bioroebe 0.10.80

data/lib/bioroebe/genbank/genbank_parser.rb

@@ -0,0 +1,291 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# === Bioroebe::GenbankParser
+#
+# This is similar to the FastaParser, but instead it will only select the
+# content between "ORIGIN" and "//" entries.
+#
+# You can pass the content of a genbank-file to this class, and it can
+# report the nucleotide sequence, e. g. the part starting after the
+# ORIGIN string.
+#
+# The reason why this class has been created was because it is sometimes
+# necessary to parse a genebank file.
+# =========================================================================== #
+# require 'bioroebe/genbank/genbank_parser.rb'
+# Bioroebe::GenbankParser.new(ARGV)
+# =========================================================================== #
+require 'bioroebe/base/commandline_application/commandline_application.rb'
+
+module Bioroebe
+
+class GenbankParser < ::Bioroebe::CommandlineApplication # === Bioroebe::GenbankParser
+ 
+  # ========================================================================= #
+  # === UPCASE_SEQUENCE
+  # ========================================================================= #
+  UPCASE_SEQUENCE = true
+
+  # ========================================================================= #
+  # === TEST_STRING
+  #
+  # Our example test-string, to see how such a genbank file usually looks
+  # like.
+  # ========================================================================= #
+  TEST_STRING = '                   /note="internal transcribed spacer 2"
+ORIGIN      
+        1 cgtaacaagg tttccgtagg tgaaccttcg gaaggatcat tgttgagacc cccaaaaaaa
+       61 cgatcgagtt aatccggagg accggtgtag tttggtctcc caggggcttt ggctactgtg
+      121 gtggccgtga atttccgtcg aacctccttg ggagaattct tgatggcaat tgaacccttg
+      181 gcccggcgca gtttcgcccc aagtcaaatg agatggaacc ggcggagggc atcgtcctcc
+      241 atggaaccgg ggagggccgg cgttcttccg ttccccccat gaattttttt ttgacaactc
+      301 tcggcaacgg atatctcggc tctttgcatc cgatgaaaga acccagcgaa atgtgataag
+      361 tggtgtgaat tgcagaatcc cgtgaaccat cgagtctttg aacgcaagtt gcgcccgagg
+      421 ccatcaggct aagggcacgc ctgcctgggc gttgcgtgct gcatctctct cccattgcta
+      481 aggctgaaca ggcatactgt tcggccggcg cggatgagtg tttggcccct tgttcttcgg
+      541 tgcgatgggt ccaagacctg ggcttttgac ggccggaaat ccggcaagag gtggacggac
+      601 ggtggctgcg acgaagctgt cgtgcgaatg ccctacgctg tcgtatttga tgggccggaa
+      661 taaatccctt ttgagcccca ttggaggcac gtcaacccgt gggcggtcga cggccatttg
+      721 gatgcaaccc caggtcaggt gagga
+//
+LOCUS       Z78510                   750 bp    DNA     linear   PLN 30-NOV-2006
+DEFINITION  P.caricinum 5.8S rRNA gene and ITS1 and ITS2 DNA.
+ACCESSION   Z78510
+VERSION     Z78510.1  GI:2765635
+KEYWORDS    5.8S ribosomal RNA; 5.8S rRNA gene; internal transcribed spacer;
+            ITS1; ITS2.
+SOURCE      Phragmipedium caricinum
+  ORGANISM  Phragmipedium caricinum
+            Eukaryota; Viridiplantae; Streptophyta; Embryophyta; Tracheophyta;
+            Spermatophyta; Magnoliophyta; Liliopsida; Asparagales; Orchidaceae;
+            Cypripedioideae; Phragmipedium.
+REFERENCE   1
+  AUTHORS   Cox,A.V., Pridgeon,A.M., Albert,V.A. and Chase,M.W.
+  TITLE     Phylogenetics of the slipper orchids (Cypripedioideae:
+            Orchidaceae): nuclear rDNA ITS sequences
+  JOURNAL   Unpublished
+REFERENCE   2  (bases 1 to 750)
+  AUTHORS   Cox,A.V.
+  TITLE     Direct Submission
+  JOURNAL   Submitted (19-AUG-1996) Cox A.V., Royal Botanic Gardens, Kew,
+            Richmond, Surrey TW9 3AB, UK
+FEATURES             Location/Qualifiers
+     source          1..750
+                     /organism="Phragmipedium caricinum"
+                     /mol_type="genomic DNA"
+                     /db_xref="taxon:53127"
+     misc_feature    1..380
+                     /note="internal transcribed spacer 1"
+     gene            381..550
+                     /gene="5.8S rRNA"
+     rRNA            381..550
+                     /gene="5.8S rRNA"
+                     /product="5.8S ribosomal RNA"
+     misc_feature    551..750
+                     /note="internal transcribed spacer 2"
+ORIGIN      
