bio-ucsc-api 0.2.0 → 0.2.1

data/lib/bio-ucsc/rhemac2/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 RheMac2
+      # = 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 => 'rheMac2',
+                                 :username => @@db_username,
+                                 :password => @@db_password, })
+        end
+      end
+    end # module
+    
+  end # module Ucsc
+end # module Bio

data/lib/bio-ucsc/rn4.rb

@@ -0,0 +1,52 @@
+#
+# = AUTOMATIC Table Definition of the Rat Nov. 2004 (Baylor 3.4/rn4) 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__)}/rn4/db_connection"
+require "#{File.dirname(__FILE__)}/table_class_detector"
+
+module Bio
+  module Ucsc
+    module Rn4
+      CHROMS = 
+%w(chrM chr1 chr2 chr3 chr4 chr5 chr6 chr7 chr8 chr9 chr10
+chr11 chr12 chr13 chr14 chr15 chr16 chr17 chr18 chr19 chr20
+chrX)
+      extend TableClassDetector
+
+      base = "#{File.dirname(__FILE__)}/rn4"
+      autoload :ChainBosTau3, "#{base}/chainbostau3"
+      autoload :ChainBosTau3Link, "#{base}/chainbostau3link"
+      autoload :ChainCanFam2, "#{base}/chaincanfam2"
+      autoload :ChainCanFam2Link, "#{base}/chaincanfam2link"
+      autoload :ChainCavPor3, "#{base}/chaincavpor3"
+      autoload :ChainCavPor3Link, "#{base}/chaincavpor3link"
+      autoload :ChainDanRer4, "#{base}/chaindanrer4"
+      autoload :ChainDanRer4Link, "#{base}/chaindanrer4link"
+      autoload :ChainEquCab1, "#{base}/chainequcab1"
+      autoload :ChainEquCab1Link, "#{base}/chainequcab1link"
+      autoload :ChainGalGal3, "#{base}/chaingalgal3"
+      autoload :ChainGalGal3Link, "#{base}/chaingalgal3link"
+      autoload :ChainMm9, "#{base}/chainmm9"
+      autoload :ChainMm9Link, "#{base}/chainmm9link"
+      autoload :ChainMonDom4, "#{base}/chainmondom4"
+      autoload :ChainMonDom4Link, "#{base}/chainmondom4link"
+      autoload :ChainPanTro2, "#{base}/chainpantro2"
+      autoload :ChainPanTro2Link, "#{base}/chainpantro2link"
+      autoload :ChainRheMac2, "#{base}/chainrhemac2"
+      autoload :ChainRheMac2Link, "#{base}/chainrhemac2link"
+      autoload :ChainXenTro2, "#{base}/chainxentro2"
+      autoload :ChainXenTro2Link, "#{base}/chainxentro2link"
+      autoload :Est, "#{base}/est"
+      autoload :Gap, "#{base}/gap"
+      autoload :Gold, "#{base}/gold"
+      autoload :IntronEst, "#{base}/intronest"
+      autoload :Mrna, "#{base}/mrna"
+      autoload :Rmsk, "#{base}/rmsk"
+    end
+  end
+end

data/lib/bio-ucsc/rn4/chainbostau3.rb

@@ -0,0 +1,79 @@
+# = chainbostau4.rb
+# Copyright::
+#  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
+# License::     The Ruby licence (Ryby's / GPLv2 dual)
+#
+# In the hg18 database, this table is actually separated
+# into "chr1_*", "chr2_*", etc. This class dynamically
+# define *::Chr1_*, *::Chr2_*, etc. The
+# Rmsk.find_by_interval calls an appropreate class automatically.
+
+module Bio
+  module Ucsc
+    module Rn4
+
+      class ChainBosTau3
+        Bio::Ucsc::Rn4::CHROMS.each do |chr|
+          class_eval %!
+            class #{chr[0..0].upcase + chr[1..-1]}_ChainBosTau3 < DBConnection
+              set_table_name "#{chr[0..0].downcase + chr[1..-1]}_chainBosTau3"
+              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
+
+        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}_ChainBosTau3")
+          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}_ChainBosTau3")
+          chr_klass.__send__(:find_all_by_interval, interval, opt)
+        end
+      end # class
+
+    end # module Hg18 
+  end # module Ucsc
+end # module Bio

