bioroebe 0.12.24 → 0.13.31

data/lib/bioroebe/utility_scripts/fetch_data_from_uniprot/fetch_data_from_uniprot.rb

@@ -0,0 +1,269 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# === Bioroebe::FetchDataFromUniprot
+#
+# This class can be used to obtain data from uniprot.
+#
+# Usage examples:
+#
+#   Bioroebe::FetchDataFromUniprot.new(ARGV)
+#
+# =========================================================================== #
+# require 'bioroebe/utility_scripts/fetch_data_from_uniprot/fetch_data_from_uniprot.rb'
+# Bioroebe::FetchDataFromUniprot.new(ARGV)
+# =========================================================================== #
+require 'bioroebe/requires/commandline_application.rb'
+
+module Bioroebe
+
+class FetchDataFromUniprot < ::Bioroebe::CommandlineApplication # === Bioroebe::FetchDataFromUniprot
+
+  require 'open-uri'
+
+  # ========================================================================= #
+  # === DEFAULT_FASTA_FILE
+  # ========================================================================= #
+  DEFAULT_FASTA_FILE = 'B5ZC00'
+
+  # ========================================================================= #
+  # === initialize
+  #
+  # The first argument to this class should be the name of the protein
+  # at hand. Alternatively it can also be just the full URL - both
+  # variants will work fine.
+  #
+  # The second argument to this class specifies whether this class will
+  # automatically try to rename the FASTA file, after it was downloaded.
+  # By default this is enabled, largely because it makes it a bit easier
+  # to know what a particular fasta sequence contains.
+  # ========================================================================= #
+  def initialize(
+      i           = nil,
+      run_already = true
+    )
+    reset
+    set_commandline_arguments(i)
+    run if run_already
+  end
+
+  # ========================================================================= #
+  # === reset                                                     (reset tag)
+  # ========================================================================= #
+  def reset
+    super()
+    # ======================================================================= #
+    # === :do_perform_rename_action
+    #
+    # By default we will rename the downloaded file, hence the true value
+    # for this variable here.
+    # ======================================================================= #
+    @internal_hash[:do_perform_rename_action] = true
+    # ======================================================================= #
+    # === :result
+    # ======================================================================= #
+    @internal_hash[:result] = ''.dup
+    reset_the_internal_variables
+  end
+
+  # ========================================================================= #
+  # === reset_the_internal_variables
+  # ========================================================================= #
+  def reset_the_internal_variables
+    # ======================================================================= #
+    # === :result
+    # ======================================================================= #
+    @internal_hash[:result].clear
+    # ======================================================================= #
+    # === :use_this_remote_URL
+    # ======================================================================= #
+    @internal_hash[:use_this_remote_URL] = DEFAULT_FASTA_FILE # Default value.
+    # ======================================================================= #
+    # === :stored_local_file
+    # ======================================================================= #
+    @internal_hash[:stored_local_file] = nil
+  end
+
+  # ========================================================================= #
+  # === do_perform_rename_action?
+  # ========================================================================= #
+  def do_perform_rename_action?
+    @internal_hash[:do_perform_rename_action]
+  end
+
+  # ========================================================================= #
+  # === use_this_remote_URL?
+  # ========================================================================= #
+  def use_this_remote_URL?
+    @internal_hash[:use_this_remote_URL]
+  end
+
+  # ========================================================================= #
+  # === set_use_this_remote_URL
+  # ========================================================================= #
+  def set_use_this_remote_URL(i)
+    @internal_hash[:use_this_remote_URL] = i
+  end
+
+  # ========================================================================= #
+  # === set_stored_local_file
+  # ========================================================================= #
+  def set_stored_local_file(i)
+    @internal_hash[:stored_local_file] = i
+  end
+
+  # ========================================================================= #
+  # === stored_local_file?
+  # ========================================================================= #
+  def stored_local_file?
+    @internal_hash[:stored_local_file]
+  end; alias store_here? stored_local_file? # === store_here? 
+
+  # ========================================================================= #
+  # === message
+  # ========================================================================= #
+  def message(
+      i,
+      use_opne = false
+    )
+    case use_opne
+    # ======================================================================= #
+    # === :use_opnn
+    # ======================================================================= #
+    when :use_opnn
+      use_opne = true
