bioroebe 0.10.80

data/lib/bioroebe/requires/require_the_bioroebe_project.rb

@@ -0,0 +1,162 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# This file will require the correct files for the Bioroebe project, in
+# the right order. All Bioroebe components must be loaded up properly.
+# =========================================================================== #
+# require 'bioroebe/requires/require_the_bioroebe_project.rb'
+# =========================================================================== #
+require 'bioroebe/project/project.rb'
+require 'bioroebe/version/version.rb'
+require 'bioroebe/colours/colours.rb'
+
+require 'bioroebe/base/prototype/prototype.rb'
+
+# =========================================================================== #
+# Files under misc/ will be required next:
+# =========================================================================== #
+require 'bioroebe/misc/ruler.rb'
+require 'bioroebe/misc/useful_formulas.rb'
+require 'bioroebe/misc/quiz/three_letter_to_aminoacid.rb'
+
+# =========================================================================== #
+# Load up the readline-support:
+# =========================================================================== #
+require 'bioroebe/readline/readline.rb'
+
+# =========================================================================== #
+# Load up support for blosum matrices:
+# =========================================================================== #
+require 'bioroebe/blosum/blosum.rb'
+
+# =========================================================================== #
+# The regexes:
+# =========================================================================== #
+require 'bioroebe/regexes/regexes.rb'
+
+# =========================================================================== #
+# Require all the bioroebe-related constants next:
+# =========================================================================== #
+require 'bioroebe/requires/require_the_constants.rb'
+
+# =========================================================================== #
+# Require .rb files for yaml-support next:
+# =========================================================================== #
+require 'bioroebe/requires/require_yaml.rb'
+
+# =========================================================================== #
+# The toplevel-methods should be required fairly early:
+# =========================================================================== #
+require 'bioroebe/requires/require_the_toplevel_methods.rb'
+
+# =========================================================================== #
+# Next, require all files under dotplots/:
+# =========================================================================== #
+require 'bioroebe/requires/require_all_dotplot_files.rb'
+
+# =========================================================================== #
+# Require all bioroebe/utility_scripts/ files next:
+# =========================================================================== #
+require 'bioroebe/requires/require_all_utility_scripts_files.rb'
+
+# =========================================================================== #
+# And the aminoacids files come next:
+# =========================================================================== #
+require 'bioroebe/requires/require_all_aminoacids_files.rb'
+
+# =========================================================================== #
+# We need the cleave-and-digest related .rb files as well
+# =========================================================================== #
+require 'bioroebe/requires/require_cleave_and_digest.rb'
+
+# =========================================================================== #
+# Next, all files under bioroebe/calculate/ will be required:
+# =========================================================================== #
+require 'bioroebe/requires/require_all_calculate_files.rb'
+
+# =========================================================================== #
+# The string_matching/*rb files come next:
+# =========================================================================== #
+require 'bioroebe/requires/require_all_string_matching_files.rb'
+
+# =========================================================================== #
+# And the various colour-scheme files:
+# =========================================================================== #
+require 'bioroebe/requires/require_all_colour_scheme_files.rb'
+
+# =========================================================================== #
+# Next, all nucleotides/ files:
+# =========================================================================== #
+require 'bioroebe/requires/require_all_nucleotides_files.rb'
+
+# =========================================================================== #
+# Require all the bioroebe/enzymes/ files next:
+# =========================================================================== #
+require 'bioroebe/requires/require_all_enzymes_files.rb'
+
+# =========================================================================== #
+# Next pull in the palindromes/ files:
+# =========================================================================== #
+require 'bioroebe/requires/require_all_palindromes_files.rb'
+
+# =========================================================================== #
+# Next, we want all bioroebe/pdb/ files:
+# =========================================================================== #
+require 'bioroebe/requires/require_all_pdb_files.rb'
+
+# =========================================================================== #
+# Require all svg/ files next:
+# =========================================================================== #
+require 'bioroebe/requires/require_all_svg_files.rb'
+
+# =========================================================================== #
+# Next the .rb files under electron_microscopy/:
+# =========================================================================== #
+require 'bioroebe/requires/require_all_electron_microscopy_files.rb'
+
+# =========================================================================== #
+# And the bioroebe/parsers/ files have to be required next:
+# =========================================================================== #
+require 'bioroebe/requires/require_all_parser_files.rb'
+
+# =========================================================================== #
+# Pull in all taxonomy/ files next:
+# =========================================================================== #
+require 'bioroebe/requires/require_all_taxonomy_files.rb'
+
+# =========================================================================== #
+# Add all fasta and fastq files next:
+# =========================================================================== #
+require 'bioroebe/requires/require_all_fasta_and_fastq_files.rb'
+
+# =========================================================================== #
+# Require all count/ files next:
+# =========================================================================== #
+require 'bioroebe/requires/require_all_count_files.rb'
+
+# =========================================================================== #
+# Require all files under bioroebe/pattern/  next:
+# =========================================================================== #
+require 'bioroebe/requires/require_all_pattern_files.rb'
+
+# =========================================================================== #
+# Next, add the bioroebe-shell component - the bioshell:
+# =========================================================================== #
+require 'bioroebe/shell/shell.rb'
+
+# =========================================================================== #
+# Next the sequence/ files - this will also require sequence/sequence.rb:
+# =========================================================================== #
+require 'bioroebe/requires/require_all_sequence_files.rb'
+
+# =========================================================================== #
+# Require all the bioroebe/codons/ files:
+# =========================================================================== #
+require 'bioroebe/requires/require_all_codon_files.rb'
+
+# =========================================================================== #
+# And the sinatra-interface:
+# =========================================================================== #
+# require 'bioroebe/requires/require_the_bioroebe_sinatra_components.rb'
+require 'bioroebe/www/sinatra/sinatra.rb'

