bioroebe 0.10.80

data/lib/bioroebe/patterns/rgg_scanner.rb

@@ -0,0 +1,160 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# === Bioroebe::RGG_Scanner
+#
+# This class can scan a DNA nucleotide sequence and discover RGG motifs
+# there.
+#
+# RGG motifs are:
+#
+#   Arginine-Glycine-Glycine
+#
+# Usage example:
+#
+#   Bioroebe::RGG_Scanner.new
+#
+# =========================================================================== #
+# require 'bioroebe/pattern/rgg_scanner.rb'
+# =========================================================================== #
+require 'bioroebe/base/commandline_application/commandline_application.rb'
+
+module Bioroebe
+
+class RGG_Scanner < ::Bioroebe::CommandlineApplication # === Bioroebe::RGG_Scanner
+
+  # ========================================================================= #
+  # === 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
+
+  # ========================================================================= #
+  # === sequence?
+  # ========================================================================= #
+  def sequence?
+    commandline_arguments?.join(' ').strip.delete(' ')
+  end
+
+  # ========================================================================= #
+  # === indices?
+  # ========================================================================= #
+  def indices?
+    @result
+  end
+
+  # ========================================================================= #
+  # === report
+  # ========================================================================= #
+  def report(array = @result)
+    sequence = sequence?
+    if array.empty?
+      e 'No result for '+orange('RGG')+' has been found.'
+    else
+      # ===================================================================== #
+      # First display the main sequence.
+      # ===================================================================== #
+      e
+      e sequence.gsub(/RGG/, slateblue('RGG')+rev)
+      e
+      report_how_many_rgg_motifs_were_found_in_that_sequence
+      e
+      erev 'The indices for the starting positions are at:'
+      e
+      e slateblue('  '+indices_as_string?(array))
+      e
+    end
+  end
+
+  # ========================================================================= #
+  # === report_how_many_rgg_motifs_were_found_in_that_sequence
+  # ========================================================================= #
+  def report_how_many_rgg_motifs_were_found_in_that_sequence
+    erev 'A total of '+slateblue(@result.size.to_s)+rev+' potential RGG '\
+         'motifs were found in this sequence.'
+  end
+
+  # ========================================================================= #
+  # === indices_as_string?
+  # ========================================================================= #
+  def indices_as_string?(
+      this_array = @result
+    )
+    _ = ''.dup
+    this_array.each {|a,b|
+      _ << b.to_s+', '
+    }
+    _.strip!
+    _.chop! if _.end_with? ','
+    return _
+  end
+
+  # ========================================================================= #
+  # === set_result
+  # ========================================================================= #
+  def set_result(i)
+    @result = i
+  end
+
+  # ========================================================================= #
+  # === scan_for_rgg_sequences
+  # ========================================================================= #
+  def scan_for_rgg_sequences
+    _ = sequence?
+    array = []
+    # ======================================================================= #
+    # Perform the scanning for the RGG sequence next:
+    # ======================================================================= #
+    _.scan(/RGG/) {|subsequence|
+      array << [
+        subsequence, Regexp.last_match.offset(0).first
+      ]
+    }
+    set_result(array)
+  end
+
+  # ========================================================================= #
+  # === run                                                         (run tag)
+  # ========================================================================= #
+  def run
+    scan_for_rgg_sequences
+    report
+  end
+
+  # ========================================================================= #
+  # === Bioroebe::RGG_Scanner[]
+  # ========================================================================= #
+  def self.[](i = '')
+    new(i)
+  end
+
+end
+
+# =========================================================================== #
+# === Bioroebe.scan_for_rgg_motifs
+# =========================================================================== #
+def self.scan_for_rgg_motifs(i)
+  Bioroebe::RGG_Scanner.new(i)
+end
+
+end
+
+if __FILE__ == $PROGRAM_NAME
+  Bioroebe::RGG_Scanner.new(ARGV)
+end # rggscanner RGGGFGG RGGFGDRGGRGGGRGG

