bio-ucsc-api 0.2.0 → 0.2.1

data/lib/bio-ucsc/saccer2.rb

@@ -0,0 +1,33 @@
+#
+# = AUTOMATIC Table Definition of the S. cerevisiae June 2008 (SGD/sacCer2) assembly
+# Copyright::   Copyright (C) 2011
+#               MISHIMA, Hiroyuki
+#               <missy at be.to / hmishima at nagasaki-u.ac.jp> 
+# License::     The Ruby licence (Ryby's / GPLv2 dual)
+#
+
+require "#{File.dirname(__FILE__)}/saccer2/db_connection"
+require "#{File.dirname(__FILE__)}/table_class_detector"
+
+module Bio
+  module Ucsc
+    module SacCer2
+      CHROMS = 
+%w(chrI chrII chrIII chrIV chrV chrVI chrVII chrVIII chrIX chrX
+chrXI chrXII chrXIII chrXIV chrXV chrXVI chrM 2micron)
+      extend TableClassDetector
+
+      base = "#{File.dirname(__FILE__)}/saccer2"
+      autoload :T2micron_est, "#{base}/t2micron_est"
+      autoload :T2micron_gap, "#{base}/t2micron_gap"
+      autoload :T2micron_gold, "#{base}/t2micron_gold"
+      autoload :T2micron_intronEst, "#{base}/t2micron_intronEst"
+      autoload :T2micron_mrna, "#{base}/t2micron_mrna"
+      autoload :Est, "#{base}/est"
+      autoload :Gap, "#{base}/gap"
+      autoload :Gold, "#{base}/gold"
+      autoload :IntronEst, "#{base}/intronest"
+      autoload :Mrna, "#{base}/mrna"
+    end
+  end
+end

data/lib/bio-ucsc/saccer2/db_connection.rb

@@ -0,0 +1,55 @@
+#
+# = db_connection.rb - UCSC DB connection
+#
+# Copyright::   Cioyrught (C) 2011
+#               MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
+# Copyright::   Copyright (C) 2010
+#               Jan Aerts <jan.aerts@gmail.com>
+# License::     Ruby licence (Ryby's / GPLv2 dual)
+
+require 'rubygems'
+require 'active_record'
+
+module Bio
+  module Ucsc
+
+    module SacCer2
+      # = DESCRIPTION
+      # Actual connection established to the UCSC mysql server.
+      class DBConnection < ActiveRecord::Base
+        # = DESCRIPTION
+        # The #connect method make the connection
+        #
+        # = USAGE
+        #  Bin::Ucsc::DB_NAME::DBConnection.connect
+        #
+        # ---
+        # *Arguments*: none
+        @@db_adapter  ||= 'mysql'
+        @@db_host     ||= 'genome-mysql.cse.ucsc.edu'
+        @@db_username ||= 'genome'
+        @@db_password ||= ''
+
+        cattr_accessor :db_adapter, :db_host, :db_username, :db_password
+
+        self.abstract_class = true
+
+        def self.default
+          @@db_adapter  = 'mysql'
+          @@db_host     = 'genome-mysql.cse.ucsc.edu'
+          @@db_username = 'genome'
+          @@db_password = ''
+        end
+
+        def self.connect
+          establish_connection({ :adapter =>  @@db_adapter,
+                                 :host =>     @@db_host,
+                                 :database => 'sacCer2',
+                                 :username => @@db_username,
+                                 :password => @@db_password, })
+        end
+      end
+    end # module
+    
+  end # module Ucsc
+end # module Bio

