bio-ucsc-api 0.0.5 → 0.1.0

data/README.rdoc

@@ -1,37 +1,40 @@
 = bio-ucsc-api
 
-A BioRuby plugin: Ruby API for the UCSC database.
+The Ruby UCSC API: accessing the UCSC Genome Database using Ruby.
 
 This package is based on the followings:
-* ruby-ucsc-api: https://github.com/jandot/ruby-ucsc-api
+* original ruby-ucsc-api: https://github.com/jandot/ruby-ucsc-api
 * ruby-ensembl-api: https://github.com/jandot/ruby-ensembl-api
 
 Major dependent gems:
 * active_record http://api.rubyonrails.org/classes/ActiveRecord/Base.html
 * bioruby-genomic-interval https://github.com/misshie/bioruby-genomic-interval
 
+Requirement:
+* Ruby version 1.9.2 or later
+
 See also a paper:
 
 Strozzi F, Aerts J: A Ruby API to query the Ensembl database for genomic features.
 Bioinformatics 2011, 27:1013-1014.
 
-NOTE: THIS PACKAGE IS STILL *EXPERIMENTAL* VERSION. 
-Your comments, suggestions and requests are welcome.
-Email to the author (missy at be.to) or create an "Issue" at
-http://github.com/misshie/bioruby-ucsc-api/issues
+Your comments, suggestions and requests are welcome. Documentation and
+feedback are available at the UserEcho site at
+http://rubyucscapi.userecho.com/.
 
 == Features
 
 * Designed as a BioRuby plugin
 * Using ActiveRecord as an O/R mapping framework
 * Using the Bin index system to improve query performance (see https://github.com/misshie/UCSCBin )
-* Automatic conversion of "1-based full-closed intervals" to internal "0-based half-closed intervals" (see also bioruby-genomic-interval)
+* Automatic conversion of "1-based full-closed intervals" to internal "0-based left-closed right-open intervals" (see also bioruby-genomic-interval)
 * Supporting non-official MySql hosts (e.g. local servers)
 * Using Rspec for the testing framework
-* Support hg19 and hg18 databases. So far, tables containing only single records of the fileName columns are not supported.
+* Support hg19 and hg18 databases. Current version does not support tables containing only single records of the fileName columns linking to bigWIG or bigBED files.
 
 == Change Log
-
+* *MODIFIED* (v0.1.0): The name of this library is now "Ruby UCSC API". The RubyGem name and the GitHub account and the library name are not changed.
+*  *MODIFIED* (v0.1.0): Bio::Ucsc::[Hg18|Hg19]::Reference is replaced by Bio::Ucsc::[Hg18|Hg19]::ReferenceSequence.
 * *UPDATE* (v0.0.5): Almost all hg18 tables are supported.
 * *UPDATE* (v0.0.5): find_by_interval and find_all_by_interval class methods accept the "partial" option. Default is true. When "partial: false" is opted, return value will be only fully-included (non-partially-included) records.  
 * *UPDATE* (v0.0.4): Almost all hg19 tables are supported. "filename" tables in ENCODE dataset are omitted. Each of them contains only single record of a path to the raw data file. Definitions of table relations are incomplete. 
@@ -49,13 +52,13 @@ You may need to be root or use "sudo". "--no-ri" and "--no-rdoc" options are rec
 
  require 'bio-ucsc'
 
- include Bio::Ucsc::Hg19
+ include Bio::Ucsc::Hg19 
  DBConnection.connect
 
  itv =  Bio::GenomicInterval.parse("chr1:1-11,000")
  Snp131.find_all_by_interval(itv).each do |e|
    i = Bio::GenomicInterval.zero_based(e.chrom, e.chromStart, e.chromEnd)
-   puts "#{i.chrom}\t#{i.chromStart}\t#{e.name}\t#{e[:class]}"
+   puts "#{i.chrom}\t#{i.chr_start}\t#{e.name}\t#{e[:class]}"
  end
 
  itv =  Bio::GenomicInterval.parse("chr17:7,579,614-7,579,700")
@@ -64,15 +67,32 @@ You may need to be root or use "sudo". "--no-ri" and "--no-rdoc" options are rec
 
  p Snp131.find_by_name("rs56289060")
 
