bioroebe 0.10.80

data/lib/bioroebe/project/project.rb

@@ -0,0 +1,264 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# === project.rb
+#
+# This file contains code that can be used for determining the
+# base-directory of the BioRoebe project, as well as some constants
+# for directories that directly depend on this, such as the
+# constant called TOPLEVEL_METHODS_DIRECTORY.
+#
+# This file also contains a constant that points to the
+# yaml-subdirectory of the bioroebe project, as well as
+# for the Taxonomy submodule within module Bioroebe.
+#
+# It is recommended to name constants pointing to directories
+# via a trailing "_DIRECTORY". In the past, this was often
+# "_DIR" instead, but this is a bit less elegant to read
+# in the code, in my opinion, so the longer variant should
+# be used instead..
+#
+# Note that since past the year 2020 it is better to make use of
+# methods rather than rely on hardcoded constants. While the old
+# constants will not be deprecated (at the least not this time),
+# it has been determined that using methods is, even though they
+# are a tiny bit slower, more flexible, leading to code that
+# can be maintained more easily as well in the long run.
+# =========================================================================== #
+# require 'bioroebe/project/project.rb'
+# Bioroebe.project_yaml_directory?
+# Bioroebe::CODON_TABLES_DIRECTORY
+# =========================================================================== #
+module Bioroebe
+
+# =========================================================================== #
+# === @project_base_directory
+#
+# This variable will keep track of the base directory for the project
+# as-is.
+# =========================================================================== #
+@project_base_directory =
+  File.absolute_path("#{__dir__}/..")+'/'
+
+# =========================================================================== #
+# === Bioroebe.project_base_dir?
+#
+# This method will return the value of @project_base_directory - that is
+# where the files that belong to Bioroebe are assumed to reside.
+# =========================================================================== #
+def self.project_base_dir?
+  @project_base_directory
+end; self.instance_eval { alias project_dir?            project_base_dir? } # === Bioroebe.project_dir?
+     self.instance_eval { alias project_base_directory? project_base_dir? } # === Bioroebe.project_base_directory?
+     self.instance_eval { alias project_directory?      project_base_dir? } # === Bioroebe.project_directory?
+
+# =========================================================================== #
+# === Bioroebe.project_yaml_directory?
+#
+# This method will return the path to the yaml-directory of the bioroebe
+# project. It will be a dynamic result rather than make use of the
+# above constant.
+# =========================================================================== #
+def self.project_yaml_directory?(i = @project_base_directory)
+  # ========================================================================= #
+  # The following code allows us to default to the HOME directory of the
+  # user.
+  # ========================================================================= #
+  use_this_as_the_project_base_directory = i
+  unless File.directory? use_this_as_the_project_base_directory
+    use_this_as_the_project_base_directory = File.expand_path('~')
+  end
+  return "#{use_this_as_the_project_base_directory}yaml/"
+end; self.instance_eval { alias yaml_dir?       project_yaml_directory? } # === Bioroebe.yaml_dir?
+     self.instance_eval { alias yaml_directory? project_yaml_directory? } # === Bioroebe.yaml_directory?
