bioroebe 0.10.80

data/lib/bioroebe/ext/sequence.cpp

@@ -0,0 +1,162 @@
+/*
+ * A sequence object carries a sequence of nucleotides or
+ * amino acids.
+ *
+ * File:
+ *
+ *   sequence.cpp
+ *
+ */
+
+#include <iostream>
+#include <algorithm>
+#include <string>
+
+using namespace std;
+
+class Sequence {
+
+private:
+
+  /* Store the sequence in this string here */
+  string sequence { "" };
+
+public:
+
+  /*
+   * A constructor is a special member function that must be defined
+   * with the same name as the class, so that the compiler can
+   * distinguish it from the class's other member functions.
+   */
+
+  // constructor initializes courseName with string supplied as argument
+  Sequence(string i = "ATCG")
+  {
+     set_sequence(i); // call set function to initialize courseName
+  } // end GradeBook constructor
+
+  string type { "nucleotide" }; /* The type can be "nucleotide" or "aminoacids". */
+
+  /*
+   * report_length()
+   */
+  void report_length() {
+    cout << "The length of this Sequence object is" << endl;
+  }
+
+  /*
+   * set_sequence()
+   *
+   * Use this method to assign to a sequence.
+   *
+   */
+  string set_sequence(string i) {
+// TODO:
+// - Must keep track of its size, through length.
+
+    /*
+     * Only store an uppercased sequence.
+     */
+    std::transform(
+      i.begin(), i.end(), i.begin(), ::toupper
+    );
+    sequence = i; /* Set the value here. */
+    return sequence;
+  }
+  /*
+   * query_sequence()
+   */
+  string query_sequence() {
+    return sequence;
+  }
+
+  /* Define an "alias" to the above method: */
+  string return_sequence() {
+    return query_sequence();
+  }
+
+  /*
+   * query_length()
+   *
+   * The C++ Standard Library's string class defines
+   * a member function length that returns the number
+   * of characters in a string object.
+   *
+   */
+  int query_length() {
+    return sequence.length();
+  }
+
+  /*
+   * obtain_user_input()
+   *
+   * This method is used to interactively ask the user for
+   * a new sequence that is to be used.
+   */
+  void obtain_user_input() {
+    string new_sequence;
+    cout << "Please enter the new sequence next:" << endl;
+    getline(cin, new_sequence);
+    set_sequence(new_sequence);
+  }
+
+  /*
+   * return_type()
+   */
+  string return_type() {
+    return type;
+  }
+
+  /*
+   * report_type()
+   *
+   * This method will report, aka output, the type of the sequence
+   * object at hand.
+   *
+   */
+  void report_type() {
+    cout << "The type of this Sequence object is a " <<
+            return_type() << "." << endl;
+  }
+  /*
+   * report_sequence()
+   *
+   * This method will report, aka output, the sequence at hand.
+   *
+   */
+  void report_sequence() {
+    cout << "The sequence is: "+return_sequence() << endl;
+  }
+
+  /*
+   * report_size()
+   */
+  void report_size() {
+    cout << "The size of this sequence is CURRENTLY NOT IMPLEMENTED." << endl;
+  }
+
+};
+
+/*
+ * Next comes some testing-code, stored in main(). It should be
+ * disabled when we distribute the class.
+ *
+ */
+
+/*
+
+int main() {
+
+  Sequence sequence_object;
+
+  sequence_object.report_size();
+  sequence_object.report_type();
+
+  sequence_object.obtain_user_input();
+  sequence_object.report_sequence();
+
+  return EXIT_SUCCESS;
+} 
+*/
+
+// run SAVE_AS_IS

data/lib/bioroebe/fasta_and_fastq/README.md

@@ -0,0 +1,6 @@
+Code that may specifically be used for .fastq and .fasta files, including
+manipulation of said files, may reside in this directory here.
+
+The file "one_random_fastq_entry.fastq" can be used as a sample file
+for FASTQ files. Since as of August 2021 this file now resides in
+another subdirectory, though, at bioroebe/data/fastaq/.

