bio-ucsc-api 0.2.0 → 0.2.1

data/lib/bio-ucsc.rb

@@ -10,30 +10,184 @@ require "bio-genomic-interval"
 
 module Bio
   module Ucsc
-    VERSION = "0.2.0"
+    VERSION = "0.2.1"
     base = "#{File.dirname(__FILE__)}/bio-ucsc"
-
+    
+    # mammmals #####################################
     # human genome assemblies
-    autoload :Hg18, "#{base}/hg18"
     autoload :Hg19, "#{base}/hg19"
+    autoload :Hg18, "#{base}/hg18"
+
+    # chimp genome assemblies
+    autoload :PanTro3, "#{base}/pantro3"
+
+    # orangutan genome assemblies
+    autoload :PonAbe2, "#{base}/ponabe2"
+
+    # rhesus macaque genome assemblies
+    autoload :RheMac2, "#{base}/rhemac2"
+
+    # marmoset genome assemblies
+    autoload :CalJac3, "#{base}/caljac3"
 
     # mouse genome assemblies
     autoload :Mm9, "#{base}/mm9"
 
-    # fruitfly genome assemblies
+    # rat genome assemblies
+    autoload :Rn4, "#{base}/rn4"
+
+    # guinea pig genome assemblies
+    autoload :CavPor3, "#{base}/cavpor3"
+
+    # rabbit genome assemblies
+    autoload :OryCun2, "#{base}/orycun2"
+
+    # cat genome assemblies
+    autoload :FelCat4, "#{base}/felcat4"
+
+    # panda genome assemblies
+    autoload :AilMel1, "#{base}/ailmel1"
+    
+    # dog genome assemblies
+    autoload :CanFam2, "#{base}/canfam2"
+
+    # horse genome assemblies
+    autoload :EquCab2, "#{base}/equcab2"
+
+    # pig genome assemblies
+    autoload :SusScr2, "#{base}/susscr2"
+
+    # sheep genome assemblies
+    autoload :OviAri1, "#{base}/oviari1"
+
+    # cow genome assemblies
+    autoload :BosTau4, "#{base}/bostau4"
+
+    # elephant genome assemblies
+    autoload :LoxAfr3, "#{base}/loxafr3"
+
+    # opossum genome assemblies
+    autoload :MonDom5, "#{base}/mondom5"
+
+    # platypus genome assemblies
+    autoload :OrnAna1, "#{base}/ornana1"
+
+    # vertebrates #################################
+    # chiken genome assemblies
+    autoload :GalGal3, "#{base}/galgal3"
+
+    # zebra finch genome assemblies
+    autoload :TaeGut1, "#{base}/taegut1"
+
+    # lizard genome assemblies
+    autoload :AnoCar2, "#{base}/anocar2"
+
+    # X. tropicalis genome assemblies
+    autoload :XenTro2, "#{base}/xentro2"
+
+    # zebrafish genome assemblies
+    autoload :DanRer7, "#{base}/danrer7"
+
+    # tetraodon genome assemblies
+    autoload :TetNig2, "#{base}/tetnig2"
+
+    # fugu genome assemblies
+    autoload :Fr2, "#{base}/fr2"
+
+    # stickleback genome assemblies
+    autoload :GasAcu1, "#{base}/gasacu1"
+
+    # medaka genome assemblies
+    autoload :OryLat2, "#{base}/orylat2"
+
+    # lamprey genome assemblies
+    autoload :PetMar1, "#{base}/petmar1"
+
+    # deuterostomes ###############################
+    # lancelet genome assemblies
+    autoload :BraFlo1, "#{base}/braflo1.rb"
+
+    # sea squirt genome assemblies
+    autoload :Ci2, "#{base}/ci2"
+
+    # sea urchin genome assemblies
+    autoload :StrPur2, "#{base}/strpur2"
+
+    # insects  ####################################
+    # D. melanogaster genome assemblies
     autoload :Dm3, "#{base}/dm3" 
 
