bio-ucsc-api 0.2.0 → 0.2.1

data/lib/bio-ucsc/caerem3/gap.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 CaeRem3
+
+      class Gap
+        KLASS = "Gap"
+        KLASS_S = "gap"
+
+        Bio::Ucsc::CaeRem3::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
+      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
+
+        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/caerem3/gold.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)
+#
+# his 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 CaeRem3
+
+      class Gold
+        KLASS = "Gold"
+        KLASS_S = "gold"
+
+        Bio::Ucsc::CaeRem3::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
+      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
+
+        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/caljac3.rb

@@ -0,0 +1,24 @@
+#
+# = AUTOMATIC Table Definition of the Marmoset March 2009 (WUGSC 3.2/calJac3) 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__)}/caljac3/db_connection"
+require "#{File.dirname(__FILE__)}/table_class_detector"
+
+module Bio
+  module Ucsc
+    module CalJac3
+      CHROMS =
+%w(Chr1 Chr2 Chr3 Chr4 Chr5 Chr6 Chr7 Chr8 Chr9
+Chr10 Chr11 Chr12 Chr13 Chr14 Chr15 Chr16 Chr17 Chr18 Chr19
+Chr20 Chr21 Chr22 ChrX ChrY)
+      extend TableClassDetector
+
+      base = "#{File.dirname(__FILE__)}/caljac3"
+    end
+  end
+end

data/lib/bio-ucsc/caljac3/db_connection.rb

@@ -0,0 +1,55 @@
+#
+# = db_connection.rb
+#
+# 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 CalJac3
+      # = 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 => 'calJac3',
+                                 :username => @@db_username,
+                                 :password => @@db_password, })
+        end
+      end
+    end # module
+    
+  end # module Ucsc
+end # module Bio

data/lib/bio-ucsc/canfam2.rb

@@ -0,0 +1,39 @@
+#
+# = AUTOMATIC Table Definition of the Dog May 2005 (Broad/canFam2) 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__)}/canfam2/db_connection"
+require "#{File.dirname(__FILE__)}/table_class_detector"
+
+module Bio
+  module Ucsc
+    module CanFam2
+      CHROMS =
+%w(ChrM Chr1 Chr2 Chr3 Chr4 Chr5 Chr6 Chr7 Chr8 Chr9
+Chr10 Chr11 Chr12 Chr13 Chr14 Chr15 Chr16 Chr17 Chr18 Chr19
+Chr20 Chr21 Chr22 Chr23 Chr24 Chr25 Chr26 Chr27 Chr28 Chr29
+Chr30 Chr31 Chr32 Chr33 Chr34 Chr35 Chr36 Chr37 Chr38 ChrX)
+      extend TableClassDetector
+
+      base = "#{File.dirname(__FILE__)}/canfam2"
+      autoload :ChainBosTau4, "#{base}/chainbostau4"
+      autoload :ChainBosTau4Link, "#{base}/chainbostau4link"
+      autoload :ChainMm9, "#{base}/chainmm9"
+      autoload :ChainMm9Link, "#{base}/chainmm9link"
+      autoload :ChainRn4, "#{base}/chainrn4"
+      autoload :ChainRn4Link, "#{base}/chainrn4link"
+      autoload :ChainSelf, "#{base}/chainself"
+      autoload :ChainSelfLink, "#{base}/chainselflink"
+      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/canfam2/chainbostau4.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 CanFam2
+
+      class ChainBosTau4
+        KLASS = "ChainBosTau4"
+        KLASS_S = KLASS[0..0].downcase + KLASS[1..-1]
+
+        Bio::Ucsc::CanFam2::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/canfam2/chainbostau4link.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 CanFam2
+
+      class ChainBosTau4Link
+        KLASS = "ChainBosTau4Link"
+        KLASS_S = KLASS[0..0].downcase + KLASS[1..-1]
+
+        Bio::Ucsc::CanFam2::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