+ # Sometimes, queries using raw SQL provide elegant solutions.
+ # 
+ sql << 'SQL'
+ SELECT name,chrom,chromStart,chromEnd,observed
+ FROM snp131 
+ WHERE name="rs56289060"
+ SQL
+ p Snp131.find_by_sql(sql)
+
  # retrieve reference sequence from a locally-stored 2bit file
- Reference.load("hg19.2bit")
+ ReferenceSequence.load("hg19.2bit")
  itv = Bio::GenomicInterval.parse("chr1:9,500-10,999")
- p Reference.find_by_interval(itv)
+ p ReferenceSequence.find_by_interval(itv)
 
-== Copyright
-Copyright:: (c) 2011 MISHIMA, Hiroyuki (missy at be.to / hmishima at nagasaki-u.ac.jp)
+ # Connetcting to non-official or local mirror MySQL servers
+ DBConnection.db_host = 'foo.example.edu'
+ DBConnection.db_username = 'genome'
+ DBConnection.db_password = ''
+ DBConnection.connect
+
+ DBConnection.default # reset to connect UCSC's public MySQL sever
+ DBConnection.connect
 
-Copyright:: (c) 2008-2010 Francesco Strozzi, Jan Aerts
+== Copyright
+Copyright:: (c) 2011 MISHIMA, Hiroyuki (missy at be.to / hmishima at nagasaki-u.ac.jp / @mishimahryk at Twitter)
+Copyright:: (c) 2010 Jan Aerts
 
 License:: Ruby license (Ruby's / GPLv2 dual). See COPYING and COPYING.ja for further details.
 

data/Rakefile

@@ -16,10 +16,10 @@ Jeweler::Tasks.new do |gem|
   gem.name = "bio-ucsc-api"
   gem.homepage = "http://github.com/misshie/bioruby-ucsc-api"
   gem.license = "Ruby (Ruby's/GPLv2 dual)"
-  gem.summary = %Q{A Bioruby plugin: an API for UCSC Genome Browser (experimental)}
-  gem.description = %Q{A Bioruby plugin: an API for UCSC Genome Browser database (experimental)}
+  gem.summary = %Q{The Ruby UCSC API: accessing the UCSC Genome Database using Ruby}
+  gem.description = %Q{Ruby UCSC API: accessing the UCSC Genome Database using Ruby}
   gem.email = "missy@be.to"
-  gem.authors = ["Hiroyuki Mishima", "Francesco Strozzi", "Jan Aerts"]
+  gem.authors = ["Hiroyuki Mishima", "Jan Aerts"]
   # Include your dependencies below. Runtime dependencies are required when using your gem,
   # and development dependencies are only needed for development
   # (ie running rake tasks, tests, etc)