-    # C.Elegans genome assemblies
+    # D. silulans genome assemblies
+    autoload :DroSim1, "#{base}/drosim1"
+
+    # D. sechellia genome assemblies
+    autoload :DroSec1, "#{base}/drosec1"
+
+    # D. yakuba genome assemblies
+    autoload :DroYak2, "#{base}/droyak2"
+
+    # D. erecta genome assemblies
+    autoload :DroEre1, "#{base}/droere1"
+
+    # D. ananassae genome assemblies
+    autoload :DroAna2, "#{base}/droana2"
+
+    # D. pseudoobscura genome assemblies
+    autoload :Dp3, "#{base}/dp3"
+
+    # D. persimilis genome assemblies
+    autoload :DroPer1, "#{base}/droper1"
+
+    # D. virilis genome assemblies
+    autoload :DroVir2, "#{base}/drovir2"
+
+    # D. mojavensis genome assemblies
+    autoload :DroMoj2, "#{base}/dromoj2"
+
+    # D. grimshawi genome assemblies
+    autoload :DroGri1, "#{base}/drogri1"
+
+    # Anopheles mosquito genome assemblies
+    autoload :AnoGam1, "#{base}/anogam1"
+
+    # honey bee genome assemblies
+    autoload :ApiMel2, "#{base}/apimel2"
+
+    # nematodes ###################################
+    # C. elegans genome assemblies
     autoload :Ce6, "#{base}/ce6"
 
-    # accessing genome sequences in the 2bit format 
-    autoload :Reference, "#{base}/reference"
+    # C. brenneri genome assemblies
+    autoload :CaePb2, "#{base}/caepb2"
+
+    # C. briggsae genome assemblies
+    autoload :Cb3, "#{base}/cb3"
+
+    # C. remanei genome assemblies
+    autoload :CaeRem3, "#{base}/caerem3"
+
+    # C. japonica genome assemblies
+    autoload :CaeJap1, "#{base}/caejap1"
 
-    # genome assembly independent dababases
+    # P. pacificus genome assemblies
+    autoload :PriPac1, "#{base}/pripac1"
+
+    # others  #####################################
+    # sea hare genome assemblies
+    autoload :AplCal1, "#{base}/aplcal1"
+
+    # yeast genome assemblies
+    autoload :SacCer2, "#{base}/saccer2"
+
+    # genome assembly independent dababases #######
     autoload :UniProt, "#{base}/uniprot"
     autoload :Proteome,  "#{base}/proteome"
     autoload :Go, "#{base}/go"
     autoload :HgFixed, "#{base}/hgfixed"
     autoload :VisiGene, "#{base}/visigene"
+
+    # utility classes #############################
+    # accessing genome sequences in the 2bit format 
+    autoload :Reference, "#{base}/reference"
   end
 end

data/lib/bio-ucsc/ailmel1.rb

@@ -0,0 +1,21 @@
+#
+# = AUTOMATIC Table Definition of the Panda Dec. 2009 (BGI-Shenzhen 1.0/ailMel1) 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__)}/ailmel1/db_connection"
+require "#{File.dirname(__FILE__)}/table_class_detector"
+
+module Bio
+  module Ucsc
+    module AilMel1
+      CHROMS = %w() # many scaffolds
+      extend TableClassDetector
+
+      base = "#{File.dirname(__FILE__)}/ailmel1"
+    end
+  end
+end

data/lib/bio-ucsc/ailmel1/db_connection.rb

@@ -0,0 +1,57 @@
+#
+# = 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 AilMel1
+      DATABASE_NAME = "ailMel1"
+
+      # = 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/anocar2.rb

@@ -0,0 +1,22 @@
+#
+# = AUTOMATIC Table Definition of the Lizard May 2010 (Broad AnoCar2.0/anoCar2) 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__)}/anocar2/db_connection"
+require "#{File.dirname(__FILE__)}/table_class_detector"
+
+module Bio
+  module Ucsc
+    module AnoCar2
+      CHROMS = %w(chr1 chr2 chr3 chr4 chr5 chr6)
+
+      extend TableClassDetector
+
+      base = "#{File.dirname(__FILE__)}/anocar2"
+    end
+  end
+end

data/lib/bio-ucsc/anocar2/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 AnoCar2
+      DATABASE_NAME = "anoCar2"
+
+      # = 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/anogam1.rb

@@ -0,0 +1,30 @@
+#
+# = AUTOMATIC Table Definition of the A. gambiae Feb. 2003 (IAGEC MOZ2/anoGam1) 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__)}/anogam1/db_connection"
+require "#{File.dirname(__FILE__)}/table_class_detector"
+
+module Bio
+  module Ucsc
+    module AnoGam1
+      CHROMS = %w(chr2L chr2R chr3L chr3R chrX chrU chrM)
+
+      extend TableClassDetector
+
+      base = "#{File.dirname(__FILE__)}/anogam1"
+      autoload :ChainDm3, "#{base}/chaindm3"
+      autoload :ChainDm3Link, "#{base}/chaindm3link"
+      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/anogam1/chaindm3.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 ChainDm3
+        KLASS = "ChainDm3"
+        KLASS_S = KLASS[0..0].downcase + KLASS[1..-1]
+
+        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