data/lib/bio-ucsc/saccer2/est.rb

@@ -0,0 +1,83 @@
+# Copyright::
+#  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
+# License::     The Ruby licence (Ryby's / GPLv2 dual)
+#
+# this table is actually separated
+# into "chr1_*", "chr2_*", etc. This class dynamically
+# define *::Chr1_*, *::Chr2_*, etc. The
+# find_by_interval calls an appropreate class automatically.
+
+module Bio
+  module Ucsc
+    module SacCer2
+
+      class Est
+        KLASS = "Est"
+        KLASS_S = KLASS[0..0].downcase + KLASS[1..-1]
+
+        def self.find_by_interval(interval, opt = {:partial => true})
+          chrom = interval.chrom[0..0].upcase + interval.chrom[1..-1]
+          chr_klass = self.const_get("#{chrom}_#{KLASS}")
+          chr_klass.__send__(:find_by_interval, interval, opt)
+        end
+
+        def self.find_all_by_interval(interval, opt = {:partial => true})
+          chrom = interval.chrom[0..0].upcase + interval.chrom[1..-1]
+          chr_klass = self.const_get("#{chrom}_#{KLASS}")
+          chr_klass.__send__(:find_all_by_interval, interval, opt)
+        end
+
+        Bio::Ucsc::SacCer2::CHROMS.each do |chr|
+          klass_header = chr[0..0].upcase + chr[1..-1]
+          klass_header.sub!(/(\A[0-9])/, "T#$1")
+          class_eval %!
+            class #{klass_header}_#{KLASS} < DBConnection
+              set_table_name "#{chr[0..0].downcase + chr[1..-1]}_#{KLASS_S}"
+              set_primary_key nil
+              set_inheritance_column nil
+
+              def self.find_by_interval(interval, opt = {:partial => true})
+                find_first_or_all_by_interval(interval, :first, opt)
+              end
+        
+              def self.find_all_by_interval(interval, opt = {:partial => true})
+                find_first_or_all_by_interval(interval, :all, opt)
+              end
+
+              def self.find_first_or_all_by_interval(interval, first_all, opt)
+                zstart = interval.zero_start
+                zend   = interval.zero_end
+                if opt[:partial] == true
+                  where = <<-SQL
+      tName = :chrom
+AND   bin in (:bins)
+AND ((tStart BETWEEN :zstart AND :zend)
+ OR  (tEnd BETWEEN :zstart AND :zend)
+ OR  (tStart <= :zstart AND tEnd >= :zend))
+                  SQL
+                else
+                  where = <<-SQL
+      tName = :chrom
+AND   bin in (:bins)
+AND ((tStart BETWEEN :zstart AND :zend)
+AND  (tEnd BETWEEN :zstart AND :zend))
+                  SQL
+                end
+                cond = {
+                  :chrom  => interval.chrom,
+                  :bins   => Bio::Ucsc::UcscBin.bin_all(zstart, zend),
+                  :zstart => zstart,
+                  :zend   => zend,
+                }
+                self.find(first_all,
+                          { :select => "*",
+                            :conditions => [where, cond], })
+              end
+            end
+          !
+        end # each chromosome
+      end # class
+
+    end # module 
+  end # module Ucsc
+end # module Bio

