bioroebe 0.10.80

data/lib/bioroebe/fasta_and_fastq/show_fasta_statistics.rb

@@ -0,0 +1,188 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# === Bioroebe::ShowFastaStatistics
+#
+# This class will show some statistics about all .fa files in the current
+# directory.
+#
+# Usage example:
+#
+#   Bioroebe::ShowFastaStatistics.new(ARGV)
+#
+# =========================================================================== #
+# require 'bioroebe/fasta_and_fastq/show_fasta_statistics.rb'
+# =========================================================================== #
+require 'bioroebe/base/commandline_application/commandline_application.rb'
+
+module Bioroebe
+
+class ShowFastaStatistics < ::Bioroebe::CommandlineApplication # === Bioroebe::ShowFastaStatistics
+
+  # ========================================================================= #
+  # === 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()
+    determine_all_local_fasta_files
+    # ======================================================================= #
+    # === @pad_filename_to
+    #
+    # This variable keeps track how much we should pad things.
+    # ======================================================================= #
+    @pad_filename_to = 35 
+  end
+
+  # ========================================================================= #
+  # === determine_all_local_fasta_files
+  # ========================================================================= #
+  def determine_all_local_fasta_files
+    # ======================================================================= #
+    # === @array_fasta_files
+    #
+    # We will populate the Array next. First we will try via .fa files,
+    # but if this is empty, we will try via .fasta as well.
+    # ======================================================================= #
+    @array_fasta_files = Dir['*.fa']
+    if @array_fasta_files.empty?
+      @array_fasta_files = Dir['*.fasta']
+    end
+  end
+
+  # ========================================================================= #
+  # === menu                                                       (menu tag)
+  # ========================================================================= #
+  def menu(
+      i = commandline_arguments?
+    )
+    if i.is_a? Array
+      i.each {|entry| menu(entry) }
+    else
+      case i # case tag
+      # ===================================================================== #
+      # === --help
+      # ===================================================================== #
+      when /help/
+        show_help
+        exit
+      # ===================================================================== #
+      # === show_fasta_statistics --start-with
+      #
+      # This entry point allows us to restrict to only certain .fa 
+      # filenames.
+      #
+      # Usage example on the commandline:
+      #
+      #   show_fasta_statistics --start-with=ncrna
+      #   show_fasta_statistics --fasta-stats=ncrna
+      #
+      # ===================================================================== #
+      when /^-?-?start(_|-)?with=(.+)$/i, # $2
+           /^-?-?fasta(-|_)?stats=(.+)$/i
+        starts_with = $2.to_s.dup
+        @array_fasta_files.select! {|line|
+          line.start_with? starts_with
+        }
+      else
+        if File.directory? i
+          _ = Dir[rds(i+'/**.fa')]
+          _ << Dir[rds(i+'/**.fasta')]
+          _.flatten!
+          add_these_entries_to_the_fasta_files(_)
+        end
+      end
+    end
+  end
+
+  # ========================================================================= #
+  # === add_these_entries_to_the_fasta_files
+  #
+  # This method can be used to append to our Array keeping track of the
+  # FASTA files.
+  # ========================================================================= #
+  def add_these_entries_to_the_fasta_files(i)
+    @array_fasta_files << i
+    @array_fasta_files.flatten!
+  end
+
+  # ========================================================================= #
+  # === show_help                                                  (help tag)
+  # ========================================================================= #
+  def show_help
+    e
+    e '  --start-with= # consider only these file names '\
+      'that start with that substring'
+    e
+  end
+
+  # ========================================================================= #
+  # === run                                                         (run tag)
+  # ========================================================================= #
+  def run
+    menu
+    regex_to_use = /(\d+)/
+    sorted = @array_fasta_files.sort_by {|line|
+      line =~ regex_to_use
+      match = $1.to_s.dup.to_i
+      if match == 0
+        match = 999 # This is just so that we sort entries such as "foo.MT.fa" last.
+      end
+      match
+    }
+    sorted.each_with_index {|this_file, index| index += 1
+      show_this_filename = this_file.sub(/\.fa$/,'')
+      padded_index = index.to_s.rjust(4)
+      # ===================================================================== #
+      # Next colourize the index a little bit.
+      # ===================================================================== #
+      result = lightgreen(padded_index)+' '+
+               sfile(
+                 show_this_filename.ljust(@pad_filename_to) # Pad the name of the file.
+               )
+      if File.exist? this_file
+        n_entries = ' '+File.readlines(this_file).select {|line|
+          line.start_with? '>'
+        }.size.to_s.rjust(5)
+        colourized_n_entries = seagreen("#{n_entries.to_s} different entries")
+        result << colourized_n_entries
+      end
+      e result
+    }
+  end
+
+  # ========================================================================= #
+  # === Bioroebe::ShowFastaStatistics[]
+  # ========================================================================= #
+  def self.[](i = '')
+    new(i)
+  end
+
+end
+
+# =========================================================================== #
+# === Bioroebe.show_fasta_statistics
+# =========================================================================== #
+def self.show_fasta_statistics(i = ARGV)
+  ::Bioroebe::ShowFastaStatistics.new(i)
+end
+
+end
+
+if __FILE__ == $PROGRAM_NAME
+  Bioroebe::ShowFastaStatistics.new(ARGV)
+end # showfastastatistics /home/Temp/bioroebe/fasta/

