        show_help
+        exit
+      end
+    end
+  end
+
+  # ========================================================================= #
+  # === show_help (help tag)
+  # ========================================================================= #
+  def show_help
+    opnn { :no_newline }
+    e
+    e '  --retain-newlines # retain newlines, aka do not dump into a single line'
+    e
+  end
+
+  # ========================================================================= #
+  # === determine_the_input_files
+  # ========================================================================= #
+  def determine_the_input_files
+    commandline_arguments?.each {|this_entry|
+      if File.file?(this_entry)
+        @internal_hash[:work_on_these_files] << this_entry
+      end
+    }
+  end
+
+  # ========================================================================= #
+  # === perform_these_actions_on_each_file?
+  # ========================================================================= #
+  def perform_these_actions_on_each_file?
+    @internal_hash[:perform_these_actions_on_each_file]
+  end
+
+  # ========================================================================= #
+  # === ask_for_user_input?
+  # ========================================================================= #
+  def ask_for_user_input?
+    @internal_hash[:ask_for_user_input]
+  end
+
+  # ========================================================================= #
+  # === consider_working_through_the_input_files
+  # ========================================================================= #
+  def consider_working_through_the_input_files
+    _ = @internal_hash[:work_on_these_files]
+    if _.empty?
+      opne 'Please provide at the least ony locally existing file as'
+      opne 'input to this class.'
+    else
+      perform_these_actions_on_each_file = perform_these_actions_on_each_file?
+      _.each {|this_file|
+        if File.exist? this_file
+          # ================================================================= #
+          # Tell the user which file we will compact next.
+          # ================================================================= #
+          opne "#{rev}Working to compact the following file next: "\
+               "`#{sfile(this_file)}#{rev}`." if be_verbose?
+          dataset = default_readlines(this_file)
+          # ================================================================= #
+          # First, check for ORIGIN; or a FASTA header that begins with '>'.
+          # ================================================================= #
+          if dataset.first
+            first = dataset.first
+            if first.start_with? 'ORIGIN'
+              dataset[0][0,'ORIGIN'.size] = ''
+              dataset[0].lstrip!
+            elsif first.start_with? '>'
+              # =============================================================== #
+              # In this case, chop off the first line completely.
+              # =============================================================== #
+              dataset[0][0, first.index("\n")] = ''
+              dataset[0].lstrip!
+            end
+          end
+          if perform_these_actions_on_each_file.include? :strip_newlines
+            dataset.map! {|entry| entry.strip }
+          end
+          if perform_these_actions_on_each_file.include? :remove_numbers
+            dataset.map! {|entry| entry.delete('[0-9]') }
+          end
+          if perform_these_actions_on_each_file.include? :upcase_the_sequence
+            dataset.map! {|entry| entry.upcase }
+          end
+          if perform_these_actions_on_each_file.include? :eliminate_spaces
+            dataset.map! {|entry| entry.delete(' ') }
+          end
+          # ================================================================= #
+          # Next we tap into class Bioroebe::SanitizeNucleotideSequence
+          # ================================================================= #
+          dataset = Bioroebe::SanitizeNucleotideSequence[dataset] { :do_not_remove_newlines }
+          if ask_for_user_input?
+            opne 'We will now store into the same file, but you must'
+            opne 'type "'+steelblue('y')+'" to confirm, else we cancel:'
+            user_input = $stdin.gets.chomp
+          else
+            user_input = 'yes'
+          end
+          if user_input.start_with? 'y'
+            opne "Now storing into the file `#{sfile(this_file)}`." if be_verbose?
+            write_what_into(dataset, this_file)
+          end
+        end # no else clause is possible here, due to a prior if-check.
+      }
+    end
+  end
+
+  # ========================================================================= #
+  # === run                                                         (run tag)
+  # ========================================================================= #
+  def run
+    menu
+    determine_the_input_files
+    consider_working_through_the_input_files
+  end
+
+  # ========================================================================= #
+  # === Bioroebe::Compacter[]
+  # ========================================================================= #
+  def self.[](i = ARGV)
+    new(i)
+  end
+
+end
+
+# =========================================================================== #
+# === Bioroebe.compacter
+#
+# Note that this variant will NEVER ask for user-input of the
+# Bioroebe::Compacter class.
+# =========================================================================== #
+def self.compacter(
+    i = ARGV
+  )
+  Bioroebe::Compacter.new(i) { :do_not_ask_for_user_input }
+end 
+
+end
+
+if __FILE__ == $PROGRAM_NAME
+  Bioroebe::Compacter.new(ARGV)
+end # fastacompact

data/lib/bioroebe/utility_scripts/compseq/compseq.rb

@@ -47,11 +47,6 @@ class Compseq < ::Bioroebe::CommandlineApplication # === Bioroebe::Compseq
   # ========================================================================= #
   COLOURIZE_CpG_ISLANDS = true
 