data/lib/bioroebe/requires/require_the_bioroebe_shell.rb

@@ -0,0 +1,7 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# require 'bioroebe/requires/require_the_bioroebe_shell.rb'
+# =========================================================================== #
+require 'bioroebe/shell/shell.rb'

data/lib/bioroebe/requires/require_the_bioroebe_sinatra_components.rb

@@ -0,0 +1,7 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# require 'bioroebe/requires/require_the_bioroebe_sinatra_components.rb'
+# =========================================================================== #
+require 'bioroebe/www/sinatra/sinatra.rb'

data/lib/bioroebe/requires/require_the_constants.rb

@@ -0,0 +1,23 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# This file will require the other .rb files that reside under
+# bioroebe/constants/.
+# =========================================================================== #
+# require 'bioroebe/requires/require_the_constants.rb'
+# =========================================================================== #
+require 'bioroebe/constants/aminoacids_and_proteins.rb'
+require 'bioroebe/constants/base_directory.rb'
+require 'bioroebe/constants/carriage_return.rb'
+require 'bioroebe/constants/codon_tables.rb'
+require 'bioroebe/constants/database_constants.rb'
+require 'bioroebe/constants/files_and_directories.rb'
+require 'bioroebe/constants/misc.rb'
+require 'bioroebe/constants/newline.rb'
+require 'bioroebe/constants/nucleotides.rb'
+require 'bioroebe/constants/regex.rb'
+require 'bioroebe/constants/roebe.rb'
+require 'bioroebe/constants/row_terminator.rb'
+require 'bioroebe/constants/tabulator.rb'
+require 'bioroebe/constants/urls.rb'

data/lib/bioroebe/requires/require_the_toplevel_methods.rb

@@ -0,0 +1,29 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# Batch-require all .rb files in the toplevel_methods/ subdirectory.
+# =========================================================================== #
+# require 'bioroebe/requires/require_the_toplevel_methods.rb'
+# =========================================================================== #
+module Bioroebe
+
+  # ========================================================================= #
+  # === Bioroebe.require_the_toplevel_methods
+  #
+  # This method can be used to require all toplevel_method files.
+  # ========================================================================= #
+  def self.require_the_toplevel_methods(
+      from_this_dir = TOPLEVEL_METHODS_DIRECTORY
+    )
+    require 'bioroebe/project/project.rb'
+    _ = Dir["#{from_this_dir}*.rb"]
+    _.each {|file|
+      file = "bioroebe/toplevel_methods/#{File.basename(file)}"
+      require file
+    }
+  end; self.instance_eval { alias require_all_toplevel_methods require_the_toplevel_methods } # === Bioroebe.require_all_toplevel_methods
+
+  require_all_toplevel_methods # Call it at once.
+
+end