data/lib/bioroebe/fasta_and_fastq/simplify_fasta_header/simplify_fasta_header.rb

@@ -0,0 +1,111 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# === Bioroebe::SimplifyFastaHeader
+#
+# This class will simplify a FASTA file. It depends on class ParseFasta
+# for its job.
+#
+# It must support commandline options such as --overwrite, so that we
+# can enable the following commandline usage:
+#
+#   simplify_fasta foo.fasta --overwrite
+#
+# Usage example:
+#
+#   Bioroebe::SimplifyFastaHeader.new(ARGV)
+#
+# =========================================================================== #
+# require 'bioroebe/fasta_and_fastq/simplify_fasta_header/simplify_fasta_header.rb'
+# Bioroebe.overwrite_fasta_header(ARGV)
+# Bioroebe::SimplifyFastaHeader.new(ARGV)
+# =========================================================================== #
+require 'bioroebe/base/commandline_application/commandline_application.rb'
+
+module Bioroebe
+
+class SimplifyFastaHeader < ::Bioroebe::CommandlineApplication # === Bioroebe::SimplifyFastaHeader
+
+  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()
+  end
+
+  # ========================================================================= #
+  # === run                                                         (run tag)
+  # ========================================================================= #
+  def run
+    of_this_file = commandline_arguments?.shift
+    _ = Bioroebe::ParseFasta.new(of_this_file, :do_not_run_yet) { :be_quiet }
+    _.set_commandline_arguments(commandline_arguments?)
+    _.do_process_the_commandline_arguments_that_are_files(of_this_file)
+    _.run
+    _.simplify_header
+  end
+
+  # ========================================================================= #
+  # === Bioroebe::SimplifyFastaHeader[]
+  # ========================================================================= #
+  def self.[](i = '')
+    new(i)
+  end
+
+end
+
+# ========================================================================= #
+# === Bioroebe.simplify_fasta_header
+# ========================================================================= #
+def self.simplify_fasta_header(
+    of_this_file = ARGV,
+    &block
+  )
+  Bioroebe::SimplifyFastaHeader.new(of_this_file, &block)
+end; self.instance_eval { alias simplify_header simplify_fasta_header } # === Bioroebe.simplify_header
+
+# ========================================================================= #
+# === Bioroebe.overwrite_fasta_header
+#
+# Usage example:
+#
+#   overwritefastaheader YP_003540719.1_cytochrome_b_mitochondrion_Columba_livia.fasta
+#
+# ========================================================================= #
+def self.overwrite_fasta_header(
+    of_this_file = ARGV,
+    &block
+  )
+  of_this_file = [of_this_file].flatten.compact # Ensure that we have an Array here.
+  of_this_file << ['--overwrite']
+  of_this_file.flatten!
+  Bioroebe::SimplifyFastaHeader.new(of_this_file, &block)
+end
+
+end
+
+if __FILE__ == $PROGRAM_NAME
+  Bioroebe::SimplifyFastaHeader.new(ARGV)
+end # simplifyfastaheader /root/Bioroebe/table_ids.md --overwrite

data/lib/bioroebe/fasta_and_fastq/split_this_fasta_file_into_chromosomes/constants.rb

@@ -0,0 +1,26 @@
+#!/user/bioinf4/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# require 'bioroebe/fasta/split_this_fasta_file_into_chromosomes/constants.rb'
+# =========================================================================== #
+module Bioroebe
+
+class SplitThisFastaFileIntoChromosomes < ::Bioroebe::CommandlineApplication # === Bioroebe::SplitThisFastaFileIntoChromosomes
+
+  # ========================================================================= #
+  # === NAMESPACE
+  # ========================================================================= #
+  NAMESPACE = inspect
+
+  # ========================================================================= #
+  # === ALLOWED_CHARACTERS
+  #
+  # This constant keeps track as to which characters are valid
+  # nucleotide strings - aka A, T, C, G, U.
+  # ========================================================================= #
+  ALLOWED_CHARACTERS = %w(
+    A T C G U
+  ) # We exclude A, T, C, G and U from the Array given above.
+
+end; end