-  # ========================================================================= #
-  # === NAMESPACE
-  # ========================================================================= #
-  NAMESPACE = inspect
-
   # ========================================================================= #
   # === EXPECTED_FREQUENCY
   # ========================================================================= #
@@ -108,10 +103,7 @@ class Compseq < ::Bioroebe::CommandlineApplication # === Bioroebe::Compseq
   # ========================================================================= #
   def reset
     super()
-    # ======================================================================= #
-    # === @namespace
-    # ======================================================================= #
-    @namespace = NAMESPACE
+    infer_the_namespace
     # ======================================================================= #
     # === @identifier
     # ======================================================================= #

data/lib/bioroebe/utility_scripts/consensus_sequence.rb

@@ -35,7 +35,7 @@ class ConsensusSequence < ::Bioroebe::CommandlineApplication # === Bioroebe::Con
       # ===================================================================== #
       when :do_not_run_yet_and_be_quiet
         run_already = false
-        @be_verbose = false
+        set_be_quiet
         @report_the_frequencies = false
       # ===================================================================== #
       # === :do_not_run_yet
@@ -46,7 +46,7 @@ class ConsensusSequence < ::Bioroebe::CommandlineApplication # === Bioroebe::Con
       # === :be_quiet
       # ===================================================================== #
       when :be_quiet
-        @be_verbose = false
+        set_be_quiet
         @report_the_frequencies = false
       end
     end
@@ -66,7 +66,7 @@ class ConsensusSequence < ::Bioroebe::CommandlineApplication # === Bioroebe::Con
     # ======================================================================= #
     # === @be_verbose
     # ======================================================================= #
-    @be_verbose = true
+    set_be_verbose
     # ======================================================================= #
     # === @report_the_frequencies
     # ======================================================================= #
@@ -195,7 +195,7 @@ class ConsensusSequence < ::Bioroebe::CommandlineApplication # === Bioroebe::Con
   # === report_on_which_sequences_we_will_operate
   # ========================================================================= #
   def report_on_which_sequences_we_will_operate(
-      be_verbose = @be_verbose
+      be_verbose = be_verbose?
     )
     case be_verbose
     when :be_verbose
@@ -217,7 +217,7 @@ class ConsensusSequence < ::Bioroebe::CommandlineApplication # === Bioroebe::Con
   # === report_the_consensus_sequence
   # ========================================================================= #
   def report_the_consensus_sequence
-    if @be_verbose
+    if be_verbose?
       the_consensus_sequence = consensus_sequence?
       erev 'The consensus sequence is: '
       e
@@ -238,7 +238,7 @@ class ConsensusSequence < ::Bioroebe::CommandlineApplication # === Bioroebe::Con
         try_to_find_and_report_an_even_simpler_consensus_sequence(
           the_consensus_sequence
         )
-        if @be_verbose
+        if be_verbose?
           erev 'Finally, we will display the above findings in a'
           erev 'shorter variant:'
           report_on_which_sequences_we_will_operate

data/lib/bioroebe/utility_scripts/create_batch_entrez_file.rb

@@ -44,11 +44,6 @@ https://www.ncbi.nlm.nih.gov/nuccore/NZ_CP007215.2
 https://www.ncbi.nlm.nih.gov/nuccore/NC_003197.2
 '
 
-  # ========================================================================= #
-  # === NAMESPACE
-  # ========================================================================= #
-  NAMESPACE = inspect
-
   # ========================================================================= #
   # === initialize
   # ========================================================================= #
@@ -66,10 +61,7 @@ https://www.ncbi.nlm.nih.gov/nuccore/NC_003197.2
   # ========================================================================= #
   def reset
     super()
-    # ======================================================================= #
-    # === @namespace
-    # ======================================================================= #
-    @namespace = NAMESPACE
+    infer_the_namespace
   end
 
   # ========================================================================= #

data/lib/bioroebe/utility_scripts/dot_alignment.rb

@@ -21,11 +21,6 @@ module Bioroebe
 
 class DotAlignment < ::Bioroebe::CommandlineApplication # === Bioroebe::DotAlignment
 
-  # ========================================================================= #
-  # === NAMESPACE
-  # ========================================================================= #
-  NAMESPACE = inspect
-
   # ========================================================================= #
   # === initialize
   # ========================================================================= #
@@ -43,10 +38,7 @@ class DotAlignment < ::Bioroebe::CommandlineApplication # === Bioroebe::DotAlign
   # ========================================================================= #
   def reset
     super()
-    # ======================================================================= #
-    # === @namespace
-    # ======================================================================= #
-    @namespace = NAMESPACE
+    infer_the_namespace
     # ======================================================================= #
     # === @may_we_continue
     # ======================================================================= #

data/lib/bioroebe/utility_scripts/move_file_to_its_correct_location.rb

@@ -42,10 +42,7 @@ class MoveFileToItsCorrectLocation < ::Bioroebe::CommandlineApplication
   # ========================================================================= #
   def reset
     super()
-    # ======================================================================= #
-    # === @namespace
-    # ======================================================================= #
-    @namespace = NAMESPACE
+    infer_the_namespace
     # ======================================================================= #
     # === @try_to_relocate_the_file_to_an_appropriate_subdirectory
     #

data/lib/bioroebe/utility_scripts/parse_taxonomy.rb

@@ -64,7 +64,7 @@ class ParseTaxonomy < ::Bioroebe::CommandlineApplication # === Bioroebe::ParseTa
     # ======================================================================= #
     # === @be_verbose
     # ======================================================================= #
-    @be_verbose = false
+    set_be_quiet
   end
 