+    end
+    if use_opne
+      opnn
+    end
+    @internal_hash[:result] << "#{i}\n"
+    erev i if be_verbose?
+  end; alias msg message # === msg
+
+  # ========================================================================= #
+  # === newline
+  # ========================================================================= #
+  def newline
+    message("\n")
+  end
+
+  # ========================================================================= #
+  # === do_download_this_file_from_uniprot
+  # ========================================================================= #
+  def do_download_this_file_from_uniprot(_)
+    _ = _.dup if _.frozen?
+    _.strip!
+    # ===================================================================== #
+    # === Support omission of leading 'http' here:
+    # ===================================================================== #
+    if _.start_with? 'http'
+      remote_URL = _
+    else
+      remote_URL = "https://www.uniprot.org/uniprot/#{_}"
+    end
+    remote_URL = remote_URL.dup if remote_URL.frozen?
+    # ======================================================================= #
+    # === Ensure trailing .fasta suffix
+    #
+    # Next, we will append a trailing ".fasta" part for uniprot, if
+    # the very last part does NOT include a '.' character.
+    # ======================================================================= #
+    _ = File.basename(remote_URL)
+    unless _.include?('.')
+      remote_URL << '.fasta' unless remote_URL.end_with? '.fasta'
+    end
+    # ======================================================================= #
+    # The remote_URL is now as it should be, so store it:
+    # ======================================================================= #
+    set_use_this_remote_URL(remote_URL)
+    # ======================================================================= #
+    # Store this protein sequence at an appropriate location:
+    # ======================================================================= #
+    newline
+    message "Trying to obtain remote data from "\
+            "#{sfancy(remote_URL)} #{rev}next:"
+    newline
+    local_uniprot_directory = LOCAL_DIRECTORY_FOR_UNIPROT
+    store_here = local_uniprot_directory+
+                 File.basename(remote_URL)
+    # ======================================================================= #
+    # Next create the directory unless it already exists.
+    # ======================================================================= #
+    unless File.directory? local_uniprot_directory
+      mkdir local_uniprot_directory
+    end
+    dataset = URI.open(remote_URL).read
+    message "The remote dataset will be stored here: "\
+            "#{sfile(store_here)}"
+    newline
+    write_what_into(dataset, store_here)
+    # ======================================================================= #
+    # Next comes a rename action - this will make the .fasta file more
+    # verbose, but I think this may still be helpful, in particular if you
+    # have a lot of .fasta files. There is a switch that controls this
+    # setting though.
+    # ======================================================================= #
+    if do_perform_rename_action?
+      # ===================================================================== #
+      # The initially assumed filename may be like this:
+      #
+      #   /root/Bioroebe/A2Z669_CSPLT_ORYSI_CASP-like_protein_5A2_OS=Oryza_sativa_subsp_indica_OX=39946_GN=OsI_33147_PE=3_SV=1.fasta
+      #
+      # ===================================================================== #
+      assumed_filename = ::Bioroebe.return_new_filename_based_on_fasta_identifier(store_here)
+      new_filename = local_uniprot_directory+
+                     File.basename(assumed_filename)
+      message "Next renaming `#{sfile(store_here)}#{rev}` to"
+      message "`#{sfile(new_filename)}#{rev}`."
+      mv(store_here, new_filename)
+      store_here = new_filename
+    end
+    set_stored_local_file(File.absolute_path(store_here))
+  end
+
+  # ========================================================================= #
+  # === result?
+  # ========================================================================= #
+  def result?
+    @internal_hash[:result]
+  end; alias string? result? # === string?
+
+  # ========================================================================= #
+  # === run                                                         (run tag)
+  # ========================================================================= #
+  def run
+    reset_the_internal_variables
+    _ = first_argument?
+    if _.nil? or _.empty?
+      _ = DEFAULT_FASTA_FILE # Use a default value in this case.
+      message 'Making use of the default value '+steelblue(_)+
+              rev+' as no specific argument was '\
+              'passed to this class.', :use_opnn
+    end
+    do_download_this_file_from_uniprot(_)
+  end
+
+  # ========================================================================= #
+  # === Bioroebe::FetchDataFromUniprot[]
+  # ========================================================================= #
+  def self.[](i = ARGV)
+    new(i)
+  end
+
+end
+
+# =========================================================================== #
+# === Bioroebe.fetch_data_from_uniprot
+# =========================================================================== #
+def self.fetch_data_from_uniprot(i = ARGV)
+  Bioroebe::FetchDataFromUniprot.new(i)
+end
+
+end
+
+if __FILE__ == $PROGRAM_NAME
+  Bioroebe::FetchDataFromUniprot.new(ARGV)
+end # unifetch
+    # unifetch B5ZC00
+    # unifetch A2Z669

