bio-ucsc-api 0.2.0 → 0.2.1

data/lib/bio-ucsc/anogam1/mrna.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 AnoGam1
+
+      class Mrna
+        KLASS = "Mrna"
+        KLASS_S = "mrna"
+
+        Bio::Ucsc::AnoGam1::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/anogam1/rmsk.rb

@@ -0,0 +1,78 @@
+# = rmsk.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, the Rmsk table is actually separated
+# into "chr1_rmsk", "chr2_rmsk", etc. The Rmsk class dynamically
+# define Rmsk::Chr1_Rmsk, Rmsk::Chr2_Rmsk, etc. The
+# Rmsk.find_by_interval calls an appropreate class automatically.
+
+module Bio
+  module Ucsc
+    module AnoGam1
+
+      class Rmsk
+        Bio::Ucsc::AnoGam1::CHROMS.each do |chr|
+          class_eval %!
+            class #{chr[0..0].upcase + chr[1..-1]}_Rmsk < DBConnection
+              set_table_name "#{chr[0..0].downcase + chr[1..-1]}_rmsk"
+              set_primary_key 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
+      genoName = :chrom
+AND   bin in (:bins)
+AND ((genoStart BETWEEN :zstart AND :zend)
+ OR  (genoEnd BETWEEN :zstart AND :zend)
+ OR  (genoStart <= :zstart AND genoEnd >= :zend))
+                  SQL
+                else
+                  where = <<-SQL
+      genoName = :chrom
+AND   bin in (:bins)
+AND ((genoStart BETWEEN :zstart AND :zend)
+AND  (genoEnd 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}_Rmsk")
+          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}_Rmsk")
+          chr_klass.__send__(:find_all_by_interval, interval, opt)
+        end
+      end # class Rmsk
+
+    end # module 
+  end # module Ucsc
+end # module Bio

data/lib/bio-ucsc/apimel2.rb

@@ -0,0 +1,26 @@
+#
+# = AUTOMATIC Table Definition of the A. mellifera Jan. 2005 (Baylor 2.0/apiMel2) 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__)}/apimel2/db_connection"
+require "#{File.dirname(__FILE__)}/table_class_detector"
+
+module Bio
+  module Ucsc
+    module ApiMel2
+      CHROMS =
+%w(Group1 Group2 Group3 Group4 Group5 Group6 Group7 Group8 Group9 Group10
+Group11 Group12 Group13 Group14 Group15 Group16 GroupUn)
+
+      extend TableClassDetector
+
+      base = "#{File.dirname(__FILE__)}/apimel2"
+      autoload :ChainDm2, "#{base}/chaindm2"
+      autoload :ChainDm2Link, "#{base}/chaindm2link"
+    end
+  end
+end

data/lib/bio-ucsc/apimel2/chaindm2.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 ApiMel2
+
+      class ChainDm2
+        KLASS = "ChainDm2"
+        KLASS_S = KLASS[0..0].downcase + KLASS[1..-1]
+
+        Bio::Ucsc::ApiMel2::CHROMS.each do |chr|
+          class_eval %!
+            class #{chr[0..0].upcase + chr[1..-1]}_#{KLASS} < DBConnection
+              set_table_name "#{chr}_#{KLASS_S}" # Group1 => Group1_chainDm2
+              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 
+  end # module Ucsc
+end # module Bio

data/lib/bio-ucsc/apimel2/chaindm2link.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 ApiMel2
+
+      class ChainDm2Link
+        KLASS = "ChainDm2Link"
+        KLASS_S = KLASS[0..0].downcase + KLASS[1..-1]
+
+        Bio::Ucsc::ApiMel2::CHROMS.each do |chr|
+          class_eval %!
+            class #{chr[0..0].upcase + chr[1..-1]}_#{KLASS} < DBConnection
+              set_table_name "#{chr}_#{KLASS_S}" # Group1 => Group1_chainDm2
+              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 
+  end # module Ucsc
+end # module Bio

data/lib/bio-ucsc/apimel2/db_connection.rb

@@ -0,0 +1,57 @@
+#
+# = 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 ApiMel2
+      DATABASE_NAME = "apiMel2"
+
+      # = 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 => DATABASE_NAME,
+                                 :username => @@db_username,
+                                 :password => @@db_password, })
+        end
+      end
+    end # module
+    
+  end # module Ucsc
+end # module Bio

data/lib/bio-ucsc/aplcal1.rb

@@ -0,0 +1,33 @@
+#
+# = AUTOMATIC Table Definition of the Sea hare Sept. 2008 (Broad 2.0/aplCal1) 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__)}/aplcal1/db_connection"
+require "#{File.dirname(__FILE__)}/table_class_detector"
+
+module Bio
+  module Ucsc
+    module AplCal1
+      CHROMS = %w() # many scaffolds
+
+      extend TableClassDetector
+
+      base = "#{File.dirname(__FILE__)}/aplcal1"
+      # autoload :ChainCaePb1, "#{base}/chaincaepb1"
+      # autoload :ChainCaePb1Link, "#{base}/chaincaepb1link"
+      # autoload :ChainCaeRem2, "#{base}/chaincaerem2"
+      # autoload :ChainCaeRem2Link, "#{base}/chaincaerem2link"
+      # autoload :ChainCb3, "#{base}/chaincb3"
+      # autoload :ChainCb3Link, "#{base}/chaincb3link"
+      # autoload :ChainCe6, "#{base}/chaince6"
+      # autoload :ChainCe6Link, "#{base}/chaince6link"
+      # autoload :Gap, "#{base}/gap"
+      # autoload :Gold, "#{base}/gold"
+      # autoload :Rmsk, "#{base}/rmsk"
+    end
+  end
+end

data/lib/bio-ucsc/aplcal1/chaincaepb1.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 PriPac1
+
+      class ChainCaePb1
+        KLASS = "ChainCaePb1"
+        KLASS_S = KLASS[0..0].downcase + KLASS[1..-1]
+
+        Bio::Ucsc::PriPac1::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