data/lib/bioroebe/requires/require_yaml.rb

@@ -0,0 +1,94 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# === require_yaml.rb
+#
+# This file can be used to either load up and subsequently use Psych,
+# or load up and then use the older Syck gem.
+#
+# Thus, this file allows us some flexibility for the bioroebe-project.
+#
+# Since as of the year 2019, using psych should have no real major
+# drawbacks really, so the BioRoebe project now defaults to psych
+# past 2019.
+# =========================================================================== #
+# require 'bioroebe/requires/require_yaml.rb'
+# =========================================================================== #
+module Bioroebe
+
+  # ========================================================================= #
+  # === @use_this_yaml_engine
+  #
+  # Please keep the value to this variable as a Symbol at all times.
+  #
+  # The two possible variants are :syck and :psych, currently.
+  #
+  # :psych is the default since some time.
+  # ========================================================================= #
+  @use_this_yaml_engine = :psych # Our default yaml engine to use.
+
+  # ========================================================================= #
+  # === Bioroebe.use_which_yaml_engine?
+  #
+  # Query-method to find out which yaml-engine is used for the
+  # BioRoebe project.
+  # ========================================================================= #
+  def self.use_which_yaml_engine?
+    @use_this_yaml_engine
+  end
+
+  # ========================================================================= #
+  # === Bioroebe.use_which_yaml_engine?
+  #
+  # Query method to determine which yaml engine is in use for the Bioroebe
+  # project.
+  # ========================================================================= #
+  def self.set_use_this_yaml_engine=(i = :default)
+    case i # case tag
+    # ======================================================================= #
+    # === :default
+    # ======================================================================= #
+    when :default
+      i = :psych
+    end
+    @use_this_yaml_engine = i
+  end; self.instance_eval { alias set_yaml_engine set_use_this_yaml_engine= } # === Bioroebe.set_yaml_engine
+
+  # ========================================================================= #
+  # === Bioroebe.load_psych_or_syck
+  # ========================================================================= #
+  def self.load_psych_or_syck(
+      i = @use_this_yaml_engine
+    )
+    case i
+    # ======================================================================= #
+    # === :default
+    # ======================================================================= #
+    when :default,
+         :psych,
+         nil
+      require 'yaml' # This should suffice as-is.
+    else # Else give more information to the user how to get syck.
+      begin
+        require 'syck'
+      rescue LoadError
+        # =================================================================== #
+        # Inform the user how syck can be installed.
+        # =================================================================== #
+        e
+        e 'syck is not available. It can be installed via:'
+        e
+        e '  gem install syck'
+        e
+      end
+    end
+  end
+
+  # ========================================================================= #
+  # Load psych or syck at once, whenever this file is loaded. By default
+  # this will load up psych past 2019.
+  # ========================================================================= #
+  load_psych_or_syck
+
+end