+        1 ctaaccaggg ttccgaggtg accttcggga ggattccttt ttaagccccc gaaaaaacga
+       61 tcgaattaaa ccggaggacc ggtttaattt ggtctcccca ggggctttcc ccccttggtg
+      121 gccgtgaatt tccatcgaac ccccctggga gaattcttgg tggccaatgg acccttggcc
+      181 cggcgcaatt tcccccccaa tcaaatgaga taggaccggc agggggcgtc cccccccatg
+      241 gaaccgggga gggccggcat tcttccgttc ccccctcgga ttttttgaca actctcgcaa
+      301 cggatatctc gcctctttgc atcggatgga agaacgcagc gaaatgtgat aagtggtgtg
+      361 aattgcagaa tcccgtgaac catcgagtct ttgaacgcaa gttgcgcccg aggccatcag
+      421 gctaagggca cgcctgcctg ggcgttgcgt gctgcatctc tcccattgct aaggttgaac
+      481 gggcatactg ttcggccggc gcggatgaga gattggcccc ttgttcttcg gtgcgatggg
+      541 tccaagacct gggcttttga cggtccaaaa tccggcaaga ggtggacgga cggtggctgc
+      601 gacaaagctg tcgtgcgaat gccctgcgtt gtcgtttttg atgggccgga ataaatccct
+      661 tttgaacccc attggaggca cgtcaaccca tgggcggttg acggccattt ggatgcaacc
+      721 ccaggtcagg tgagccaccc gctgagttta
+//
+LOCUS       Z78509                   731 bp    DNA     linear   PLN 30-NOV-2006
+DEFINITION  P.pearcei 5.8S rRNA gene and ITS1 and ITS2 DNA.
+ACCESSION   Z78509
+VERSION     Z78509.1  GI:2765634
+KEYWORDS    5.8S ribosomal RNA; 5.8S rRNA gene; internal transcribed spacer;
+            ITS1; ITS2.
+SOURCE      Phragmipedium pearcei
+  ORGANISM  Phragmipedium pearcei
+            Eukaryota; Viridiplantae; Streptophyta; Embryophyta; Tracheophyta;
+            Spermatophyta; Magnoliophyta; Liliopsida; Asparagales; Orchidaceae;
+            Cypripedioideae; Phragmipedium.
+REFERENCE   1
+  AUTHORS   Cox,A.V., Pridgeon,A.M., Albert,V.A. and Chase,M.W.
+  TITLE     Phylogenetics of the slipper orchids (Cypripedioideae:
+            Orchidaceae): nuclear rDNA ITS sequences
+  JOURNAL   Unpublished
+REFERENCE   2  (bases 1 to 731)
+  AUTHORS   Cox,A.V.
+  TITLE     Direct Submission
+  JOURNAL   Submitted (19-AUG-1996) Cox A.V., Royal Botanic Gardens, Kew,
+            Richmond, Surrey TW9 3AB, UK
+FEATURES             Location/Qualifiers
+     source          1..731
+                     /organism="Phragmipedium pearcei"
+                     /mol_type="genomic DNA"
+                     /db_xref="taxon:53135"
+     misc_feature    1..380
+                     /note="internal transcribed spacer 1"
+     gene            381..550
+                     /gene="5.8S rRNA"
+     rRNA            381..550
+                     /gene="5.8S rRNA"
+                     /product="5.8S ribosomal RNA"
+     misc_feature    551..731
+                     /note="internal transcribed spacer 2"
+ORIGIN      
+        1 cgtaacaagg tttccgtagg tgaacctgcg gaaggatcat tgttgagacc gccaaatata
+       61 cgatcgagtt aatccggagg accggtgtag tttggtctcc caggggcttt cgccgctgtg
+      121 gtgaccgtga tttgccatcg agcctccttg ggagatttct tgatggcaat tgaacccttg
+      181 gcccggcgca gtttcgcgcc aagtcatatg agatagaacc ggcggagggc gtcgtcctcc
+      241 atggagcggg gagggccggc atgctccgtg cccccccatg aatttttctg acaactctcg
+      301 gcaacggacg taacaaggtt taaatgtgat aagcaggtgt gaattgcaga atcccgtgaa
+      361 ccatcgagtc tttgaacgca agttgcgccc gaggccatca ggttaagggc acgcctgcct
+      421 gggcgttgcg tgctgcatct ctcccattgc taaggttgaa cgggcatact gttcggccgg
+      481 cgcggatgag agtttggccc cttgttcttc ggtgcgatgg gtccaagacc tgggcttttg
+      541 acggtccaaa atccggcaag aggtggacgg acggtggctg cgacagagct gtcgtgcgaa
+      601 tgccctacgt tgtcgttttt gatgggccag aataaatccc ttttgaaccc cattggaggc
+      661 acgtcaaccc aatggggggt gacgggcatt tggttaaccc cggcaagtta aggcacccgt
+      721 taattttagg a
+//
+LOCUS       Z78508                   741 bp    DNA     linear   PLN 30-NOV-2006'
+
+  # ========================================================================= #
+  # === initialize
+  # ========================================================================= #
+  def initialize(
+      dataset     = ARGV,
+      run_already = true
+    )
+    reset
+    determine_dataset(
+      dataset
+    )
+    if block_given?
+      yielded = yield
+      case yielded
+      # ===================================================================== #
+      # === :do_not_report_anything