data/VERSION

@@ -1 +1 @@
-0.0.5
+0.1.0

data/bio-ucsc-api.gemspec

@@ -5,12 +5,12 @@
 
 Gem::Specification.new do |s|
   s.name = %q{bio-ucsc-api}
-  s.version = "0.0.5"
+  s.version = "0.1.0"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
-  s.authors = ["Hiroyuki Mishima", "Francesco Strozzi", "Jan Aerts"]
-  s.date = %q{2011-06-19}
-  s.description = %q{A Bioruby plugin: an API for UCSC Genome Browser database (experimental)}
+  s.authors = ["Hiroyuki Mishima", "Jan Aerts"]
+  s.date = %q{2011-07-07}
+  s.description = %q{Ruby UCSC API: accessing the UCSC Genome Database using Ruby}
   s.email = %q{missy@be.to}
   s.extra_rdoc_files = [
     "README.rdoc"
@@ -1009,6 +1009,7 @@ Gem::Specification.new do |s|
     "lib/bio-ucsc/hg18/rdmrraw.rb",
     "lib/bio-ucsc/hg18/recombrate.rb",
     "lib/bio-ucsc/hg18/reference.rb",
+    "lib/bio-ucsc/hg18/reference_sequence.rb",
     "lib/bio-ucsc/hg18/refflat.rb",
     "lib/bio-ucsc/hg18/refgene.rb",
     "lib/bio-ucsc/hg18/reflink.rb",
@@ -3651,6 +3652,7 @@ Gem::Specification.new do |s|
     "lib/bio-ucsc/hg19/polyapredict.rb",
     "lib/bio-ucsc/hg19/recombrate.rb",
     "lib/bio-ucsc/hg19/reference.rb",
+    "lib/bio-ucsc/hg19/reference_sequence.rb",
     "lib/bio-ucsc/hg19/refflat.rb",
     "lib/bio-ucsc/hg19/refgene.rb",
     "lib/bio-ucsc/hg19/reflink.rb",
@@ -4939,7 +4941,7 @@ Gem::Specification.new do |s|
   s.licenses = ["Ruby (Ruby's/GPLv2 dual)"]
   s.require_paths = ["lib"]
   s.rubygems_version = %q{1.7.2}
-  s.summary = %q{A Bioruby plugin: an API for UCSC Genome Browser (experimental)}
+  s.summary = %q{The Ruby UCSC API: accessing the UCSC Genome Database using Ruby}
   s.test_files = [
     "spec/hg18/acembly_spec.rb",
     "spec/hg18/acemblyclass_spec.rb",
@@ -5921,7 +5923,7 @@ Gem::Specification.new do |s|
     "spec/hg18/rdmr_spec.rb",
     "spec/hg18/rdmrraw_spec.rb",
     "spec/hg18/recombrate_spec.rb",
-    "spec/hg18/reference_spec.rb",
+    "spec/hg18/reference_sequence_spec.rb",
     "spec/hg18/refflat_spec.rb",
     "spec/hg18/refgene_spec.rb",
     "spec/hg18/reflink_spec.rb",
@@ -8560,7 +8562,7 @@ Gem::Specification.new do |s|
     "spec/hg19/polyadb_spec.rb",
     "spec/hg19/polyapredict_spec.rb",
     "spec/hg19/recombrate_spec.rb",
-    "spec/hg19/reference_spec.rb",
+    "spec/hg19/reference_sequence_spec.rb",
     "spec/hg19/refflat_spec.rb",
     "spec/hg19/refgene_spec.rb",
     "spec/hg19/reflink_spec.rb",

data/lib/bio-ucsc/hg18.rb

@@ -2,8 +2,6 @@
 # = hg18/ucsc.rb
 # Copyright::   Cioyrught (C) 2011
 #               MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-# Copyright::   Copyright (C) 2008
-#               Jan Aerts <jan.aerts@gmail.com>
 # License::     Ruby's
 
 base = "#{File.dirname(__FILE__)}/hg18"
@@ -15,7 +13,8 @@ module Bio
     module Hg18
       base = "#{File.dirname(__FILE__)}/hg18"
       # Reference sequence retrieval via the 2bit file
-      autoload :Reference,    "#{base}/reference"
+      autoload :Reference, "#{base}/reference" # OBSOLETE
+      autoload :ReferenceSequence, "#{base}/reference_sequence"
 