data/lib/bio-ucsc/saccer2/gap.rb

@@ -0,0 +1,83 @@
+# Copyright::
+#  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
+# License::     The Ruby licence (Ryby's / GPLv2 dual)
+#
+# this table is actually separated
+# into "chr1_*", "chr2_*", etc. This class dynamically
+# define *::Chr1_*, *::Chr2_*, etc. The
+# find_by_interval calls an appropreate class automatically.
+
+module Bio
+  module Ucsc
+    module SacCer2
+
+      class Gap
+        KLASS = "Gap"
+        KLASS_S = KLASS[0..0].downcase + KLASS[1..-1]
+
+        def self.find_by_interval(interval, opt = {:partial => true})
+          chrom = interval.chrom[0..0].upcase + interval.chrom[1..-1]
+          chr_klass = self.const_get("#{chrom}_#{KLASS}")
+          chr_klass.__send__(:find_by_interval, interval, opt)
+        end
+
+        def self.find_all_by_interval(interval, opt = {:partial => true})
+          chrom = interval.chrom[0..0].upcase + interval.chrom[1..-1]
+          chr_klass = self.const_get("#{chrom}_#{KLASS}")
+          chr_klass.__send__(:find_all_by_interval, interval, opt)
+        end
+
+        Bio::Ucsc::SacCer2::CHROMS.each do |chr|
+          klass_header = chr[0..0].upcase + chr[1..-1]
+          klass_header.sub!(/(\A[0-9])/, "T#$1")
+          class_eval %!
+            class #{klass_header}_#{KLASS} < DBConnection
+              set_table_name "#{chr[0..0].downcase + chr[1..-1]}_#{KLASS_S}"
+              set_primary_key nil
+              set_inheritance_column nil
+
+              def self.find_by_interval(interval, opt = {:partial => true})
+                find_first_or_all_by_interval(interval, :first, opt)
+              end
+        
+              def self.find_all_by_interval(interval, opt = {:partial => true})
+                find_first_or_all_by_interval(interval, :all, opt)
+              end
+
+              def self.find_first_or_all_by_interval(interval, first_all, opt)
+                zstart = interval.zero_start
+                zend   = interval.zero_end
+                if opt[:partial] == true
+                  where = <<-SQL
+      chrom = :chrom
+AND   bin in (:bins)
+AND ((chromStart BETWEEN :zstart AND :zend)
+ OR  (chromEnd BETWEEN :zstart AND :zend)
+ OR  (chromStart <= :zstart AND chromEnd >= :zend))
+                  SQL
+                else
+                  where = <<-SQL
+      chrom = :chrom
+AND   bin in (:bins)
+AND ((chromStart BETWEEN :zstart AND :zend)
+AND  (chromEnd BETWEEN :zstart AND :zend))
+                  SQL
+                end
+                cond = {
+                  :chrom  => interval.chrom,
+                  :bins   => Bio::Ucsc::UcscBin.bin_all(zstart, zend),
+                  :zstart => zstart,
+                  :zend   => zend,
+                }
+                self.find(first_all,
+                          { :select => "*",
+                            :conditions => [where, cond], })
+              end
+            end
+          !
+        end # each chromosome
+      end # class
+
+    end # module 
+  end # module Ucsc
+end # module Bio

data/lib/bio-ucsc/saccer2/gold.rb

@@ -0,0 +1,83 @@
+# Copyright::
+#  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
+# License::     The Ruby licence (Ryby's / GPLv2 dual)
+#
+# this table is actually separated
+# into "chr1_*", "chr2_*", etc. This class dynamically
+# define *::Chr1_*, *::Chr2_*, etc. The
+# find_by_interval calls an appropreate class automatically.
+
+module Bio
+  module Ucsc
+    module SacCer2
+
+      class Gold
+        KLASS = "Gold"
+        KLASS_S = KLASS[0..0].downcase + KLASS[1..-1]
+
+        def self.find_by_interval(interval, opt = {:partial => true})
+          chrom = interval.chrom[0..0].upcase + interval.chrom[1..-1]
+          chr_klass = self.const_get("#{chrom}_#{KLASS}")
+          chr_klass.__send__(:find_by_interval, interval, opt)
+        end
+
+        def self.find_all_by_interval(interval, opt = {:partial => true})
+          chrom = interval.chrom[0..0].upcase + interval.chrom[1..-1]
+          chr_klass = self.const_get("#{chrom}_#{KLASS}")
+          chr_klass.__send__(:find_all_by_interval, interval, opt)
+        end
+
+        Bio::Ucsc::SacCer2::CHROMS.each do |chr|
+          klass_header = chr[0..0].upcase + chr[1..-1]
+          klass_header.sub!(/(\A[0-9])/, "T#$1")
+          class_eval %!
+            class #{klass_header}_#{KLASS} < DBConnection
+              set_table_name "#{chr[0..0].downcase + chr[1..-1]}_#{KLASS_S}"
+              set_primary_key nil
+              set_inheritance_column nil
+
+              def self.find_by_interval(interval, opt = {:partial => true})
+                find_first_or_all_by_interval(interval, :first, opt)
+              end
+        
+              def self.find_all_by_interval(interval, opt = {:partial => true})
+                find_first_or_all_by_interval(interval, :all, opt)
+              end
+
+              def self.find_first_or_all_by_interval(interval, first_all, opt)
+                zstart = interval.zero_start
+                zend   = interval.zero_end
+                if opt[:partial] == true
+                  where = <<-SQL
+      chrom = :chrom
+AND   bin in (:bins)
+AND ((chromStart BETWEEN :zstart AND :zend)
+ OR  (chromEnd BETWEEN :zstart AND :zend)
+ OR  (chromStart <= :zstart AND chromEnd >= :zend))
+                  SQL
+                else
+                  where = <<-SQL
+      chrom = :chrom
+AND   bin in (:bins)
+AND ((chromStart BETWEEN :zstart AND :zend)
+AND  (chromEnd BETWEEN :zstart AND :zend))
+                  SQL
+                end
+                cond = {
+                  :chrom  => interval.chrom,
+                  :bins   => Bio::Ucsc::UcscBin.bin_all(zstart, zend),
+                  :zstart => zstart,
+                  :zend   => zend,
+                }
+                self.find(first_all,
+                          { :select => "*",
+                            :conditions => [where, cond], })
+              end
+            end
+          !
+        end # each chromosome
+      end # class
+
+    end # module 
+  end # module Ucsc
+end # module Bio