data/lib/bioroebe/utility_scripts/find_gene.rb

@@ -17,12 +17,14 @@
 # =========================================================================== #
 # require 'bioroebe/utility_scripts/find_gene.rb' 
 # =========================================================================== #
-require 'bioroebe/base/commandline_application/commandline_application.rb'
+require 'bioroebe/requires/commandline_application.rb'
 
 module Bioroebe
 
 class FindGene < ::Bioroebe::CommandlineApplication # === Bioroebe::FindGene
 
+  require 'bioroebe/codons/codons.rb'
+
   # ========================================================================= #
   # === DEFAULT_STRING
   # ========================================================================= #
@@ -191,7 +193,7 @@ class FindGene < ::Bioroebe::CommandlineApplication # === Bioroebe::FindGene
   # ========================================================================= #
   # === Bioroebe::FindGene[]
   # ========================================================================= #
-  def self.[](i)
+  def self.[](i = ARGV)
     new(i)
   end
 

data/lib/bioroebe/utility_scripts/{mirror_repeat.rb → mirror_repeat/mirror_repeat.rb}

@@ -16,10 +16,10 @@
 #   Bioroebe::MirrorRepeat.new(ARGV)
 #
 # =========================================================================== #
-# require 'bioroebe/utility_scripts/mirror_repeat.rb'
+# require 'bioroebe/utility_scripts/mirror_repeat/mirror_repeat.rb'
 # Bioroebe::MirrorRepeat.new(ARGV)
 # =========================================================================== #
-require 'bioroebe/base/commandline_application/commandline_application.rb'
+require 'bioroebe/requires/commandline_application.rb'
 
 module Bioroebe
 
@@ -43,12 +43,12 @@ class MirrorRepeat < CommandlineApplication # === Bioroebe::MirrorRepeat
         # === :use_separator_token
         # =================================================================== #
         if yielded.has_key? :use_separator_token
-          @commandline_arguments << '--use-spacer'
+          commandline_arguments? << '--use-spacer'
         # =================================================================== #
         # === :separator_token
         # =================================================================== #
         elsif yielded.has_key? :separator_token
-          @commandline_arguments << '--use-spacer'
+          commandline_arguments? << '--use-spacer'
         end
         # =================================================================== #
         # === :use_colours
@@ -140,7 +140,7 @@ class MirrorRepeat < CommandlineApplication # === Bioroebe::MirrorRepeat
   # === menu                                                       (menu tag)
   # ========================================================================= #
   def menu(
-      i = @commandline_arguments
+      i = commandline_arguments?
     )
     if i.is_a? Array
       i.each {|entry| menu(entry) }

data/lib/bioroebe/utility_scripts/move_file_to_its_correct_location.rb

@@ -17,13 +17,13 @@
 # Bioroebe.move_file_to_its_correct_location(ARGV)
 # Bioroebe::MoveFileToItsCorrectLocation.new(ARGV)
 # =========================================================================== #
-require 'bioroebe/base/commandline_application/commandline_application.rb'
+require 'bioroebe/requires/commandline_application.rb'
 
 module Bioroebe
 
 class MoveFileToItsCorrectLocation < ::Bioroebe::CommandlineApplication
 
-  require 'bioroebe/toplevel_methods/file_and_directory_related_actions.rb'
+  require 'bioroebe/toplevel_methods/toplevel_methods.rb'
 
   # ========================================================================= #
   # === initialize
@@ -128,7 +128,7 @@ class MoveFileToItsCorrectLocation < ::Bioroebe::CommandlineApplication
   # ========================================================================= #
   # === Bioroebe::MoveFileToItsCorrectLocation[]
   # ========================================================================= #
-  def self.[](i)
+  def self.[](i = ARGV)
     new(i)
   end
 

data/lib/bioroebe/utility_scripts/{parse_taxonomy.rb → parse_taxonomy/parse_taxonomy.rb}

@@ -4,7 +4,8 @@
 # =========================================================================== #
 # === Bioroebe::ParseTaxonomy
 #
-# This class may display, on the commandline, something like this:
+# This class may display, on the commandline, something like the
+# following:
 #
 #   ======================================================================
 #   Title: Taxonomy browser (Hepatitis delta virus (isolate US-2))
@@ -20,9 +21,9 @@
 #   Bioroebe::ParseTaxonomy.new
 #
 # =========================================================================== #
-# require 'bioroebe/utility_scripts/parse_taxonomy.rb'
+# require 'bioroebe/utility_scripts/parse_taxonomy/parse_taxonomy.rb'
 # =========================================================================== #
-require 'bioroebe/base/commandline_application/commandline_application.rb'
+require 'bioroebe/requires/commandline_application.rb'
 
 module Bioroebe
 
@@ -40,7 +41,7 @@ class ParseTaxonomy < ::Bioroebe::CommandlineApplication # === Bioroebe::ParseTa
     'https://www.ncbi.nlm.nih.gov/Taxonomy/Browser/'\
     'wwwtax.cgi?mode=Info&id=261991&lvl=3&p=protein&lin=f&keep=1&srchmode=1&unlock'
 
