bioroebe 0.10.80 → 0.12.24

data/lib/bioroebe/sequence/sequence.rb

@@ -188,6 +188,27 @@ class Sequence < ::Bioroebe::RawSequence
     @internal_hash[:type]
   end; alias type type? # === type
 
+  # ========================================================================= #
+  # === index
+  # ========================================================================= #
+  def index(i)
+    @sequence.index(i)
+  end
+
+  # ========================================================================= #
+  # === map
+  # ========================================================================= #
+  def map(&block)
+    @sequence.map(&block)
+  end
+
+  # ========================================================================= #
+  # === size?
+  # ========================================================================= #
+  def size?
+    @sequence.size
+  end
+
   # ========================================================================= #
   # === to_regexp
   #
@@ -245,25 +266,6 @@ class Sequence < ::Bioroebe::RawSequence
     @internal_hash[:description]
   end; alias desc? description? # === desc?
 
-  # ========================================================================= #
-  # === return_string_nucleotides_or_aminoacids
-  #
-  # This will either return the String "nucleotides" or "aminoacids".
-  #
-  # This functionality may be useful in downstream applications that
-  # try to display the correct terminology/word.
-  # ========================================================================= #
-  def return_string_nucleotides_or_aminoacids(
-      type = type?
-    )
-    case type
-    when :rna, :dna
-      'nucleotides'
-    when :protein
-      'aminoacids'
-    end
-  end; alias nucleotides_or_aminoacids? return_string_nucleotides_or_aminoacids # === nucleotides_or_aminoacids
-
   # ========================================================================= #
   # === n_uracil?
   #
@@ -334,7 +336,7 @@ class Sequence < ::Bioroebe::RawSequence
       # ===================================================================== #
       sequence?.tr!('T','U') if sequence?
     end
-  end
+  end; alias normalize sanitize_dataset # === normalize
 
   # ========================================================================= #
   # === set_type
@@ -372,13 +374,6 @@ class Sequence < ::Bioroebe::RawSequence
     @internal_hash[:save_file] = i
   end
 
-  # ========================================================================= #
-  # === is_a_protein?
-  # ========================================================================= #
-  def is_a_protein?
-    @internal_hash[:type] == :protein
-  end; alias is_protein? is_a_protein? # === is_protein?
-
   # ========================================================================= #
   # === is_a_protein_now
   #
@@ -403,18 +398,6 @@ class Sequence < ::Bioroebe::RawSequence
     @internal_hash[:type] == :rna
   end; alias is_rna? is_RNA? # === is_rna?
 
-  require 'bioroebe/toplevel_methods/infer.rb'
-  # ========================================================================= #
-  # === infer_type
-  # ========================================================================= #
-  def infer_type
-    set_type(
-      Bioroebe.infer_type_from_this_sequence(
-        sequence?
-      )
-    )
-  end
-
   # ========================================================================= #
   # === set_description
   #
@@ -484,7 +467,8 @@ class Sequence < ::Bioroebe::RawSequence
       # ===================================================================== #
       # === :do_not_downcase
       # ===================================================================== #
-      when :do_not_downcase
+      when :do_not_downcase,
+           :make_no_modification
         # Make no modification in this case.
       # ===================================================================== #
       # === :do_upcase
@@ -526,6 +510,16 @@ class Sequence < ::Bioroebe::RawSequence
   end; alias set_rna_type   set_rna # === set_rna_type
        alias convert_to_rna set_rna # === convert_to_rna
 
+  # ========================================================================= #
+  # === automatic_support_for_nucleotides
+  #
+  # This adds automatic support for RNA and DNA to this sequence object.
+  # ========================================================================= #
+  def automatic_support_for_nucleotides
+    require 'bioroebe/sequence/nucleotide_module/nucleotide_module.rb'
+    extend(Bioroebe::NucleotideModule)
+  end
+
   # ========================================================================= #
   # === set_dna
   # ========================================================================= #
@@ -544,6 +538,82 @@ class Sequence < ::Bioroebe::RawSequence
     }.join
   end
 
+  # ========================================================================= #
+  # === remove_invalid_entries_from_the_dna_sequence!
+  # ========================================================================= #
+  def remove_invalid_entries_from_the_dna_sequence!(i = sequence?)
+    result = i.chars.select {|character|
+      DNA_NUCLEOTIDES.include? character.upcase
+    }.join
+    set_sequence(result)
+  end
+
+  # ========================================================================= #
+  # === return_string_nucleotides_or_aminoacids
+  #
+  # This will either return the String "nucleotides" or "aminoacids".
+  #
+  # This functionality may be useful in downstream applications that
+  # try to display the correct terminology/word.
+  # ========================================================================= #
+  def return_string_nucleotides_or_aminoacids(
+      type = type?
+    )
+    case type
+    # ======================================================================= #
+    # === :rna
+    # ======================================================================= #
+    when :rna,
+         :dna
+      'nucleotides'
+    # ======================================================================= #
+    # === :protein
+    # ======================================================================= #
+    when :protein
+      'aminoacids'
+    end
+  end; alias nucleotides_or_aminoacids? return_string_nucleotides_or_aminoacids # === nucleotides_or_aminoacids
+
+  # ========================================================================= #
+  # === randomize
+  #
+  # Usage example:
+  #
+  #   x = Bioroebe::Sequence.new; x.randomize
+  #
+  # ========================================================================= #
+  def randomize(
+      i = { 'A'=>1,'C'=>2,'G'=>3,'T'=>4 } 
+    )
+    if i.is_a? Hash
+      i = i.map{|key, value| "#{key * value}" }.join
+    end
+    ::Bioroebe.random_dna(size?, i) # => "GGTAGGGGGGGGTAGGGGGG"
+  end
+
+  # ========================================================================= #
+  # === is_a_protein?
+  # ========================================================================= #
+  def is_a_protein?
+    @internal_hash[:type] == :protein
+  end; alias is_protein? is_a_protein? # === is_protein?
+
+  require 'bioroebe/toplevel_methods/infer.rb'
+  # ========================================================================= #
+  # === infer_type
+  #
+  # This method attempts to determine whether the main sequence is a DNA,
+  # RNA or protein. Right now this is not very sophisticated, so we have
+  # to improve this at a later time (April 2023).
+  # ========================================================================= #
+  def infer_type
+    set_type(
+      Bioroebe.infer_type_from_this_sequence(
+        sequence?
+      )
+    )
+  end; alias try_to_infer_the_type infer_type # === try_to_infer_the_type
+
   # ========================================================================= #
   # === Bioroebe::Sequence.sequence_from_file
   #
@@ -658,6 +728,24 @@ def self.create_new_sequence(i = ARGV, &block)
   @sequence = ::Bioroebe::Sequence.new(i, &block)
 end; self.instance_eval { alias sequence create_new_sequence } # === Bioroebe.sequence
 
+# =========================================================================== #
+# === Bioroebe.is_a_protein?
+#
+# Usage example:
+#
+#   puts Bioroebe.is_a_protein?('AGATCGTACTGAGCTATCGTAC') # => false
+#   puts Bioroebe.is_a_protein?('WWW') # => true
+#
+# =========================================================================== #
+def self.is_a_protein?(i = ARGV, &block)
+  if i.is_a? Array
+    i = i.join(' ').strip
+  end
+  _ = Bioroebe.sequence(i)
+  _.try_to_infer_the_type
+  return _.is_a_protein?
+end
+
 end
 
 # =========================================================================== #