data/lib/bioroebe/fasta_and_fastq/split_this_fasta_file_into_chromosomes/menu.rb

@@ -0,0 +1,41 @@
+#!/user/bioinf4/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# require 'bioroebe/fasta/split_this_fasta_file_into_chromosomes/menu.rb'
+# =========================================================================== #
+module Bioroebe
+
+class SplitThisFastaFileIntoChromosomes < ::Bioroebe::CommandlineApplication # === Bioroebe::SplitThisFastaFileIntoChromosomes
+
+  # ========================================================================= #
+  # === menu                                                       (menu tag)
+  # ========================================================================= #
+  def menu(
+      i = commandline_arguments?
+    )
+    if i.is_a? Array
+      i.each {|entry| menu(entry) }
+    else
+      case i
+      # ===================================================================== #
+      # === prefix=ncrna
+      #
+      # This can be used like so:
+      #
+      #   splitthisfasta Mus_musculus.GRCm38.ncrna.fa --prefix=ncrna.
+      #
+      # ===================================================================== #
+      when /^-?-?prefix=(.+)$/i
+        set_use_this_prefix($1.to_s.dup)
+      # ===================================================================== #
+      # === help
+      # ===================================================================== #
+      when /help/
+        show_help
+        exit
+      end
+    end
+  end
+
+end; end

data/lib/bioroebe/fasta_and_fastq/split_this_fasta_file_into_chromosomes/misc.rb

@@ -0,0 +1,23 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# require 'bioroebe/fasta/split_this_fasta_file_into_chromosomes/misc.rb
+# =========================================================================== #
+module Bioroebe
+
+class SplitThisFastaFileIntoChromosomes < ::Bioroebe::CommandlineApplication # === Bioroebe::SplitThisFastaFileIntoChromosomes.new
+
+  # ========================================================================= #
+  # === set_commandline_arguments
+  # ========================================================================= #
+  def set_commandline_arguments(i = '')
+    i = [i].flatten.compact
+    if i.any? {|line| File.exist? line }
+      @this_file = i.select {|line| File.exist? line }.first
+      i.reject! {|line| File.exist? line}
+    end
+    @commandline_arguments = i
+  end
+
+end; end

data/lib/bioroebe/fasta_and_fastq/split_this_fasta_file_into_chromosomes/reset.rb

@@ -0,0 +1,68 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# require 'bioroebe/fasta/split_this_fasta_file_into_chromosomes/reset.rb'
+# =========================================================================== #
+module Bioroebe
+
+class SplitThisFastaFileIntoChromosomes < ::Bioroebe::CommandlineApplication # === Bioroebe::SplitThisFastaFileIntoChromosomes
+
+  # ========================================================================= #
+  # === reset                                                     (reset tag)
+  # ========================================================================= #
+  def reset
+    super()
+    # ======================================================================= #
+    # === @namespace
+    # ======================================================================= #
+    @namespace = NAMESPACE
+    # ======================================================================= #
+    # === @be_verbose
+    # ======================================================================= #
+    @be_verbose = false
+    # ======================================================================= #
+    # === @this_file
+    #
+    # The variable @this_file will store which file we are working with.
+    # ======================================================================= #
+    @this_file = nil
+    # ======================================================================= #
+    # === @n_unidentified_lines
+    #
+    # The next variable keeps track as to how many unidentified lines we
+    # have found.
+    # ======================================================================= #
+    @n_unidentified_lines = 0
+    # ======================================================================= #
+    # === @store_the_fasta_body_as_a_single_line
+    #
+    # If the following variable is set to true then the FASTA body will
+    # be stored as a single line - thus ignoring newlines. If this option
+    # is set to false then we may end up with FASTA files that will have
+    # entries such as "ATGCG\nGCGC", which is not as convenient to process
+    # further.
+    # ======================================================================= #
+    @store_the_fasta_body_as_a_single_line = true
+    # ======================================================================= #
+    # === @last_line
+    #
+    # This line will keep track of the last line.
+    # ======================================================================= #
+    @last_line = ''.dup
+    # ======================================================================= #
+    # === @use_this_genome
+    #
+    # The variable @use_this_genome keeps track of which genome we will
+    # use, based on the number.
+    # ======================================================================= #
+    @use_this_genome = 'GRCm38'
+    # ======================================================================= #
+    # We must determine the prefix for the autogenerated files. This also
+    # requires the user to be in the proper directory, such as ncrna/ or
+    # cdna/.
+    # ======================================================================= #
+    determine_the_prefix_for_the_autogenerated_files
+  end
+
+end; end