       # group: Mapping and Sequencing Tracks ----------
       #
@@ -3867,4 +3866,3 @@ module Bio
     end
   end
 end
-

data/lib/bio-ucsc/hg18/activerecord.rb

@@ -3,7 +3,7 @@
 # 
 # Copyright::
 #   Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp>
-#   Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
+#   Copyright (C) 2010 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = DESCRIPTION

data/lib/bio-ucsc/hg18/all_bacends.rb

@@ -2,7 +2,6 @@
 # = hg18/cnpiafrate2.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser

data/lib/bio-ucsc/hg18/cnpiafrate2.rb

@@ -2,7 +2,6 @@
 # = hg18/cnpiafrate2.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser

data/lib/bio-ucsc/hg18/cnplocke.rb

@@ -2,7 +2,6 @@
 # = hg18/cnplocke.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser

data/lib/bio-ucsc/hg18/cnpredon.rb

@@ -2,7 +2,6 @@
 # = hg18/cnpredon.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser

data/lib/bio-ucsc/hg18/cnpsebat2.rb

@@ -2,7 +2,6 @@
 # = hg18/cnpsebat2.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser

data/lib/bio-ucsc/hg18/cnpsharp2.rb

@@ -2,7 +2,6 @@
 # = hg18/cnpsharp2.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser

data/lib/bio-ucsc/hg18/cnptuzun.rb

@@ -2,7 +2,6 @@
 # = hg18/cnptuzun.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser

data/lib/bio-ucsc/hg18/cytoband.rb

@@ -2,7 +2,6 @@
 # = hg18/wikitrack.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser The Wiki track displays

data/lib/bio-ucsc/hg18/db_connection.rb

@@ -3,7 +3,7 @@
 #
 # Copyright::   Cioyrught (C) 2011
 #               MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-# Copyright::   Copyright (C) 2008
+# Copyright::   Copyright (C) 2010
 #               Jan Aerts <jan.aerts@gmail.com>
 # License::     Ruby licence (Ryby's / GPLv2 dual)
 

data/lib/bio-ucsc/hg18/delconrad2.rb

@@ -2,7 +2,6 @@
 # = hg18/DelConrad2.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser

data/lib/bio-ucsc/hg18/delhinds2.rb

@@ -2,7 +2,6 @@
 # = hg18/DelHinds2.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser

data/lib/bio-ucsc/hg18/delmccarroll.rb

@@ -2,7 +2,6 @@
 # = hg18/DelMccarroll.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser

data/lib/bio-ucsc/hg18/dgv.rb

@@ -2,7 +2,6 @@
 # = hg18/dgv.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser

data/lib/bio-ucsc/hg18/ensgene.rb

@@ -2,7 +2,6 @@
 # = hg18/ensgene.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser

data/lib/bio-ucsc/hg18/exaptedrepeats.rb

@@ -2,7 +2,6 @@
 # = hg18/exaptedrepeats.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser

data/lib/bio-ucsc/hg18/hgcentral_wikitrack.rb

@@ -2,7 +2,6 @@
 # = hg18/wikitrack.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser The Wiki track displays

data/lib/bio-ucsc/hg18/kgprotmap2.rb

@@ -2,7 +2,6 @@
 # = hg18/kgprotmap2.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser

data/lib/bio-ucsc/hg18/kgtargetali.rb

@@ -2,7 +2,6 @@
 # = hg18/kgtargetali.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser

data/lib/bio-ucsc/hg18/kiddeichlerdiscabc10.rb

@@ -2,7 +2,6 @@
 # = hg18/kiddeichlerdiscabc10.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser

data/lib/bio-ucsc/hg18/kiddeichlerdiscabc11.rb

@@ -2,7 +2,6 @@
 # = hg18/kiddeichlerdiscabc11.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser

data/lib/bio-ucsc/hg18/kiddeichlerdiscabc12.rb

@@ -2,7 +2,6 @@
 # = hg18/kiddeichlerdiscabc12.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser

data/lib/bio-ucsc/hg18/kiddeichlerdiscabc13.rb

@@ -2,7 +2,6 @@
 # = hg18/kiddeichlerdiscabc13.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser

data/lib/bio-ucsc/hg18/kiddeichlerdiscabc14.rb

@@ -2,7 +2,6 @@
 # = hg18/kiddeichlerdiscabc14.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser

data/lib/bio-ucsc/hg18/kiddeichlerdiscabc8.rb

@@ -2,7 +2,6 @@
 # = hg18/kiddeichlerdiscabc8.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser

data/lib/bio-ucsc/hg18/kiddeichlerdiscabc9.rb

@@ -2,7 +2,6 @@
 # = hg18/kiddeichlerdiscabc9.rb
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser

data/lib/bio-ucsc/hg18/kiddeichlerdiscg248.rb

@@ -2,7 +2,6 @@
 # = hg18/kiddeichlerdiscg248
 # Copyright::
 #  Copyright (C) 2011 MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
-#  Copyright (C) 2008 Jan Aerts <jan.aerts@gmail.com>
 # License::     The Ruby licence (Ryby's / GPLv2 dual)
 #
 # = Table desfription in UCSC Table Browser

data/lib/bio-ucsc/hg18/reference.rb

@@ -2,170 +2,11 @@
 # = reference.rb
 # handle UCSC's 2bit file (locally stored) to retrieve the reference sequence
 #
+# In version 0.1.0, this file is OBSOLETE.
+# Use Ucsc::Hg18::ReferenceSequence instead.
+#
 # Copyright::   Cioyrught (C) 2011
 #               MISHIMA, Hiroyuki <missy at be.to / hmishima at nagasaki-u.ac.jp> 
 # License::     Ruby license (Ryby's / GPLv2 dual)
 