+      # ===================================================================== #
+      when :do_not_report_anything
+        @report_dataset = false
+      end
+    end
+    run if run_already
+  end
+
+  # ========================================================================= #
+  # === reset
+  # ========================================================================= #
+  def reset
+    super()
+    # ======================================================================= #
+    # === @dataset
+    # ======================================================================= #
+    @dataset = nil
+    # ======================================================================= #
+    # === @report_dataset
+    # ======================================================================= #
+    @report_dataset = true
+  end
+
+  # ========================================================================= #
+  # === determine_dataset
+  # ========================================================================= #
+  def determine_dataset(i = nil)
+    if i.is_a? Array
+      i = i.first
+    end
+    if i.nil?
+      e 'Please provide some input to this class.'
+      return
+    end
+    # ======================================================================= #
+    # === Handle existing .gb files here
+    # ======================================================================= #
+    if File.exist?(i) and File.file?(i)
+      i = File.read(i)
+    end
+    case i # case tag
+    # ======================================================================= #
+    # === --test
+    #
+    # This entry point can be used to test the default TEST_STRING.
+    # ======================================================================= #
+    when nil,
+         /^-?-?test$/i,
+         /^-?-?test(-|_)?string$/i
+      i = TEST_STRING
+    end
+    # ======================================================================= #
+    # Store the results in the following array.
+    # ======================================================================= #
+    array = []
+    open = false
+    splitted = i.split(N)
+    splitted.each {|line|
+      if line.start_with? '//'
+        open = false
+      end
+      if open
+        array.last << line.strip.delete(' ').gsub(/\d/,'')
+        array.flatten!
+      end
+      if line.start_with? 'ORIGIN'
+        open = true
+        array << [] # Append a new Array.
+      end
+    }
+    string = array.first
+    string.upcase! if UPCASE_SEQUENCE
+    @dataset = string # Assign to the main @dataset here.
+  end
+
+  # ========================================================================= #
+  # === report_dataset?
+  # ========================================================================= #
+  def report_dataset?
+    @report_dataset
+  end
+
+  # ========================================================================= #
+  # === dataset?
+  # ========================================================================= #
+  def dataset?
+    @dataset
+  end; alias sequence? dataset? # === sequence?
+
+  # ========================================================================= #
+  # === report_dataset                                           (report tag)
+  #
+  # Simply output the dataset without any further processing to be done.
+  # ========================================================================= #
+  def report_dataset
+    if @dataset
+      e @dataset if report_dataset?
+    end
+  end; alias report report_dataset # === report
+
+  # ========================================================================= #
+  # === run
+  # ========================================================================= #
+  def run
+    report_dataset
+  end
+
+  # ========================================================================= #
+  # === Bioroebe::GenbankParser[]
+  # ========================================================================= #
+  def self.[](i)
+    new(i)
+  end
+
+end; end
+
+if __FILE__ == $PROGRAM_NAME
+  Bioroebe::GenbankParser.new(ARGV) # Report the dataset as well.
+end # genbankparser
+    # genbankparser sample_file.genbank
+    # genbankparser --test