data/lib/bioroebe/fasta_and_fastq/autocorrect_the_name_of_this_fasta_file.rb

@@ -0,0 +1,88 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# === Bioroebe::AutocorrectTheNameOfThisFastaFile
+#
+# This class can autocorrect the filename of a local .fasta file.
+#
+# It will take the header line for this.
+#
+# Usage example:
+#
+#   Bioroebe::AutocorrectTheNameOfThisFastaFile.new(ARGV)
+#
+# =========================================================================== #
+# require 'bioroebe/fasta_and_fastq/autocorrect_the_name_of_this_fasta_file.rb'
+# =========================================================================== #
+require 'bioroebe/base/commandline_application/commandline_application.rb'
+
+module Bioroebe
+
+class AutocorrectTheNameOfThisFastaFile < CommandlineApplication # === Bioroebe::AutocorrectTheNameOfThisFastaFile
+
+  require 'bioroebe/fasta_and_fastq/parse_fasta/parse_fasta.rb'
+
+  # ========================================================================= #
+  # === NAMESPACE
+  # ========================================================================= #
+  NAMESPACE = inspect
+
+  # ========================================================================= #
+  # === initialize
+  # ========================================================================= #
+  def initialize(
+      commandline_arguments = nil,
+      run_already           = true
+    )
+    reset
+    set_commandline_arguments(
+      commandline_arguments
+    )
+    run if run_already
+  end
+
+  # ========================================================================= #
+  # === reset                                                     (reset tag)
+  # ========================================================================= #
+  def reset
+    super()
+    # ======================================================================= #
+    # === @namespace
+    # ======================================================================= #
+    @namespace = NAMESPACE
+  end
+
+  # ========================================================================= #
+  # === run                                                         (run tag)
+  # ========================================================================= #
+  def run
+    _ = first_argument?
+    if _ and File.exist?(_)
+      original_filename = _
+      dataset = Bioroebe::ParseFasta.new(_) { :be_quiet }
+      new_name = dataset.first_key? # such as: "sp|O58412|VORD_PYRHO (frag)"
+      if new_name.include? ' '
+        new_name = new_name.split(' ').first
+      end
+      new_name = new_name.tr('|','_')+'.fasta'
+      opnn; erev 'Now renaming from `'+sfile(original_filename)+
+                 rev+'` to `'+sfile(new_name)+'`.'
+      rename(original_filename, new_name)
+    else
+      no_file_exists_at(_)
+    end
+  end
+
+  # ========================================================================= #
+  # === Bioroebe::AutocorrectTheNameOfThisFastaFile[]
+  # ========================================================================= #
+  def self.[](i = '')
+    new(i)
+  end
+
+end; end
+
+if __FILE__ == $PROGRAM_NAME
+  Bioroebe::AutocorrectTheNameOfThisFastaFile.new(ARGV)
+end # autocorrectthenameofthisfastafile

data/lib/bioroebe/fasta_and_fastq/compact_fasta_file/compact_fasta_file.rb

@@ -0,0 +1,151 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# === Bioroebe::CompactFastaFile
+#
+# This class will simply take a local .fasta file and put all its sequence
+# body into a single line. In other words, all newlines will be removed
+# (if there are any).
+#
+# Note that this will overwrite the existing file, so be careful when
+# using this class.
+# =========================================================================== #
+# require 'bioroebe/fasta_and_fastq/compact_fasta_file/compact_fasta_file.rb'
+# Bioroebe::CompactFastaFile.new(ARGV)
+# =========================================================================== #
+require 'bioroebe/base/commandline_application/commandline_application.rb'
+
+module Bioroebe
+
+class CompactFastaFile < ::Bioroebe::CommandlineApplication # === Bioroebe::CompactFastaFile
+
+  # ========================================================================= #
+  # === NAMESPACE
+  # ========================================================================= #
+  NAMESPACE = inspect
+
+  # ========================================================================= #
+  # === initialize
+  # ========================================================================= #
+  def initialize(
+      i           = ARGV,
+      run_already = true,
+      &block
+    )
+    reset
+    # ======================================================================= #
+    # === Handle blocks next
+    # ======================================================================= #
+    if block_given?
+      yielded = yield
+      # ===================================================================== #
+      # First handle Symbols.
+      # ===================================================================== #
+      case yielded
+      # ===================================================================== #
+      # === :be_verbose
+      # ===================================================================== #
+      when :be_verbose,
+           :verbose
+        set_be_verbose_and_report_the_sequence
+      # ===================================================================== #
+      # === :be_quiet
+      # ===================================================================== #
+      when :be_quiet,
+           :be_silent
+        be_quiet
+      end
+    end
+    set_commandline_arguments(i)
+    case run_already
+    # ======================================================================= #
+    # === :dont_run_yet
+    # ======================================================================= #
+    when :dont_run_yet
+      run_already = false
+    end
+    run if run_already
+  end
+
+  # ========================================================================= #
+  # === reset                                                     (reset tag)
+  # ========================================================================= #
+  def reset
+    # ======================================================================= #
+    # === @namespace
+    # ======================================================================= #
+    @namespace = NAMESPACE
+    # ======================================================================= #
+    # === @user_configuration
+    # ======================================================================= #
+    @user_configuration = {}
+    # ======================================================================= #
+    # === :do_upcase_the_sequence
+    # ======================================================================= #
+    @user_configuration[:do_upcase_the_sequence] = false
+    enable_colours
+  end
+
+  # ========================================================================= #
+  # === do_upcase_the_sequence
+  # ========================================================================= #
+  def do_upcase_the_sequence
+    @user_configuration[:do_upcase_the_sequence] = true
+  end
+
+  # ========================================================================= #
+  # === do_upcase_the_sequence?
+  # ========================================================================= #
+  def do_upcase_the_sequence?
+    @user_configuration[:do_upcase_the_sequence]
+  end
+
+  # ========================================================================= #
+  # === menu                                                       (menu tag)
+  # ========================================================================= #
+  def menu(
+      i = commandline_arguments_starting_with_leading_hyphens?
+    )
+    if i.is_a? Array
+      i.each {|entry| menu(entry) }
+    else
+      case i # case tag
+      # ===================================================================== #
+      # === compactfile --upcase small_proline_rich_protein_4.fasta
+      # ===================================================================== #
+      when /^-?-?upcase$/i
+        do_upcase_the_sequence
+      end
+    end
+  end
+
+  # ========================================================================= #
+  # === run                                                         (run tag)
+  # ========================================================================= #
+  def run
+    require 'bioroebe/fasta_and_fastq/parse_fasta/parse_fasta.rb'
+    this_file = first_non_hyphened_argument?.to_s
+    menu(
+      commandline_arguments_starting_with_leading_hyphens?
+    )
+    _ = Bioroebe::ParseFasta.new(this_file) { :be_quiet }
+    what = '> '+_.header?+"\n".dup
+    if do_upcase_the_sequence?
+      what << _.body?.upcase.delete(' ')
+    else
+      what << _.body?.delete(' ')
+    end
+    into = this_file
+    opne 'Storing into the file:'
+    e
+    e "  #{sfile(into)}"
+    e
+    write_what_into(what, into)
+  end
+
+end; end
+
+if __FILE__ == $PROGRAM_NAME
+  Bioroebe::CompactFastaFile.new(ARGV)
+end # compactfastafile