-  attr_accessor :title
+  attr_writer :title
 
   # ========================================================================= #
   # === initialize
@@ -123,9 +124,16 @@ class ParseTaxonomy < ::Bioroebe::CommandlineApplication # === Bioroebe::ParseTa
   # ========================================================================= #
   def show_title
     @title = @dataset.scan(/\<title\>(.+)\<\/title\>/)[0][0]
-    e steelblue('Title: ')+royalblue(@title)
+    e steelblue('Title: ')+royalblue(title?)
   end
 
+  # ========================================================================= #
+  # === title?
+  # ========================================================================= #
+  def title?
+    @title
+  end; alias title title? # === title
+
   # ========================================================================= #
   # === run                                                         (run tag)
   # ========================================================================= #
@@ -165,4 +173,5 @@ end; end
 
 if __FILE__ == $PROGRAM_NAME
   Bioroebe::ParseTaxonomy.new(ARGV)
-end # parsetaxonomy
+end # parsetaxonomy 261991
+    # parsetaxonomy "https://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?lvl=0&id=261991"

data/lib/bioroebe/utility_scripts/{pathways.rb → pathways/pathways.rb}

@@ -11,9 +11,9 @@
 #   Bioroebe.pathways
 #
 # =========================================================================== #
-# require 'bioroebe/utility_scripts/pathways.rb'
+# require 'bioroebe/utility_scripts/pathways/pathways.rb'
 # =========================================================================== #
-require 'bioroebe/base/commandline_application/commandline_application.rb'
+require 'bioroebe/requires/commandline_application.rb'
 
 module Bioroebe
 
@@ -79,7 +79,8 @@ class Pathways < ::Bioroebe::CommandlineApplication # === Bioroebe::Pathways
   def show_every_pathway
     if ALL_PATHWAYS.empty?
       erev 'No pathways were found. Does the directory exist at '
-      erev sdir(PATHWAYS_DIRECTORY)+' and is it not empty?'
+      erev sdir(PATHWAYS_DIRECTORY)+rev+
+           ' and is it not empty?'
     else
       e "#{Pathways.rev}Now showing the available pathways."
       e

data/lib/bioroebe/utility_scripts/{permutations.rb → permutations/permutations.rb}

@@ -19,9 +19,9 @@
 #   Bioroebe.permutations 5
 #
 # =========================================================================== #
-# require 'bioroebe/utility_scripts/permutations.rb'
+# require 'bioroebe/utility_scripts/permutations/permutations.rb'
 # =========================================================================== #
-require 'bioroebe/base/commandline_application/commandline_application.rb'
+require 'bioroebe/requires/commandline_application.rb'
 
 module Bioroebe
 
@@ -109,7 +109,7 @@ class Permutations < ::Bioroebe::CommandlineApplication # === Bioroebe::Permutat
   # ========================================================================= #
   # === Bioroebe::Permutations[]
   # ========================================================================= #
-  def self.[](i = '')
+  def self.[](i = ARGV)
     new(i)
   end
 

data/lib/bioroebe/utility_scripts/punnet/punnet.rb

@@ -13,11 +13,11 @@
 # =========================================================================== #
 # require 'bioroebe/utility_scripts/punnet/punent.rb'
 # =========================================================================== #
-require 'bioroebe/base/base.rb'
+require 'bioroebe/requires/commandline_application.rb'
 
 module Bioroebe
 
-class Punnet < Base # === Bioroebe::Punnet
+class Punnet < ::Bioroebe::CommandlineApplication # === Bioroebe::Pathways
 
   # ========================================================================= #
   # === DEFAULT_INPUT
@@ -64,8 +64,10 @@ class Punnet < Base # === Bioroebe::Punnet
   # ========================================================================= #
   def show_usage
     e 'Simply input commandline options here:'
+    e
     e '  0.7 x 0.3'
     e '  punnett 0.7x0.3'
+    e
   end
 
   # ========================================================================= #

data/lib/bioroebe/utility_scripts/{show_this_dna_sequence.rb → show_this_dna_sequence/show_this_dna_sequence.rb}

@@ -17,7 +17,7 @@
 # require 'bioroebe/utility_scripts/show_this_dna_sequence.rb'
 # Bioroebe::ShowThisDNASequence.new(ARGV) 
 # =========================================================================== #
-require 'bioroebe/base/commandline_application/commandline_application.rb'
+require 'bioroebe/requires/commandline_application.rb'
 
 module Bioroebe