data/lib/bioroebe/gene/gene.rb

@@ -0,0 +1,64 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# === Bioroebe::Gene
+#
+# Every Gene has a DNA sequence. This is mandatory (there is no gene
+# without a DNA sequence, by definition; though we could accept the
+# mRNA as a representative towards DNA, with the enzyme reverse
+# transcriptase) - but the sequence can be '', hence empty.
+#
+# Every Gene may have a name - this is optional, though, as it is man-made.
+# =========================================================================== #
+# require 'bioroebe/gene/gene.rb'
+# =========================================================================== #
+require 'bioroebe/requires/require_sequence.rb'
+
+module Bioroebe
+
+class Gene < Sequence # === Bioroebe::Gene
+
+  # ========================================================================= #
+  # === initialize
+  # ========================================================================= #
+  def initialize(
+      sequence_to_use        = '',
+      optional_use_this_name = nil
+    )
+    reset
+    super(sequence_to_use)
+    if optional_use_this_name
+      set_name(optional_use_this_name)
+    end
+  end
+
+  # ========================================================================= #
+  # === reset                                                     (reset tag)
+  # ========================================================================= #
+  def reset
+    # ======================================================================= #
+    # === @hash
+    # ======================================================================= #
+    @hash = {}
+  end
+
+  # ========================================================================= #
+  # === set_name
+  # ========================================================================= #
+  def set_name(i)
+    @hash[:name] = i
+  end
+
+  # ========================================================================= #
+  # === name?
+  # ========================================================================= #
+  def name?
+    @hash.fetch :name
+  end
+
+end; end
+
+if __FILE__ == $PROGRAM_NAME
+  Bioroebe::Gene.new
+end

data/lib/bioroebe/genomes/genome_pattern.rb

@@ -0,0 +1,165 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# === Bioroebe::GenomePattern
+#
+# This will allow us to "reshuffle" genes in an organism, but purely from
+# a theoretical point of view alone.
+#
+# It must be able to respond to methods such as:
+#
+#   add_gene(:foo)
+#   remove_gene(:bla)
+#   genome? # return an Array of all genes available in said organism.
+#
+# Usage example:
+#
+#   Bioroebe::GenomePattern.new
+#
+# =========================================================================== #
+# require 'bioroebe/genomes/genome_pattern.rb'
+# =========================================================================== #
+require 'bioroebe/base/commandline_application/commandline_application.rb'
+
+module Bioroebe
+
+class GenomePattern < ::Bioroebe::CommandlineApplication # === Bioroebe::GenomePattern
+
+  require 'bioroebe/gene/gene.rb'
+
+  # ========================================================================= #
+  # === NAMESPACE
+  # ========================================================================= #
+  NAMESPACE = inspect
+
+  # ========================================================================= #
+  # === initialize
+  # ========================================================================= #
+  def initialize(
+      i           = nil,
+      run_already = true
+    )
+    reset
+    set_cmdline(i)
+    run if run_already
+  end
+
+  # ========================================================================= #
+  # === reset                                                     (reset tag)
+  # ========================================================================= #
+  def reset
+    super()
+    # ======================================================================= #
+    # === @namespace
+    # ======================================================================= #
+    @namespace = NAMESPACE
+    # ======================================================================= #
+    # === @hash
+    # ======================================================================= #
+    @hash = {}
+    @hash[:array_genes] = []
+    @hash[:be_verbose] = true # If true then we tell the user what we will do.
+  end
+
+  # ========================================================================= #
+  # === be_verbose?
+  # ========================================================================= #
+  def be_verbose?
+    @hash[:be_verbose]
+  end
+
+  # ========================================================================= #
+  # === set_cmdline
+  # ========================================================================= #
+  def set_cmdline(i = '')
+    i = [i] unless i.is_a? Array
+    @hash[:cmdline] = i
+  end
+
+  # ========================================================================= #
+  # === cmdline?
+  # ========================================================================= #
+  def cmdline?
+    @hash[:cmdline]
+  end; alias input? cmdline? # === input?
+
+  # ========================================================================= #
+  # === array_genes?
+  # ========================================================================= #
+  def array_genes?
+    @hash[:array_genes]
+  end
+
+  # ========================================================================= #
+  # === add_gene
+  # ========================================================================= #
+  def add_gene(i = :foo)
+    _ = array_genes?
+    unless _.include? i
+      if be_verbose?
+        opnn; erev "Now adding the gene `#{sfancy(i.to_s)}#{rev}`."
+      end
+      _ << i
+    end
+  end; alias add add_gene # === add
+       alias <<  add_gene # === <<
+
+  # ========================================================================= #
+  # === genome?
+  #
+  # return an Array of all genes available in said organism.
+  # ========================================================================= #
+  def genome?
+    array_genes?
+  end; alias genes? genome? # === genes?
+
+  # ========================================================================= #
+  # === report
+  # ========================================================================= #
+  def report
+    if genome?.empty?
+      opnn; erev 'We have no genes stored.'
+    else
+      opnn; erev 'We have these genes:'
+      genes?.each {|entry|
+        opnn; erev '  '+sfancy(entry.to_s)
+      }
+    end
+  end
+
+  # ========================================================================= #
+  # === remove_gene
+  #
+  # Use this method to remove a Gene again.
+  # ========================================================================= #
+  def remove_gene(i = :foo, be_verbose = be_verbose?)
+    case be_verbose
+    when :be_verbose
+      be_verbose = true
+    end
+    _ = array_genes?
+    if _.include? i
+      if be_verbose
+        opnn; erev "Now removing the gene `#{sfancy(i.to_s)}#{rev}`."
+      end
+      _.delete i
+    end
+  end; alias remove remove_gene # === remove
+
+  # ========================================================================= #
+  # === run                                                         (run tag)
+  # ========================================================================= #
+  def run
+  end
+
+end; end
+
+if __FILE__ == $PROGRAM_NAME
+  _ = Bioroebe::GenomePattern.new(ARGV)
+  _ << 'yo'
+  _ << 'there'
+  _.remove 'yo', :be_verbose
+  _.remove 'ya'
+  _.report
+end # genomepattern