data/lib/bio-ucsc/saccer2/intronest.rb

@@ -0,0 +1,83 @@
+# Copyright::
+#  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
+# License::     The Ruby licence (Ryby's / GPLv2 dual)
+#
+# this table is actually separated
+# into "chr1_*", "chr2_*", etc. This class dynamically
+# define *::Chr1_*, *::Chr2_*, etc. The
+# find_by_interval calls an appropreate class automatically.
+
+module Bio
+  module Ucsc
+    module SacCer2
+
+      class IntronEst
+        KLASS = "IntronEst"
+        KLASS_S = KLASS[0..0].downcase + KLASS[1..-1]
+
+        def self.find_by_interval(interval, opt = {:partial => true})
+          chrom = interval.chrom[0..0].upcase + interval.chrom[1..-1]
+          chr_klass = self.const_get("#{chrom}_#{KLASS}")
+          chr_klass.__send__(:find_by_interval, interval, opt)
+        end
+
+        def self.find_all_by_interval(interval, opt = {:partial => true})
+          chrom = interval.chrom[0..0].upcase + interval.chrom[1..-1]
+          chr_klass = self.const_get("#{chrom}_#{KLASS}")
+          chr_klass.__send__(:find_all_by_interval, interval, opt)
+        end
+
+        Bio::Ucsc::SacCer2::CHROMS.each do |chr|
+          klass_header = chr[0..0].upcase + chr[1..-1]
+          klass_header.sub!(/(\A[0-9])/, "T#$1")
+          class_eval %!
+            class #{klass_header}_#{KLASS} < DBConnection
+              set_table_name "#{chr[0..0].downcase + chr[1..-1]}_#{KLASS_S}"
+              set_primary_key nil
+              set_inheritance_column nil
+
+              def self.find_by_interval(interval, opt = {:partial => true})
+                find_first_or_all_by_interval(interval, :first, opt)
+              end
+        
+              def self.find_all_by_interval(interval, opt = {:partial => true})
+                find_first_or_all_by_interval(interval, :all, opt)
+              end
+
+              def self.find_first_or_all_by_interval(interval, first_all, opt)
+                zstart = interval.zero_start
+                zend   = interval.zero_end
+                if opt[:partial] == true
+                  where = <<-SQL
+      tName = :chrom
+AND   bin in (:bins)
+AND ((tStart BETWEEN :zstart AND :zend)
+ OR  (tEnd BETWEEN :zstart AND :zend)
+ OR  (tStart <= :zstart AND tEnd >= :zend))
+                  SQL
+                else
+                  where = <<-SQL
+      tName = :chrom
+AND   bin in (:bins)
+AND ((tStart BETWEEN :zstart AND :zend)
+AND  (tEnd BETWEEN :zstart AND :zend))
+                  SQL
+                end
+                cond = {
+                  :chrom  => interval.chrom,
+                  :bins   => Bio::Ucsc::UcscBin.bin_all(zstart, zend),
+                  :zstart => zstart,
+                  :zend   => zend,
+                }
+                self.find(first_all,
+                          { :select => "*",
+                            :conditions => [where, cond], })
+              end
+            end
+          !
+        end # each chromosome
+      end # class
+
+    end # module 
+  end # module Ucsc
+end # module Bio