data/lib/bioroebe/patterns/scan_for_repeat.rb

@@ -0,0 +1,157 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# === Bioroebe::ScanForRepeat
+#
+# The idea behind this class is to scan for motifes such as "(TAAG)n".
+#
+# Usage example:
+#
+#   Bioroebe::ScanForRepeat.new(ARGV)
+#
+# =========================================================================== #
+# require 'bioroebe/patterns/scan_for_repeat.rb'
+# =========================================================================== #
+require 'bioroebe/requires/commandline_application.rb'
+
+module Bioroebe
+
+class ScanForRepeat < ::Bioroebe::CommandlineApplication # === Bioroebe::ScanForRepeat
+
+  # ========================================================================= #
+  # === NAMESPACE
+  # ========================================================================= #
+  NAMESPACE = inspect
+
+  # ========================================================================= #
+  # === initialize
+  # ========================================================================= #
+  def initialize(
+      commandline_arguments = nil,
+      run_already           = true
+    )
+    reset
+    set_commandline_arguments(
+      commandline_arguments
+    )
+    # ======================================================================= #
+    # === Handle blocks given next
+    # ======================================================================= #
+    if block_given?
+      yielded = yield
+      if yielded.is_a? Hash
+        # =================================================================== #
+        # === :pattern
+        # =================================================================== #
+        if yielded.has_key? :pattern
+          set_use_this_pattern(yielded[:pattern])
+        end
+        # =================================================================== #
+        # === :verbose
+        # =================================================================== #
+        if yielded.has_key? :verbose
+          set_be_verbose(:be_quiet)
+        end
+      end
+    end
+    run if run_already
+  end
+
+  # ========================================================================= #
+  # === reset                                                     (reset tag)
+  # ========================================================================= #
+  def reset
+    super()
+    # ======================================================================= #
+    # === @n_repeats
+    # ======================================================================= #
+    @n_repeats = 0
+    # ======================================================================= #
+    # === @scan_for_this_repeat
+    #
+    # Try to find the following repeat.
+    #
+    # We may also need to allow for syntax such as '(TAAG)n' as valid
+    # input.
+    # ======================================================================= #
+    set_scan_for_this_repeat
+  end
+
+  # ========================================================================= #
+  # === set_scan_for_this_repeat
+  # ========================================================================= #
+  def set_scan_for_this_repeat(i = 'TAAG')
+    i = i.dup
+    i.chop! if i.end_with? 'n'
+    i.delete!('()') if i.include?('(')
+    @scan_for_this_repeat = i
+  end; alias set_use_this_pattern set_scan_for_this_repeat # === set_use_this_pattern
+
+  # ========================================================================= #
+  # === determine_the_main_sequence
+  # ========================================================================= #
+  def determine_the_main_sequence
+    @main_sequence = first?
+  end
+
+  # ========================================================================= #
+  # === do_scan_for_the_main_pattern_in_use
+  # ========================================================================= #
+  def do_scan_for_the_main_pattern_in_use
+    @n_repeats = @main_sequence.scan(/#{@scan_for_this_repeat}/).size
+    report if be_verbose?
+  end
+
+  # ========================================================================= #
+  # === @n_repeats
+  # ========================================================================= #
+  def n_repeats?
+    @n_repeats
+  end; alias n_patterns? n_repeats? # === n_patterns?
+
+  # ========================================================================= #
+  # === report                                                   (report tag)
+  # ========================================================================= #
+  def report
+    e @n_repeats
+  end
+
+  # ========================================================================= #
+  # === run                                                         (run tag)
+  # ========================================================================= #
+  def run
+    determine_the_main_sequence
+    do_scan_for_the_main_pattern_in_use
+  end
+
+  # ========================================================================= #
+  # === Bioroebe::ScanForRepeat[]
+  # ========================================================================= #
+  def self.[](i = '')
+    new(i)
+  end
+
+end
+
+# =========================================================================== #
+# === Bioroebe.return_n_repeats
+#
+# This willy not output anything but simply return the number of matches matches.
+# =========================================================================== #
+def self.return_n_repeats(
+    i                              = ARGV,
+    use_this_as_the_repeat_pattern = 'ATG'
+  )
+  return Bioroebe::ScanForRepeat.new(i) {{
+    pattern: use_this_as_the_repeat_pattern,
+    verbose: :be_quiet
+  }}.n_patterns?
+end
+
+end
+
+if __FILE__ == $PROGRAM_NAME
+  Bioroebe::ScanForRepeat.new(ARGV)
+end # scanforrepeat TTAATT
+    # scanforrepeat TTAATTATATGATATGAATGTAAGAATTAAG