data/lib/bioroebe/fasta_and_fastq/display_how_many_fasta_entries_are_in_this_directory.rb

@@ -0,0 +1,111 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# === Bioroebe::DisplayHowManyFastaEntriesAreInThisDirectory
+#
+# This class will simply display the Fasta-entries as a table,
+# on the commandline.
+#
+# Usage example:
+#
+#   Bioroebe::DisplayHowManyFastaEntriesAreInThisDirectory.new
+#
+# =========================================================================== #
+# require 'bioroebe/fasta_and_fastq/display_how_many_fasta_entries_are_in_this_directory.rb'
+# =========================================================================== #
+require 'bioroebe/base/commandline_application/commandline_application.rb'
+
+module Bioroebe
+
+class DisplayHowManyFastaEntriesAreInThisDirectory < ::Bioroebe::CommandlineApplication # === Bioroebe::DisplayHowManyFastaEntriesAreInThisDirectory
+
+  # ========================================================================= #
+  # === initialize
+  # ========================================================================= #
+  def initialize(
+      commandline_arguments = nil,
+      run_already           = true
+    )
+    reset
+    set_commandline_arguments(
+      commandline_arguments
+    )
+    run if run_already
+  end
+
+  # ========================================================================= #
+  # === reset                                                     (reset tag)
+  # ========================================================================= #
+  def reset
+    super()
+    # ======================================================================= #
+    # === @array_count_from_these_files
+    #
+    # Both .fa and .fasta are considered to be valid FASTA files for this
+    # class.
+    # ======================================================================= #
+    @array_count_from_these_files = (
+      Dir['*.fasta']+
+      Dir['*.fa']
+    ).flatten
+  end
+
+  # ========================================================================= #
+  # === run                                                         (run tag)
+  # ========================================================================= #
+  def run
+    if @array_count_from_these_files.empty?
+      e "No fasta file could be found in #{sdir(return_pwd)}."
+    else
+      # ===================================================================== #
+      # We have to sort this:
+      # ===================================================================== #
+      sorted = @array_count_from_these_files.sort_by {|line|
+        middle_part = line.split('.')[1].to_i
+        # =================================================================== #
+        # If middle_part is 0, then we assume it is not a number.
+        # =================================================================== #
+        middle_part = 100 if middle_part == 0
+        middle_part
+      }
+      e
+      e steelblue(sorted.size.to_s+' FASTA files')+
+        rev+' '+was_or_were(sorted.size)+
+        ' found in this directoy:'
+      e
+      sorted.each {|this_entry|
+        index = this_entry.split('.')[1]
+        if File.exist? this_entry
+          dataset = File.readlines(this_entry)
+          n_fasta_headers_in_that_file = dataset.select {|line| line.start_with? '>' }.size
+          # ================================================================= #
+          # Remove the file-extension part.
+          # ================================================================= #
+          basename = File.basename(this_entry).
+                     sub(/#{File.extname(this_entry)}$/,'')
+          if basename.include? '.'
+            basename = basename.split('.')[0]
+          end
+          result = '  '+index.to_s.rjust(2)+' '+
+                   basename+' '+
+                   n_fasta_headers_in_that_file.to_s.rjust(4)
+          e result
+        end
+      }
+      e
+    end
+  end
+
+  # ========================================================================= #
+  # === Bioroebe::DisplayHowManyFastaEntriesAreInThisDirectory[]
+  # ========================================================================= #
+  def self.[](i = '')
+    new(i)
+  end
+
+end; end
+
+if __FILE__ == $PROGRAM_NAME
+  Bioroebe::DisplayHowManyFastaEntriesAreInThisDirectory.new(ARGV)
+end # displayhowmanyfastaentriesareinthisdirectory