+
+# =========================================================================== #
+# === FILE_EXPAND_CD_ALIASES
+# =========================================================================== #
+FILE_EXPAND_CD_ALIASES =
+  "#{@project_base_directory}hash_expand_cd_aliases.rb"
+
+# =========================================================================== #
+# === CONFIGURATION_DIRECTORY
+# =========================================================================== #
+CONFIGURATION_DIRECTORY = 
+  "#{project_yaml_directory?}configuration/"
+
+# =========================================================================== #
+# === TOPLEVEL_METHODS_DIRECTORY
+# =========================================================================== #
+TOPLEVEL_METHODS_DIRECTORY =
+  "#{@project_base_directory}toplevel_methods/"
+
+# =========================================================================== #
+# === CODON_TABLES_DIRECTORY
+# =========================================================================== #
+CODON_TABLES_DIRECTORY = 
+  "#{@project_base_directory}codon_tables/"
+
+# =========================================================================== #
+# === CLEAVE_AND_DIGEST_DIRECTORY
+# =========================================================================== #
+CLEAVE_AND_DIGEST_DIRECTORY = 
+  "#{@project_base_directory}cleave_and_digest/"
+
+# =========================================================================== #
+# === ELECTRON_MICROSCOPY_DIRECTORY
+# =========================================================================== #
+ELECTRON_MICROSCOPY_DIRECTORY = 
+  "#{@project_base_directory}electron_microscopy/"
+
+# =========================================================================== #
+# === CODON_TABLES_DIRECTORY_FREQUENCY
+# =========================================================================== #  
+CODON_TABLES_DIRECTORY_FREQUENCY =
+  "#{CODON_TABLES_DIRECTORY}frequencies/"
+
+# =========================================================================== #
+# === PDB_DIRECTORY
+# =========================================================================== #
+PDB_DIRECTORY =
+  "#{@project_base_directory}pdb/"
+
+# =========================================================================== #
+# === PARSERS_DIRECTORY
+# =========================================================================== #
+PARSERS_DIRECTORY =
+  "#{@project_base_directory}parsers/"
+
+# =========================================================================== #
+# === ENZYMES_DIRECTORY
+# =========================================================================== #
+ENZYMES_DIRECTORY =
+  "#{@project_base_directory}enzymes/"
+
+# =========================================================================== #
+# === PALINDROMES_DIRECTORY
+# =========================================================================== #
+PALINDROMES_DIRECTORY =
+  "#{@project_base_directory}palindromes/"
+
+# =========================================================================== #
+# === PATTERN_DIRECTORY
+# =========================================================================== #
+PATTERN_DIRECTORY =
+  "#{@project_base_directory}pattern/"
+
+# =========================================================================== #
+# === NUCLEOTIDES_DIRECTORY
+# =========================================================================== #
+NUCLEOTIDES_DIRECTORY =
+  "#{@project_base_directory}nucleotides/"
+
+# =========================================================================== #
+# === COUNT_DIRECTORY
+# =========================================================================== #
+COUNT_DIRECTORY =
+  "#{@project_base_directory}count/"
+
+# =========================================================================== #
+# === AMINOACIDS_DIRECTORY
+# =========================================================================== #
+AMINOACIDS_DIRECTORY =
+  "#{@project_base_directory}aminoacids/"
+
+# =========================================================================== #
+# === BLOSUM_DIRECTORY
+# =========================================================================== #
+BLOSUM_DIRECTORY =
+  "#{project_yaml_directory?}blosum/"
+
+# =========================================================================== #
+# === CALCULATE_DIRECTORY
+# =========================================================================== #
+CALCULATE_DIRECTORY =
+  "#{@project_base_directory}calculate/"
+
+# =========================================================================== #
+# === CODONS_DIRECTORY
+# =========================================================================== #
+CODONS_DIRECTORY =
+  "#{@project_base_directory}codons/"
+
+# =========================================================================== #
+# === DOTPLOTS_DIRECTORY
+# =========================================================================== #
+DOTPLOTS_DIRECTORY =
+  "#{@project_base_directory}dotplots/"
+
+# =========================================================================== #
+# === SEQUENCE_DIRECTORY
+# =========================================================================== #
+SEQUENCE_DIRECTORY =
+  "#{@project_base_directory}sequence/"
+
+# =========================================================================== #
+# === PATHWAYS_DIRECTORY
+#
+# This constant will point to e. g.
+# "/Programs/Ruby/2.6.4/lib/ruby/site_ruby/2.6.0/bioroebe/yaml/pathways/".
+# =========================================================================== #
+PATHWAYS_DIRECTORY =
+  "#{project_yaml_directory?}pathways/"
+
+# =========================================================================== #
+# === BIOROEBE_YAML_AMINOACIDS_DIRECTORY
+# =========================================================================== #
+BIOROEBE_YAML_AMINOACIDS_DIRECTORY =
+  "#{project_yaml_directory?}aminoacids/"
+
+# =========================================================================== #
+# === STRING_MATCHING_DIRECTORY
+# =========================================================================== #
+STRING_MATCHING_DIRECTORY =
+  "#{@project_base_directory}string_matching/"
+
+# =========================================================================== #
+# === FASTA_AND_FASTQ_DIRECTORY
+# =========================================================================== #
+FASTA_AND_FASTQ_DIRECTORY =
+  "#{@project_base_directory}fasta_and_fastq/"
+
+# =========================================================================== #
+# === Bioroebe::Taxonomy
+#
+# Add some code for the Taxonomy submodule next.