data/lib/bioroebe/pdb/download_this_pdb.rb

@@ -0,0 +1,67 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# require 'bioroebe/pdb/download_this_pdb.rb'
+# Bioroebe.download_this_pdb(ARGV)
+# =========================================================================== #
+module Bioroebe
+
+  require 'bioroebe/toplevel_methods/e.rb'
+  require 'bioroebe/toplevel_methods/esystem.rb'
+  require 'bioroebe/colours/colours.rb'
+
+  require 'open-uri'
+
+  require 'bioroebe/utility_scripts/move_file_to_its_correct_location.rb'
+  # ========================================================================= #
+  # === Bioroebe.download_this_pdb
+  #
+  # This method can be used to download a remote .pdb file to the local
+  # file-system. If the default pdb/ directory exists as well locally
+  # then the downloaded .pdb file will be relocated into that file.
+  #
+  # An example for a remote URL to a .pdb file would be:
+  #
+  #   https://files.rcsb.org/view/2BTS.pdb
+  #   https://files.rcsb.org/view/355D.pdb
+  #
+  # ========================================================================= #
+  def self.download_this_pdb(
+      i = '355D'
+    )
+    # ======================================================================= #
+    # Treat all input as an Array past the next point.
+    # ======================================================================= #
+    [i].flatten.compact.each {|this_entry|
+      if this_entry.frozen?
+        this_entry = this_entry.dup
+      end
+      if this_entry.end_with? '.pdb' # This will lateron be appended again anyway.
+        this_entry.sub!(/\.pdb$/,'')
+      end
+      this_entry.upcase! # For convenience.
+      unless this_entry.end_with? '.pdb'
+        this_entry << '.pdb'
+      end
+      e this_entry
+      # ===================================================================== #
+      # Build up our remote URL next:
+      # ===================================================================== #
+      remote_url = "https://files.rcsb.org/view/#{this_entry}"
+      e steelblue(remote_url)
+      esystem "wget #{remote_url}"
+      _ = File.basename(remote_url)
+      if File.exist? _
+        ::Bioroebe.move_file_to_its_correct_location(_)
+      end
+    }
+  end
+
+end
+
+if __FILE__ == $PROGRAM_NAME
+  Bioroebe.download_this_pdb(ARGV)
+end # downloadthispdb 355D
+    # downloadthispdb 2BTS.pdb
+    # downloadthispdb 1K4R

