Prototype for getting the species prefixes for Ensembl identifiers for all species, using the production database

ensembl_production_prototype.rb

# This script is released under a GPL version 3 license or newer. Copyright by the author Ben Woodcroft.

require 'rubygems'
require 'ensembl'
require 'pp'

module Ensembl
  module Production
    class DBConnection < Ensembl::DBRegistry::Base
      self.abstract_class = true
      self.pluralize_table_names = false

      def self.connect(release = Ensembl::ENSEMBL_RELEASE, args = {})
        database = "ensembl_production_#{release}"
        self.generic_connect(nil,nil, release,{:database => database})
      end
    end # Core::DBConnection

    class Species < DBConnection
      set_primary_key 'species_id'
    end
  end # Production

  class SpeciesPrefixExtractor
    def initialize
      @@species_info = nil
    end
    
    def [](db_name)
      Ensembl::SpeciesPrefixExtractor.cache
      @@species_info[db_name]
    end

    def infos
      Ensembl::SpeciesPrefixExtractor.cache
      @@species_info
    end

    def self.cache
      return nil unless @@species_info.nil?
      
      Ensembl::Production::DBConnection.connect
      # A hash of species db_names to SpeciesInformation objects
      @@species_info = {}

      # Get the list of species, db_name and common_names for each of species
      Ensembl::Production::Species.all(:limit => 2).each do |sp|
        info = SpeciesInformation.new
        info.db_name = sp.db_name
        info.common_name = sp.common_name
        @@species_info[sp.db_name] = info
      end

      # Get the prefix information, which are in each of the individual databases
      @@species_info.each do |db_name, info|
        $stderr.puts db_name
        Ensembl::Core::DBConnection.connect(db_name)
        prefix = Ensembl::Core::Meta.first(:conditions => ['meta_key = ?','species.stable_id_prefix'])
        @@species_info[db_name].prefix = prefix.meta_value unless prefix.nil?
      end
      nil
    end

    class SpeciesInformation
      attr_accessor :common_name, :db_name, :prefix
    end

  end

end # Ensembl

ex = Ensembl::SpeciesPrefixExtractor.new
hash = {}
ex.infos.collect do |db_name, info|
  unless info.prefix.nil?
    hash[info.prefix] = [info.common_name, db_name]
  end
end

pp hash