   # ========================================================================= #
@@ -112,7 +112,7 @@ class ParseTaxonomy < ::Bioroebe::CommandlineApplication # === Bioroebe::ParseTa
   # ========================================================================= #
   def read_in_url(i = @url)
     require 'open-uri'
-    e "Reading in from #{sfancy(i.to_s)} now." if @be_verbose
+    e "Reading in from #{sfancy(i.to_s)} now." if be_verbose?
     @dataset = URI.open(i).read
   end
 

data/lib/bioroebe/utility_scripts/permutations.rb

@@ -4,9 +4,9 @@
 # =========================================================================== #
 # === Bioroebe::Permutations
 #
-# Pass a number into this class. The number should be smaller than 8,
+# Pass a number into this class. This number should be smaller than 8,
 # so 1, 2, 3 etc... is all ok; 8 is beginning to become difficult, as it
-# leads to more calculations and thus more time it takes to display the
+# leads to more calculations and thus more time required to display the
 # result - after all there are already 40321 lines generated when the
 # number is 8, so the higher the input, the more lines generated.
 #
@@ -81,6 +81,13 @@ class Permutations < ::Bioroebe::CommandlineApplication # === Bioroebe::Permutat
     @result
   end; alias result result? # === result
 
+  # ========================================================================= #
+  # === report
+  # ========================================================================= #
+  def report
+    e @result
+  end
+
   # ========================================================================= #
   # === run                                                         (run tag)
   # ========================================================================= #
@@ -99,13 +106,6 @@ class Permutations < ::Bioroebe::CommandlineApplication # === Bioroebe::Permutat
     report
   end
 
-  # ========================================================================= #
-  # === report
-  # ========================================================================= #
-  def report
-    e @result
-  end
-
   # ========================================================================= #
   # === Bioroebe::Permutations[]
   # ========================================================================= #
@@ -115,6 +115,33 @@ class Permutations < ::Bioroebe::CommandlineApplication # === Bioroebe::Permutat
 
 end
 
+# =========================================================================== #
+# === Bioroebe.unique_two_paired_permutations
+#
+# This method will return all unique permutations of a given Array.
+#
+# The method will return an Array of all possible, unique combinations,
+# as a pair of two entries per inner array.
+#
+# Invocation example:
+#
+#   Bioroebe.unique_two_paired_permutations # Would yield an Array of 28 members.
+#
+# =========================================================================== #
+def self.unique_two_paired_permutations(
+    array = %w( B E G I K L P S )
+  )
+  # ========================================================================= #
+  # Example: ["B", "E", "G", "I", "K", "L", "P", "S"]
+  # ========================================================================= #
+  all_permutations = array.permutation(2).to_a
+  all_unique_permutations = []
+  all_permutations.each {|this_array|
+    all_unique_permutations << this_array unless all_unique_permutations.include?(this_array.reverse)
+  }
+  return all_unique_permutations
+end
+
 # =========================================================================== #
 # === Bioroebe.permutations
 #

data/lib/bioroebe/utility_scripts/showorf/constants.rb

@@ -6,11 +6,6 @@ module Bioroebe
 
 class ShowOrf < ::Bioroebe::CommandlineApplication
 
-  # ========================================================================= #
-  # === NAMESPACE
-  # ========================================================================= #
-  NAMESPACE = inspect
-
   # ========================================================================= #
   # === COLOURIZE_VERTICAL_TOKEN
   # ========================================================================= #

data/lib/bioroebe/utility_scripts/showorf/reset.rb

@@ -11,10 +11,7 @@ class ShowOrf < ::Bioroebe::CommandlineApplication
   # ========================================================================= #
   def reset
     super()
-    # ======================================================================= #
-    # === @namespace
-    # ======================================================================= #
-    @namespace = NAMESPACE
+    infer_the_namespace
     # ======================================================================= #
     # === @show_these_frames
     #

data/lib/bioroebe/version/version.rb

@@ -9,7 +9,7 @@ module Bioroebe
   # ========================================================================= #
   # === VERSION
   # ========================================================================= #
-  VERSION = '0.10.80'
+  VERSION = '0.12.24'
 
   # ========================================================================= #
   # === LAST_UPDATE
@@ -17,7 +17,7 @@ module Bioroebe
   # This variable keeps track as to when the bioroebe project was last
   # updated. The notation is: DD.MM.YYYY
   # ========================================================================= #
-  LAST_UPDATE = '24.06.2022'
+  LAST_UPDATE = '12.09.2023'
 
   # ========================================================================= #
   # === URL_TO_THE_DOCUMENTATION