data/lib/bioroebe/pdb/fetch_fasta_sequence_from_pdb.rb

@@ -0,0 +1,164 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# === Bioroebe::FetchFastaSequenceFromPdb
+#
+# This class can fetch a FASTA sequence from the official pdb website.
+#
+# Usage example:
+#
+#   Bioroebe::FetchFastaSequenceFromPdb.new(ARGV)
+#
+# =========================================================================== #
+# require 'bioroebe/pdb/fetch_fasta_sequence_from_pdb.rb'
+# =========================================================================== #
+require 'bioroebe/base/commandline_application/commandline_application.rb'
+
+module Bioroebe
+
+class FetchFastaSequenceFromPdb < ::Bioroebe::CommandlineApplication # === Bioroebe::FetchFastaSequenceFromPdb
+
+  require 'open-uri'
+
+  # ========================================================================= #
+  # === NAMESPACE
+  # ========================================================================= #
+  NAMESPACE = inspect
+
+  # ========================================================================= #
+  # === URL_FOR_FASTA_ENTRIES_AT_THE_PDB
+  # ========================================================================= #
+  URL_FOR_FASTA_ENTRIES_AT_THE_PDB =
+    'https://www.rcsb.org/fasta/entry/'
+
+  # ========================================================================= #
+  # === initialize
+  # ========================================================================= #
+  def initialize(
+      i           = ARGV,
+      run_already = true
+    )
+    reset
+    set_commandline_arguments(i)
+    if block_given?
+      yielded = yield
+      case yielded
+      # ===================================================================== #
+      # === :be_silent
+      # ===================================================================== #
+      when :be_silent
+        set_be_silent
+      end
+    end
+    run if run_already
+  end
+
+  # ========================================================================= #
+  # === reset                                                     (reset tag)
+  # ========================================================================= #
+  def reset
+    super()
+    # ======================================================================= #
+    # === @namespace
+    # ======================================================================= #
+    @namespace = NAMESPACE
+    # ======================================================================= #
+    # === @result
+    # ======================================================================= #
+    @result = ''.dup
+  end
+
+  # ========================================================================= #
+  # === set_search_for_this_file
+  # ========================================================================= #
+  def set_search_for_this_file(i)
+    if i.include? '.'
+      i = i.split('.').first
+    end
+    @search_for_this_file = i.upcase
+  end
+
+  # ========================================================================= #
+  # === set_result
+  # ========================================================================= #
+  def set_result(i)
+    if i.is_a? Array
+      i = i.first
+    end
+    @result = i
+  end
+
+  # ========================================================================= #
+  # === result?
+  # ========================================================================= #
+  def result?
+    @result
+  end; alias aminoacid_sequence? result? # === aminoacid_sequence?
+
+  # ========================================================================= #
+  # === store_this_dataset
+  # ========================================================================= #
+  def store_this_dataset(
+      what = @result
+    )
+    into = log_dir?+'pdb/'+@search_for_this_file+'.fasta'
+    erev 'Next storing into `'+sfile(into)+'`.' if be_verbose?
+    write_what_into(what, into)
+  end
+
+  # ========================================================================= #
+  # === run                                                         (run tag)
+  # ========================================================================= #
+  def run
+    @commandline_arguments.each {|this_file|
+      set_search_for_this_file(this_file)
+      if File.exist? this_file
+        data_from_the_remote_webpage = File.read(this_file)
+      else # else download it
+        remote_url = "#{URL_FOR_FASTA_ENTRIES_AT_THE_PDB}#{@search_for_this_file}/display"
+        erev "Working on #{steelblue(remote_url)} #{rev}next." if be_verbose?
+        data_from_the_remote_webpage = URI.open(remote_url).read
+      end
+      set_result(data_from_the_remote_webpage)
+      store_this_dataset(@result)
+      @result # Return it here too, just in case.
+    }
+  end
+
+end
+
+# =========================================================================== #
+# === Bioroebe.fetch_fasta_sequence_from_pdb
+# =========================================================================== #
+def self.fetch_fasta_sequence_from_pdb(i = ARGV)
+  Bioroebe::FetchFastaSequenceFromPdb.new(i)
+end
+
+# =========================================================================== #
+# === Bioroebe.return_fasta_sequence_from_this_pdb_file
+#
+# This variant will (silently) return the aminoacid sequence.
+#
+# Invocation example:
+#
+#   x = Bioroebe.return_fasta_sequence_from_this_pdb_file "2bts" # => "MLSDEDFKAVFGMTRSAFANLPLWKQQNLKKEKGLF"
+#
+# =========================================================================== #
+def self.return_fasta_sequence_from_this_pdb_file(i = ARGV)
+  require 'bioroebe/pdb/parse_pdb_file.rb'
+  sequence = Bioroebe::FetchFastaSequenceFromPdb.new(i) { :be_silent }.aminoacid_sequence?
+  if sequence.start_with? '>'
+    sequence = sequence.split("\n")[1..-1]
+  end
+  if sequence.is_a? Array
+    sequence = sequence.first
+  end
+  return sequence
+end
+
+end
+
+if __FILE__ == $PROGRAM_NAME
+   Bioroebe::FetchFastaSequenceFromPdb.new(ARGV)
+end # fetchfastasequencefrompdb 2bts

data/lib/bioroebe/pdb/parse_mmCIF_file.rb

@@ -0,0 +1,63 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# === Bioroebe::ParsemmCIFFile
+#
+# This class shall eventually be able to parse .cif files. Currently it
+# does not work - I need to test it on a real dataset.
+# =========================================================================== #
+require 'bioroebe/base/commandline_application/commandline_application.rb'
+
+module Bioroebe
+
+class ParsemmCIFFile < ::Bioroebe::CommandlineApplication # === Bioroebe::ParsemmCIFFile
+
+  # ========================================================================= #
+  # === NAMESPACE
+  # ========================================================================= #
+  NAMESPACE = inspect
+
+  # ========================================================================= #
+  # === initialize
+  # ========================================================================= #
+  def initialize(
+      i           = ARGV,
+      run_already = true
+    )
+    reset
+    if block_given?
+      yielded = yield
+      case yielded
+      # ===================================================================== #
+      # === :be_silent
+      # ===================================================================== #
+      when :be_silent
+        set_be_silent
+      end
+    end
+    run if run_already
+  end
+
+  # ========================================================================= #
+  # === reset                                                     (reset tag)
+  # ========================================================================= #
+  def reset
+    super()
+    # ======================================================================= #
+    # === @namespace
+    # ======================================================================= #
+    @namespace = NAMESPACE
+  end
+
+  # ========================================================================= #
+  # === run
+  # ========================================================================= #
+  def run
+  end
+
+end; end
+
+if __FILE__ == $PROGRAM_NAME
+  Bioroebe::ParsemmCIFFile.new(ARGV)
+end # parsemmcifffile *mmcif