+#
+# The Taxonomy submodule will make use of the toplevel yaml/ subdirectory
+# of the Bioroebe project in particular.
+# =========================================================================== #
+module Taxonomy
+
+  # ========================================================================= #
+  # === Bioroebe::Taxonomy::PROJECT_YAML_DIR
+  # ========================================================================= #
+  PROJECT_YAML_DIR = ::Bioroebe.project_base_directory?
+
+  # ========================================================================= #
+  # === Bioroebe::Taxonomy.project_base_dir?
+  #
+  # The Taxonomy project has been fully integrated into the Bioroebe
+  # namespace in the year ~2015.
+  #
+  # This method will return a String such as:
+  #
+  #   "/Programs/Ruby/2.6.4/lib/ruby/site_ruby/2.6.0/bioroebe/taxonomy/"
+  #
+  # ========================================================================= #
+  def self.project_base_dir?
+    "#{::Bioroebe.project_base_directory?}taxonomy/"
+  end
+
+  # ========================================================================= #
+  # === Bioroebe::Taxonomy.project_yaml_dir?
+  # ========================================================================= #
+  def self.project_yaml_dir?
+    Taxonomy::PROJECT_YAML_DIR
+  end
+
+end; end
+
+if __FILE__ == $PROGRAM_NAME
+  puts Bioroebe.project_base_dir?
+end

data/lib/bioroebe/protein_structure/alpha_helix.rb

@@ -0,0 +1,96 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# === Bioroebe::AlphaHelix
+#
+# This class will help us answer some questions related to the
+# Alpha-Helix motif (https://en.wikipedia.org/wiki/Structural_motif).
+#
+# The Alpha-helix is a common secondary structure of proteins and it is
+# righthanded.
+#
+# The pitch of the alpha-helix, which is the vertical distance between one
+# consecutive turn of the helix, 0.54 nm, aka 5.4 Angstrom.
+#
+# A full turn takes 3.6 Aminoacids in an alpha-helix. This will be the
+# basis upon which to do calculations.
+# =========================================================================== #
+# require 'bioroebe/alpha_helix.rb'
+# =========================================================================== #
+require 'bioroebe/base/commandline_application/commandline_application.rb'
+
+module Bioroebe
+
+class AlphaHelix < ::Bioroebe::CommandlineApplication # === Bioroebe::AlphaHelix
+
+  # ========================================================================= #
+  # === PITCH_DISTANCE
+  # ========================================================================= #
+  PITCH_DISTANCE = 0.54 # in nm
+
+  # ========================================================================= #
+  # === FULL_TURN_TAKES_N_AMINO_ACIDS
+  # ========================================================================= #
+  FULL_TURN_TAKES_N_AMINO_ACIDS = 3.6
+
+  # ========================================================================= #
+  # === DEFAULT_HOW_MANY_AMINO_ACIDS
+  # ========================================================================= #
+  DEFAULT_HOW_MANY_AMINO_ACIDS = 60
+
+  # ========================================================================= #
+  # === initialize
+  # ========================================================================= #
+  def initialize(
+      how_many_amino_acids = DEFAULT_HOW_MANY_AMINO_ACIDS
+    )
+    reset
+    set_how_many_amino_acids(how_many_amino_acids)
+    do_the_calculation
+  end
+
+  # ========================================================================= #
+  # === reset
+  # ========================================================================= #
+  def reset
+    super()
+  end
+
+  # ========================================================================= #
+  # === set_how_many_amino_acids
+  # ========================================================================= #
+  def set_how_many_amino_acids(i = DEFAULT_HOW_MANY_AMINO_ACIDS)
+    i = i.first.to_s if i.is_a? Array
+    i = DEFAULT_HOW_MANY_AMINO_ACIDS if i.empty?
+    @how_many_amino_acids = i.to_i
+  end
+
+  # ========================================================================= #
+  # === do_the_calculation
+  # ========================================================================= #
+  def do_the_calculation
+    result = AlphaHelix.length?(@how_many_amino_acids)
+    e result.to_s
+  end
+
+  # ========================================================================= #
+  # === Bioroebe::AlphaHelix.length?