data/lib/bio-ucsc/rn4/chainbostau3link.rb

@@ -0,0 +1,79 @@
+# = chainbostau3link.rb
+# Copyright::
+#  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
+# License::     The Ruby licence (Ryby's / GPLv2 dual)
+#
+# In the hg18 database, this table is actually separated
+# into "chr1_*", "chr2_*", etc. This class dynamically
+# define *::Chr1_*, *::Chr2_*, etc. The
+# Rmsk.find_by_interval calls an appropreate class automatically.
+
+module Bio
+  module Ucsc
+    module Rn4
+
+      class ChainBosTau3Link
+        Bio::Ucsc::Rn4::CHROMS.each do |chr|
+          class_eval %!
+            class #{chr[0..0].upcase + chr[1..-1]}_ChainBosTau3Link < DBConnection
+              set_table_name "#{chr[0..0].downcase + chr[1..-1]}_chainBosTau3Link"
+              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
+
+        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}_ChainBosTau3Link")
+          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}_ChainBosTau3Link")
+          chr_klass.__send__(:find_all_by_interval, interval, opt)
+        end
+      end # class
+
+    end # module
+  end # module Ucsc
+end # module Bio

data/lib/bio-ucsc/rn4/chaincanfam2.rb

@@ -0,0 +1,82 @@
+# = chaincanfam2.rb
+# Copyright::
+#  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
+# License::     The Ruby licence (Ryby's / GPLv2 dual)
+#
+# In the hg18 database, this table is actually separated
+# into "chr1_*", "chr2_*", etc. This class dynamically
+# define *::Chr1_*, *::Chr2_*, etc. The
+# Rmsk.find_by_interval calls an appropreate class automatically.
+
+module Bio
+  module Ucsc
+    module Rn4
+
+      class ChainCanFam2
+        KLASS = "ChainCanFam2"
+        KLASS_S = "chainCanFam2"
+
+         Bio::Ucsc::Rn4::CHROMS.each do |chr|
+          class_eval %!
+            class #{chr[0..0].upcase + chr[1..-1]}_#{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
+
+        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
+      end # class
+
+    end # module Hg18 
+  end # module Ucsc
+end # module Bio

data/lib/bio-ucsc/rn4/chaincanfam2link.rb

@@ -0,0 +1,82 @@
+# = chainbraflo1link.rb
+# Copyright::
+#  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
+# License::     The Ruby licence (Ryby's / GPLv2 dual)
+#
+# In the hg18 database, this table is actually separated
+# into "chr1_*", "chr2_*", etc. This class dynamically
+# define *::Chr1_*, *::Chr2_*, etc. The
+# Rmsk.find_by_interval calls an appropreate class automatically.
+
+module Bio
+  module Ucsc
+    module Rn4
+
+      class ChainCanFam2Link
+        KLASS = "ChainCanFam2Link"
+        KLASS_S = "chainCanFam2Link"
+
+         Bio::Ucsc::Rn4::CHROMS.each do |chr|
+          class_eval %!
+            class #{chr[0..0].upcase + chr[1..-1]}_#{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
+
+        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
+      end # class
+
+    end # module Hg18 
+  end # module Ucsc
+end # module Bio

data/lib/bio-ucsc/rn4/chaincavpor3.rb

@@ -0,0 +1,81 @@
+# Copyright::
+#  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
+# License::     The Ruby licence (Ryby's / GPLv2 dual)
+#
+# In the hg18 database, this table is actually separated
+# into "chr1_*", "chr2_*", etc. This class dynamically
+# define *::Chr1_*, *::Chr2_*, etc. The
+# Rmsk.find_by_interval calls an appropreate class automatically.
+
+module Bio
+  module Ucsc
+    module Rn4
+
+      class ChainCavPor3
+        KLASS = "ChainCavPor3"
+        KLASS_S = "chainCavPor3"
+
+        Bio::Ucsc::Rn4::CHROMS.each do |chr|
+          class_eval %!
+            class #{chr[0..0].upcase + chr[1..-1]}_#{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
+
+        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
+      end # class
+
+    end # module Hg18 
+  end # module Ucsc
+end # module Bio