-# require 'bio'
-
-module Bio
-  module Ucsc
-    module Hg18
-      TwoBitHeader = 
-        Struct.new(:signature, :version, :sequence_count, :reserved)
-      TwoBitRecord =
-        Struct.new(:dna_size,
-                   :n_block_intervals, :mask_block_intervals,
-                   :reserved, :packed_dna_offset)
-
-      class ByteQueue
-        def initialize(str)
-          @str = str
-          @index = 0
-        end
-
-        attr_accessor :index
-
-        def next(n)
-          result = @str[@index, n]
-          @index += n
-          result
-        end
-      end # class ByteQueue
-
-      class Reference
-        BINCODE = {0b00 => "T", 0b01 => "C", 0b10 => "A", 0b11 => "G"}
-
-        cattr_reader :filename, :header, :offsets
-
-        def self.load(filename)
-          two_bit = nil
-          open(filename, 'rb') {|f| two_bit = f.read}
-          @@tbq = ByteQueue.new(two_bit)
-          @@filename = filename
-
-          twobit_header = TwoBitHeader.new
-          twobit_header.signature      = @@tbq.next(4).unpack('L').first
-          twobit_header.version        = @@tbq.next(4).unpack('L').first
-          twobit_header.sequence_count = @@tbq.next(4).unpack('L').first
-          twobit_header.reserved       = @@tbq.next(4).unpack('L').first
-          @@header = twobit_header
-
-          @@offsets = Hash.new
-          @@header.sequence_count.times do
-            name_length = @@tbq.next(1).unpack('C').first
-            @@offsets[@@tbq.next(name_length).unpack('a*').first] =
-              @@tbq.next(4).unpack('L').first
-          end
-          @@records = Hash.new
-        end
-
-        def self.records(chrom)
-          return @@records[chrom] if @@records[chrom]
-          
-          @@tbq.index = @@offsets[chrom]
-          @@records[chrom] = TwoBitRecord.new
-          @@records[chrom].dna_size = @@tbq.next(4).unpack('L').first
- 
-          n_block_count = @@tbq.next(4).unpack('L').first
-          n_block_starts = Array.new
-          n_block_count.times do
-            n_block_starts << @@tbq.next(4).unpack('L').first
-          end
-          n_block_sizes = Array.new
-          n_block_count.times do
-            n_block_sizes << @@tbq.next(4).unpack('L').first
-          end
-          @@records[chrom].n_block_intervals = Array.new
-          n_block_count.times do |idx|
-            @@records[chrom].n_block_intervals << 
-              Bio::GenomicInterval.zero_based(chrom,
-                                             n_block_starts[idx],
-                                             n_block_starts[idx]+n_block_sizes[idx])
-          end
-          
-          mask_block_count = @@tbq.next(4).unpack('L').first
-          mask_block_starts = Array.new
-          mask_block_count.times do
-            mask_block_starts << @@tbq.next(4).unpack('L').first
-          end
-          mask_block_sizes = Array.new
-          mask_block_count.times do
-            mask_block_sizes << @@tbq.next(4).unpack('L').first
-          end
-          @@records[chrom].mask_block_intervals = Array.new
-          mask_block_count.times do |idx|
-            @@records[chrom].mask_block_intervals << 
-              Bio::GenomicInterval.zero_based(chrom,
-                                             mask_block_starts[idx],
-                                             mask_block_starts[idx]+mask_block_sizes[idx])
-          end
-  
-          @@records[chrom].reserved = @@tbq.next(4).unpack('L').first
-          @@records[chrom].packed_dna_offset = @@tbq.index
-
-          @@records[chrom]
-        end
-
-        def self.find_by_interval(interval)
-          seq = self.find_by_interval_raw(interval)
-          @@records[interval.chrom].n_block_intervals.map do |nb|
-            if interval.overlapped?(nb)
-              case interval.compare(nb)
-              when :equal,:contained_by
-                seq = 'N' * interval.overlap(nb)
-              when :contains
-                left_len  = nb.chr_start - interval.chr_start + 1
-                right_len = interval.chr_end - nb.chr_end + 1
-                seq[0, left_len] = 'N' * left_len
-                seq[-right_len, right_len] = 'N' * right_len
-              when :left_overlapped
-                left_len = nb.chr_end - interval.chr_start + 1
-                seq[0, left_len] = 'N' * left_len
-              when :right_overlapped
-                right_len = interval.chr_end - nb.chr_start + 1
-                seq[-right_len, right_len] = 'N' * right_len
-              when :right_adjacent, :right_off
-                # expecting that N-blocks are sorted
-                # return Bio::Sequence::NA.new(seq) 
-                seq
-              end
-            end
-          end
-          #Bio::Sequence::NA.new(seq)
-          seq
-        end
-
-        def self.find_by_interval_raw(interval)
-          byte_count, byte_mod = interval.zero_start.divmod 4
-          chrom_top = self.records(interval.chrom).packed_dna_offset
-          div_start, mod_start = interval.zero_start.divmod 4
-          div_end, mod_end     = interval.zero_end.divmod 4
-          div_len, mod_len     = interval.length.divmod 4
-
-          byte_length = div_end - div_start + 1
-          @@tbq.index = chrom_top + div_start
-          bytes = @@tbq.next(byte_length).unpack('C*')
-          seq = Bio::Ucsc::Hg18::Reference.bytes_to_nucleotides(bytes)
-          seq[mod_start..(-1-(4-mod_end))]
-        end
-
-        def self.bytes_to_nucleotides(bytes) 
-          results = ""
-          bytes.each do |byte|
-            results << Bio::Ucsc::Hg18::Reference.byte_to_nucleotides(byte)
-          end
-          results
-        end
-
-        def self.byte_to_nucleotides(byte)
-          BINCODE[byte >> 6] +
-            BINCODE[(byte >> 4) & 0b11] +
-            BINCODE[(byte >> 2) & 0b11] +
-            BINCODE[byte & 0b11]
-        end
-      end # class Reference
-
-    end # module Hg18
-  end # module Ucsc
-end # module Bio
+raise "Bio::Ucsc::Hg18::Reference is OBSOLETE. Use Bio::Ucsc::Hg18::ReferenceSequence instead."