+  # ========================================================================= #
+  def self.length?(i)
+    i = i.join.strip if i.is_a? Array
+    if i.is_a? String
+      i = i.size
+    end
+    # ========================================================================= #
+    # Next calculate that length in an AlphaHelix.
+    # ========================================================================= #
+    length = (i.to_i / FULL_TURN_TAKES_N_AMINO_ACIDS) * PITCH_DISTANCE
+    return length.to_s+' nm'
+  end
+
+end; end
+
+if __FILE__ == $PROGRAM_NAME
+  Bioroebe::AlphaHelix.new(ARGV)
+end # alphahelix

data/lib/bioroebe/protein_structure/helical_wheel.rb

@@ -0,0 +1,205 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# === Bioroebe::HelixWheel
+#
+# Unfortunately, this class does not work very well.
+#
+# Usage examples:
+#
+#   Bioroebe::HelixWheel.new
+#
+# hwheel
+#
+# More specifically this class can be used in this way:
+#
+#   echo DVAGHGQDILIRLFKSH    | ruby_helical_wheel > output.ps
+#   echo 20DVAGHGQDILIRLFKSH  | ruby_helical_wheel > output.ps
+#
+# =========================================================================== #
+# require 'helix_wheel'
+# =========================================================================== #
+require 'bioroebe/base/commandline_application/commandline_application.rb'
+
+module Bioroebe
+
+class HelixWheel < ::Bioroebe::CommandlineApplication # === Bioroebe::HelixWheel
+
+  # ========================================================================= #
+  # === GS_HEADER_START
+  # ========================================================================= #
+  GS_HEADER_START = <<EOF;
+%!PS-Adobe-
+%%BoundingBox: (atend)
+%1 0 0 setrgbcolor
+%newpath
+%37.5 161 moveto 557.5 161 lineto 557.5 681 lineto 37.5 681 lineto
+%closepath stroke
+297.5 421. translate 2 setlinewidth 1 setlinecap
+/Helvetica findfont 20 scalefont setfont 0 0 moveto
+EOF
+
+  HELVETICA_FONT = 'Helvetica'
+  HELVETICA_FONT_BOLD = 'Helvetica-Bold'
+
+  # ========================================================================= #
+  # === FONT
+  #
+  # Define fonts to associate with each amino acid
+  # ========================================================================= #
+  FONT = {}
+  FONT['G'] = HELVETICA_FONT
+  FONT['A'] = HELVETICA_FONT
+  FONT['S'] = HELVETICA_FONT
+  FONT['T'] = HELVETICA_FONT
+  FONT['C'] = HELVETICA_FONT
+  FONT['V'] = HELVETICA_FONT
+  FONT['I'] = HELVETICA_FONT
+  FONT['L'] = HELVETICA_FONT
+  FONT['F'] = HELVETICA_FONT
+  FONT['Y'] = HELVETICA_FONT
+  FONT['P'] = HELVETICA_FONT
+  FONT['M'] = HELVETICA_FONT
+  FONT['W'] = HELVETICA_FONT
+  FONT['H'] = HELVETICA_FONT_BOLD
+  FONT['D'] = HELVETICA_FONT_BOLD
+  FONT['R'] = HELVETICA_FONT_BOLD
+  FONT['Q'] = HELVETICA_FONT_BOLD
+  FONT['K'] = HELVETICA_FONT_BOLD
+  FONT['N'] = HELVETICA_FONT_BOLD
+  FONT['E'] = HELVETICA_FONT_BOLD
+
+  # ========================================================================= #
+  # === initialize
+  # ========================================================================= #
+  def initialize(
+      run_already = true
+    )
+    reset
+    run if run_already
+  end
+
+  # ========================================================================= #
+  # === reset                                                     (reset tag)
+  # ========================================================================= #
+  def reset
+    super()
+    # ======================================================================= #
+    # === @_
+    # ======================================================================= #
+    @_ = ''.dup
+  end
+
+  # ========================================================================= #
+  # === _
+  # ========================================================================= #
+  def _(i)
+    @_ << i
+  end
+
+  # ========================================================================= #
+  # === build_string
+  # ========================================================================= #
+  def build_string
+    _ GS_HEADER_START
+    fetch_input
+  end
+
+  # ========================================================================= #
+  # === cos
+  # ========================================================================= #
+  def cos(i)
+    Math.cos(i)
+  end
+