data/lib/bio-ucsc/saccer2/mrna.rb

@@ -0,0 +1,83 @@
+# Copyright::
+#  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
+# License::     The Ruby licence (Ryby's / GPLv2 dual)
+#
+# this table is actually separated
+# into "chr1_*", "chr2_*", etc. This class dynamically
+# define *::Chr1_*, *::Chr2_*, etc. The
+# find_by_interval calls an appropreate class automatically.
+
+module Bio
+  module Ucsc
+    module SacCer2
+
+      class Mrna
+        KLASS = "Mrna"
+        KLASS_S = KLASS[0..0].downcase + KLASS[1..-1]
+
+        def self.find_by_interval(interval, opt = {:partial => true})
+          chrom = interval.chrom[0..0].upcase + interval.chrom[1..-1]
+          chr_klass = self.const_get("#{chrom}_#{KLASS}")
+          chr_klass.__send__(:find_by_interval, interval, opt)
+        end
+
+        def self.find_all_by_interval(interval, opt = {:partial => true})
+          chrom = interval.chrom[0..0].upcase + interval.chrom[1..-1]
+          chr_klass = self.const_get("#{chrom}_#{KLASS}")
+          chr_klass.__send__(:find_all_by_interval, interval, opt)
+        end
+
+        Bio::Ucsc::SacCer2::CHROMS.each do |chr|
+          klass_header = chr[0..0].upcase + chr[1..-1]
+          klass_header.sub!(/(\A[0-9])/, "T#$1")
+          class_eval %!
+            class #{klass_header}_#{KLASS} < DBConnection
+              set_table_name "#{chr[0..0].downcase + chr[1..-1]}_#{KLASS_S}"
+              set_primary_key nil
+              set_inheritance_column nil
+
+              def self.find_by_interval(interval, opt = {:partial => true})
+                find_first_or_all_by_interval(interval, :first, opt)
+              end
+        
+              def self.find_all_by_interval(interval, opt = {:partial => true})
+                find_first_or_all_by_interval(interval, :all, opt)
+              end
+
+              def self.find_first_or_all_by_interval(interval, first_all, opt)
+                zstart = interval.zero_start
+                zend   = interval.zero_end
+                if opt[:partial] == true
+                  where = <<-SQL
+      tName = :chrom
+AND   bin in (:bins)
+AND ((tStart BETWEEN :zstart AND :zend)
+ OR  (tEnd BETWEEN :zstart AND :zend)
+ OR  (tStart <= :zstart AND tEnd >= :zend))
+                  SQL
+                else
+                  where = <<-SQL
+      tName = :chrom
+AND   bin in (:bins)
+AND ((tStart BETWEEN :zstart AND :zend)
+AND  (tEnd BETWEEN :zstart AND :zend))
+                  SQL
+                end
+                cond = {
+                  :chrom  => interval.chrom,
+                  :bins   => Bio::Ucsc::UcscBin.bin_all(zstart, zend),
+                  :zstart => zstart,
+                  :zend   => zend,
+                }
+                self.find(first_all,
+                          { :select => "*",
+                            :conditions => [where, cond], })
+              end
+            end
+          !
+        end # each chromosome
+      end # class
+
+    end # module 
+  end # module Ucsc
+end # module Bio