data/lib/bioroebe/genomes/genome_retriever.rb

@@ -0,0 +1,79 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# === Bioroebe::GenomeRetriever
+#
+# This class will retrieve genome sequences from organisms.
+#
+# This is incomplete - but it will be kept, so that we may extend it
+# at some later point in the future.
+# =========================================================================== #
+# require 'bioroebe/genomes/genome_retriever.rb'
+# =========================================================================== #
+require 'bioroebe/base/commandline_application/commandline_application.rb'
+
+module Bioroebe
+
+class GenomeRetriever < ::Bioroebe::CommandlineApplication # === Bioroebe::GenomeRetriever
+
+  # ========================================================================= #
+  # === DEFAULT_NAME
+  # ========================================================================= #
+  DEFAULT_NAME = :humans
+
+  # ========================================================================= #
+  # === initialize
+  # ========================================================================= #
+  def initialize(
+      name_of_genome = DEFAULT_NAME
+    )
+    reset
+    fetch(name_of_genome)
+  end
+
+  # ========================================================================= #
+  # === reset
+  # ========================================================================= #
+  def reset
+    super()
+  end
+
+  # ========================================================================= #
+  # === fetch
+  #
+  # Use this to fetch remote data.
+  # ========================================================================= #
+  def fetch(what)
+    case what.to_sym
+    when :human,
+         :humans
+      open_via_the_browser 'https://www.genome.jp/dbget-bin/www_bget?hsa+20+21+22'
+    end
+  end
+
+  # ========================================================================= #
+  # === open_via_the_browser
+  # ========================================================================= #
+  def open_via_the_browser(
+      url, use_this_browser = :firefox
+    )
+    system_cmd "#{use_this_browser} --new-tab #{url}"
+  end
+
+  # ========================================================================= #
+  # === system_cmd
+  # ========================================================================= #
+  def system_cmd(this_cmd)
+    e 'Running this command next:'
+    e
+    e "  #{sfancy(this_cmd)}"
+    e
+    system this_cmd  
+  end
+
+end; end
+
+if __FILE__ == $PROGRAM_NAME
+  Bioroebe::GenomeRetriever.new
+end # genomeretriever

data/lib/bioroebe/gui/experimental/README.md

@@ -0,0 +1 @@
+This directory exists just for testing purposes really.