rbfam 0.0.6 → 0.0.7

data/lib/rbfam.rb

@@ -4,7 +4,9 @@ require "entrez"
 require "httparty"
 require "active_support/inflector"
 
-Dir[File.join(File.dirname(__FILE__), "rbfam", "modules", "*.rb")].each { |name| require "rbfam/modules/#{File.basename(name, '.rb')}" }
+%W|helpers modules|.each do |folder|
+  Dir[File.join(File.dirname(__FILE__), "rbfam", folder, "*.rb")].each { |name| require "rbfam/#{folder}/#{File.basename(name, '.rb')}" }
+end
 
 module Rbfam
   def self.script(name)

data/lib/rbfam/helpers/common.rb

@@ -0,0 +1,14 @@
+module Rbfam
+  module CommonHelpers
+    def self.included(base)
+      base.send(:include, InstanceMethods)
+    end
+    
+    module InstanceMethods
+      def entries!(options = {})
+        remove_instance_variable(:@parsed_entries)
+        entries(options)
+      end
+    end
+  end
+end

data/lib/rbfam/modules/alignment.rb

@@ -1,5 +1,7 @@
 module Rbfam
   class Alignment
+    include Rbfam::CommonHelpers
+    
     LINE_REGEXP = /^([\w\.]+)\/(\d+)\-(\d+)\s+([AUGC\.]+)$/
     
     attr_reader :family, :seed
@@ -8,13 +10,15 @@ module Rbfam
       @family = family
     end
     
-    def entries(alignment = :seed)
-      @parsed_entries ||= pull_from_server(alignment).split(/\n/).reject do |line| 
+    def entries(options = {})
+      options = { alignment: :seed, limit: false }.merge(options)
+      
+      @parsed_entries ||= pull_from_server(options[:alignment]).split(/\n/).reject do |line| 
         line =~ /^#/
       end.select do |line| 
         line =~ LINE_REGEXP
-      end.map(&method(:parse_line)).tap do
-        @seed = alignment == :seed
+      end[options[:limit] ? 0...options[:limit] : 0..-1].map(&method(:parse_line)).tap do
+        @seed = options[:alignment] == :seed
       end
     end
     
@@ -23,11 +27,9 @@ module Rbfam
     end
     
     def load_entries!(options = {})
-      Rbfam.script("sequences_in_mysql")
+      options = { extended: false }.merge(options)
       
-      @parsed_entries = SequenceTable.where({ family: family.family_name }.merge(options)).map do |entry|
-        entry.to_rbfam_sequence(family)
-      end
+      @parsed_entries = family.load_entries!(options)
     end
     
     private
@@ -43,7 +45,7 @@ module Rbfam
         puts "RESPONSE: 200 OK"
         party.parsed_response
       else
-        raise RuntimeError.new("HTTParty raised the following error when retrieving family %s: %s %s" % [
+        raise RuntimeError.new("HTTParty raised the following error when retrieving alignment %s: %s %s" % [
           family_name,
           party.response.code,
           party.response.message

data/lib/rbfam/modules/family.rb

@@ -1,11 +1,17 @@
 module Rbfam
   class Family
+    include Rbfam::CommonHelpers
+    
     attr_reader :family_name
     
     class << self
       def purine;  new("RF00167"); end
       def tpp;     new("RF00059"); end
       def secis_1; new("RF00031"); end
+      def trna;    new("RF00005"); end
+      def let_7;   new("RF00027"); end
+      def snora71; new("RF00056"); end
+      def u7;      new("RF00066"); end
     end
     
     def initialize(family_name)
@@ -16,15 +22,19 @@ module Rbfam
       Rbfam::Alignment.new(self)
     end
     
-    def entries
-      @parsed_entries ||= pull_from_server.split(/\n/).reject { |line| line =~ /^#/ }.map(&method(:parse_line))
+    def entries(options = {})
+      options = { limit: false }.merge(options)
+      
+      @parsed_entries ||= pull_from_server.split(/\n/).reject { |line| line =~ /^#/ }[options[:limit] ? 0...options[:limit] : 0..-1].map(&method(:parse_line))
     end
     
     def load_entries!(options = {})
+      options = { extended: false }.merge(options)
+      
       Rbfam.script("sequences_in_mysql")
       
       @parsed_entries = SequenceTable.where({ family: family_name }.merge(options)).map do |entry|
-        entry.to_rbfam_sequence(self)
+        entry.to_rbfam(self)
       end
     end
     

data/lib/rbfam/modules/sequence.rb

@@ -1,16 +1,20 @@
 module Rbfam
   class Sequence
-    attr_reader :family, :accession, :from, :to
+    attr_reader :family, :accession, :from, :to, :coord_options
 
     def initialize(family, accession, from, to, options = {})
-      @family, @accession, @from, @to = family, accession, from, to
+      @family        = family
+      @accession     = accession
+      @from          = from
+      @to            = to
+      @coord_options = options[:autoload].is_a?(Hash) ? options[:autoload] : {}
       
       if options[:sequence]
-        @raw_sequence = (options[:sequence].is_a?(String) ? Bio::Sequence::NA.new(options[:sequence]) : options[:sequence])
+        @raw_sequence = (options[:sequence].is_a?(String) ? Bio::Sequence::NA.new(options[:sequence]) : options[:sequence]).upcase
       end
       
       if options[:autoload]
-        sequence(options[:autoload].is_a?(Hash) ? options[:autoload] : {})
+        sequence
       end
     end
     