data/lib/bioroebe/sequence/alignment.rb

@@ -0,0 +1,214 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# === Bioroebe::Alignment
+#
+# This class, when given a file, will read the lines in that file and
+# then perform an alignment profile. The entries in that line are
+# assumed to be nucleotides, e. g. a typical FASTA listing, without
+# the header.
+#
+# The consensus sequence will also be shown.
+#
+# To test this class for a given input have a look at the end of
+# this file.
+#
+# Usage example:
+#
+#   Bioroebe::Alignment.new(ARGV)
+#
+# =========================================================================== #
+# require 'bioroebe/sequence/alignment.rb'
+# Bioroebe::Alignment.new(ARGV)
+# =========================================================================== #
+require 'bioroebe/base/commandline_application/commandline_application.rb'
+
+module Bioroebe
+
+class Alignment < ::Bioroebe::CommandlineApplication # === Bioroebe::Alignment
+
+  # ========================================================================= #
+  # === initialize
+  # ========================================================================= #
+  def initialize(
+      commandline_arguments = nil,
+      run_already           = true
+    )
+    reset
+    set_commandline_arguments(
+      commandline_arguments
+    )
+    case run_already
+    # ======================================================================= #
+    # === :do_not_run
+    # ======================================================================= #
+    when :do_not_run,
+         :do_not_run_yet
+      run_already = false
+    end
+    run if run_already
+  end
+
+  # ========================================================================= #
+  # === reset                                                     (reset tag)
+  # ========================================================================= #
+  def reset
+    super()
+  end
+
+  # ========================================================================= #
+  # === use_this_as_input
+  #
+  # This method is assumed to split a String into an Array, before then
+  # passing this into work_on_this_array().
+  # ========================================================================= #
+  def use_this_as_input(i)
+    work_on_this_array(
+      i.split("\n")
+    )
+  end
+
+  # ========================================================================= #
+  # === work_on_this_array
+  # ========================================================================= #
+  def work_on_this_array(i)
+    @array_A = []
+    @array_T = []
+    @array_C = []
+    @array_G = []
+    if i.respond_to? :first
+      first = i.first
+    else
+      first = i
+    end
+    first.size.times {|index|
+      hash = Hash.new(0)
+      hash['A'] = 0
+      hash['T'] = 0
+      hash['C'] = 0
+      hash['G'] = 0
+      # ===================================================================== #
+      # ["A", "C", "A", "A", "C", "C", "A", "G", "G", "G"]
+      # ===================================================================== #
+      vertical_row = i.map {|entry| entry[index] }
+      vertical_row.each {|key| hash[key] += 1 }
+      # ===================================================================== #
+      # hash may be: { "A"=>4, "C"=>3, "G"=>3 }
+      # ===================================================================== #
+      hash.each_pair {|key, value|
+        total_count = vertical_row.size
+        case key
+        when 'A'
+          @array_A << (value.to_i) / total_count.to_f
+        when 'T'
+          @array_T << (value.to_i) / total_count.to_f
+        when 'C'
+          @array_C << (value.to_i) / total_count.to_f
+        when 'G'
+          @array_G << (value.to_i) / total_count.to_f
+        end
+      }
+    }
+    print_row_for('A', @array_A)
+    print_row_for('T', @array_T)
+    print_row_for('C', @array_C)
+    print_row_for('G', @array_G)
+    print_consensus_sequence
+  end
+
+  # ========================================================================= #
+  # === print_consensus_sequence
+  # ========================================================================= #
+  def print_consensus_sequence
+    e
+    erev 'The consensus sequence, derived from the above matrix, is:'
+    e
+    print '  '
+    @array_G.size.times {|index|
+      hash = {}
+      hash['A'] = @array_A[index]
+      hash['T'] = @array_T[index]
+      hash['C'] = @array_C[index]
+      hash['G'] = @array_G[index]
+      _ = hash.max_by {|key, value| value }
+      print mediumslateblue(_[0])+rev # Print each individual character here.
+    }
+    e Colours.rev
+    e
+  end
+
+  # ========================================================================= #
+  # === consensus_sequence?
+  # ========================================================================= #
+  def consensus_sequence?
+    _ = ''.dup
+    @array_G.size.times {|index|
+      hash = {}
+      hash['A'] = @array_A[index]
+      hash['T'] = @array_T[index]
+      hash['C'] = @array_C[index]
+      hash['G'] = @array_G[index]
+      max = hash.max_by {|key, value| value }
+      _ << max.first
+    }
+    return _
+  end
+
+  # ========================================================================= #
+  # === print_row_for
+  # ========================================================================= #
+  def print_row_for(
+      i = 'A', array = @array_A
+    )
+    print "#{rev}  #{i}: " # First output the identifier, e. g. "A" or "T" and so forth.
+    array.each {|entry| # entry is a Float.
+      print steelblue(entry.round(1).to_s.ljust(4))
+    }; e
+  end
+
+  # ========================================================================= #
+  # === run                                                         (run tag)
+  # ========================================================================= #
+  def run
+    first = first?
+    # ======================================================================= #
+    # === If a file was given, read in that file next.
+    # ======================================================================= #
+    if first and File.exist?(first)
+      dataset = file_readlines(first).map(&:strip)
+      work_on_this_array(dataset)
+    elsif first.nil? or first.empty?
+      erev 'Using the __END__ dataset of '+sfile(__FILE__)+rev+'.'
+      e
+      work_on_this_array(
+        DATA.read.split("\n")
+      )
+    else
+      work_on_this_array(commandline_arguments?)
+    end
+  end
+
+  # ========================================================================= #
+  # === Bioroebe::Alignment[]
+  # ========================================================================= #
+  def self.[](i = '')
+    new(i)
+  end
+
+end; end
+
+if __FILE__ == $PROGRAM_NAME
+  Bioroebe::Alignment.new(ARGV)
+end # alignment
+__END__
+ACTCC
+CACCA
+AGCCA
+AACGC
+CGCAT
+CGACC
+ACCGC
+GGCCG
+GGCGT
+GGCGT