+  # ========================================================================= #
+  # === sin
+  # ========================================================================= #
+  def sin(i)
+    Math.sin(i)
+  end
+
+  # ========================================================================= #
+  # === fetch_input
+  # ========================================================================= #
+  def fetch_input
+    @input = $stdin.gets.chomp.strip # Read line of input
+    # @input =~ s/\s//g # remove terminal carriage return and blanks
+    use_this_regex = /^(\d+)/
+    if ( @input =~ use_this_regex ) # if input begins with integer
+      resno = $1 # extract it as initial residue number
+    else
+      resno = 1  # if not, set initial residue number = 1
+    end
+    radius =  50 # initialize values for radius,
+    x      =   0
+    y      = -50
+    theta  = -90 # x, y and angle theta
+    # print light gray spiral arc as succession of
+    # line segments, 10 per residue
+    npoints = 10 * (@input.size - 1)
+    _ '0.8 0.8 0.8 setrgbcolor'+N # set colour to light gray
+    _ 'newpath'+N                 # draw spiral arc
+    _ sprintf('%8.3f %8.3f moveto'+N,x,y)
+    (1 .. npoints).each {|d|             # 10 points per residue
+      theta  += 10
+      radius += 0.6          # increase radius and theta
+      x = radius * cos(theta * 0.01747737)   # calculate new value of x
+      y = radius * sin(theta * 0.01747737)   # and y
+      printf("%8.3f %8.3f lineto"+N,x,y)
+    }
+    _ 'stroke'+N
+    # ======================================================================= #
+    # print residues and residue numbers
+    # ======================================================================= #
+    radius =  50 # reinitialize values for radius,
+    x      =   0
+    y      = -50
+    theta  = -90 # x, y and angle theta
+    _ '0 setgray'+N  # set colour to black
+    @input.split('').each {|line|       # loop over characters from input line
+      _ "/FONT{@input} findfont"        # set font appropriate
+      _ ' 20 scalefont setfont'+N       # for this amino acid
+      _ sprintf('%8.3f %8.3f moveto'+N,x,y)  # move to current point
+      _ " (resno@input) stringwidth"    # adjust position to center residue
+      _ ' pop -0.5 mul -7 rmoveto'+N    #    identification on point on spiral
+      _ " (resno@input) show"+N         # print residue number and id
+      _ "% theta resno@input"+N
+      theta += 100
+      radius += 6           #  set new values of angle, radius
+      x = radius * cos(theta * 0.01747737) #  compute new values of x
+      y = radius * sin(theta * 0.01747737) #     and y
+      resno += 1                           #  increase residue number
+    }
+    _ 'showpage'+N             # postscript signals to
+    _ "%%BoundingBox:"         # print
+    xl = 297.5 - 1.05 * radius #   x
+    xr = 297.5 + 1.05 * radius # and
+    yb = 421.  - 1.05 * radius #   y
+    yt = 421.  + 1.05 * radius # limits
+    _ sprintf('%8.3f %8.3f %8.3f %8.3f'+N,xl,xr,yb,yt)
+    _ 'showpage'+N
+    _ '%%EOF'+N              #  and wind up
+  end
+
+  # ========================================================================= #
+  # === show_result
+  # ========================================================================= #
+  def show_result
+    e @_
+  end
+
+  # ========================================================================= #
+  # === run
+  # ========================================================================= #
+  def run
+    build_string
+    show_result
+  end
+
+end; end
+
+if __FILE__ == $PROGRAM_NAME
+  Bioroebe::HelixWheel.new
+end # hwheel

data/lib/bioroebe/raw_sequence/README.md

@@ -0,0 +1,17 @@
+A RawSequence is the base for Bioroebe::Sequence.
+
+It contains the bare minimum of a "sequence object".
+
+Specifically it does NOT contain any colour-related code or 
+output-related methods, so "alias e puts" can NOT be found
+in a RawSequence.
+
+The reason this class was added in 2021 was because there 
+were some bugs in the Bioroebe::Sequence, which in particular
+caused the bioshell interface to show odd formatting results
+on the commandline.
+
+RawSequence bypasses this issue completely, by being much,
+much simpler than Bioroebe::Sequence. Internally
+Bioroebe::Sequence will rely on RawSequence for sequence
+data.