@@ -24,8 +28,8 @@ module Rbfam
         sequence_length: sequence.length, 
         from:            from, 
         to:              to, 
-        seq_from:        up_coord + coord_window(options[:coord_window] || {}).min, 
-        seq_to:          up_coord + coord_window(options[:coord_window] || {}).max,
+        seq_from:        seq_from, 
+        seq_to:          seq_to,
         seed:            options[:seed]
       })
     end
@@ -38,6 +42,14 @@ module Rbfam
       [from, to].max
     end
     
+    def seq_from
+      up_coord + coord_window.min
+    end
+    
+    def seq_to
+      up_coord + coord_window.max
+    end
+    
     def strand
       plus_strand? ? :plus : :minus
     end
@@ -50,8 +62,8 @@ module Rbfam
       !plus_strand?
     end
 
-    def sequence(options = {})
-      @raw_sequence ||= Rbfam::Utils.rna_sequence_from_entrez(accession, up_coord, coord_window(options))
+    def sequence
+      @raw_sequence ||= Rbfam::Utils.rna_sequence_from_entrez(accession, up_coord, coord_window)
       @raw_sequence   = minus_strand? ? @raw_sequence.complement : @raw_sequence
     end
     
@@ -61,23 +73,27 @@ module Rbfam
       @mfe_structure ||= ViennaRna::Fold.run(seq).structure
     end
     
+    def description
+      ("%s %s %s" % [accession, from, to]).gsub(/\W+/, "_")
+    end
+    
     def fftbor
       @fftbor ||= ViennaRna::Fftbor.run(seq: seq, str: mfe_structure)
     end
     
-    def coord_window(options = {})
-      # Ex: { length: 300, extend: 3 }
+    def coord_window
+      # Options from @coord_options ex: { length: 300, extend: 3 }
       
       range = 0..(down_coord - up_coord)
       
-      if options[:length] && options[:extend]
-        if range.count < options[:length]
-          length_difference = options[:length] - range.count
+      if @coord_options[:length] && @coord_options[:extend]
+        if range.count < @coord_options[:length]
+          length_difference = @coord_options[:length] - range.count
           
-          case [options[:extend], strand]
+          case [@coord_options[:extend], strand]
           when [3, :plus], [5, :minus] then Range.new(range.min, range.max + length_difference)
           when [5, :plus], [3, :minus] then Range.new(range.min - length_difference, range.max)
-          else puts "WARNING: value for :extend key in sequence retreival needs to be one of 5, 3 - found (%s)" % options[:extend]
+          else puts "WARNING: value for :extend key in sequence retreival needs to be one of 5, 3 - found (%s)" % @coord_options[:extend]
           end
         else
           puts "WARNING: %s %d-%d (%s) is length %d, but only %d nt. have been requested. Providing the full sequence anyways." % [
@@ -86,12 +102,16 @@ module Rbfam
             to,
             strand,
             range.count,
-            options[:length]
+            @coord_options[:length]
           ]
         end
       else
         range
       end
     end
+    
+    def inspect
+      "#<Rbfam::Sequence #{description} #{seq[0, 20] + ('...' if seq.length > 20)}>"
+    end
   end
 end

data/lib/rbfam/scripts/sequences_in_mysql.rb

@@ -1,3 +1,5 @@
+# SequenceTable(id: integer, family: string, accession: string, sequence: text, sequence_length: integer, from: integer, to: integer, seq_from: integer, seq_to: integer, seed: boolean, created_at: datetime, updated_at: datetime, extended: boolean)
+
 require "mysql2"
 require "active_record"
 
@@ -6,7 +8,7 @@ class Object; def this; self; end; end
 class SequenceTable < ActiveRecord::Base
   self.table_name = "sequences"
   
-  validates_uniqueness_of :accession, scope: [:sequence, :structure]
+  validates_uniqueness_of :accession, scope: [:sequence, :seq_from, :seq_to]
   
   def self.connect
     ActiveRecord::Base.establish_connection(config = { adapter: "mysql2", username: "root", reconnect: true })
@@ -20,34 +22,10 @@ class SequenceTable < ActiveRecord::Base
     inline_rails if defined?(inline_rails)
   end
   
-  def clean_description
-    ("%s %s %s" % [accession, seq_from, seq_to]).gsub(/[^A-Za-z0-9]/, "_")
-  end
-  
-  def to_rbfam_sequence(family = nil)
+  def to_rbfam(family = nil)
+    # Should use a singleton pattern here to look up the family.
     Rbfam::Sequence.new(family || Rbfam::Family.new(family), accession, from, to, sequence: sequence)
   end
 end
 
-SequenceTable.connect
-
-class BuildSequence < ActiveRecord::Migration
-  def self.up
-    create_table :sequences do |table|
-      table.string  :family
-      table.string  :accession
-      table.text    :sequence
-      table.integer :sequence_length
-      table.integer :from
-      table.integer :to
-      table.integer :seq_from
-      table.integer :seq_to
-      table.boolean :seed, default: false
-      table.timestamps
-    end 
-  end
-end
-
-unless ActiveRecord::Base.connection.execute("show tables").map(&:this).flatten.include?("sequences")
-  BuildSequence.up
-end
+SequenceTable.connect

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: rbfam
 version: !ruby/object:Gem::Version
-  version: 0.0.6
+  version: 0.0.7
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-08-16 00:00:00.000000000 Z
+date: 2012-09-05 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bio
@@ -81,6 +81,7 @@ executables: []
 extensions: []
 extra_rdoc_files: []
 files:
+- lib/rbfam/helpers/common.rb
 - lib/rbfam/modules/alignment.rb
 - lib/rbfam/modules/family.rb
 - lib/rbfam/modules/sequence.rb