ensembl 0.0.6 → 0.0.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: f643d5fd17097e73e1556a8e722410572d1f9a4c
-  data.tar.gz: c522eb0b8e1ea26ffa138237807ba8783bbf114a
+  metadata.gz: 16d2860f71a33c2c465b0fccf217377d71d583d9
+  data.tar.gz: bb8022c713d11e6fcc6c5baed00676bb3d9edd93
 SHA512:
-  metadata.gz: 93adb22a4770ee822b65ac3e183368f5462db87acdc691ce6f7ed717e75e2904bd0c403f2878cba669271f52a75f0b04eff5e53790dbda4b6f88bc2a86de13b9
-  data.tar.gz: 7927a1042a1cf6b7eed99c799551d5a39694efb912d89d348b40c809dc2ec3b38bb3ed06e45e874630ffd3addfd1eb14d281722cf3caab48b7cc2197f3051813
+  metadata.gz: cdf044fbd9e8c20534d035448afc3b2ee272ed229ebdca887e66f982f17881039b91e3ea8cc1186993893c45a130a9900f48e019046143807bf8d39dd47b3d95
+  data.tar.gz: 8f0aba8e3951d048296d4baab25e65e82d3c4244918856bd5890e43c9be2262d191a37b5e7c0d42e6ddc32659983ea34b28167a957021b04158d73b58e26bb1f

data/.gitignore

@@ -22,3 +22,5 @@ tmp
 mkmf.log
 
 .idea
+
+config/database.yml

data/Gemfile

@@ -2,6 +2,3 @@ source 'https://rubygems.org'
 
 # Specify your gem's dependencies in ensembl.gemspec
 gemspec
-
-gem 'pry', group: :pry
-gem 'pry-doc', group: :pry

data/README.md

@@ -20,25 +20,13 @@ Or install it yourself as:
 
 ## Usage
 
-    require 'ensembl'
+    Copy config/database.yml.example to database.yml and change parameters if needed 
 
-### ENSEMBL.ORG database - Default configuration
+### Using
     
     Ensembl::Variation::Variation.first
     
     Ensembl::Variation::Variation.first.source
-
-### Custom database
-    
-    # Set following values before using - Only Human databases is somewhat tested.
-    
-    Ensembl.host = 'myhost.example.com'
-    Ensembl.port = 3306                                 # default
-    Ensembl.username = 'anonymous'                      # default
-    Ensembl.password = ''                               # default
-    Ensembl.species = 'homo_sapiens'                    # default
-    Ensembl.version = 75                                # default
-    Ensembl.hg_version = 37                             # default
     
 
 ## Contributing

data/config/database.yml.example

@@ -0,0 +1,17 @@
+default: &default
+  adapter: mysql2
+  host: ensembldb.ensembl.org
+  port: 3306
+  username: anonymous
+  password: ''
+  reconnect: true
+  timeout: 1000
+  pool: 5
+
+core:
+  <<: *default
+  database: homo_sapiens_core_75_37
+
+variation:
+  <<: *default
+  database: homo_sapiens_variation_75_37

data/ensembl.gemspec

@@ -27,5 +27,5 @@ Gem::Specification.new do |spec|
   spec.add_development_dependency "bundler", "~> 1.6"
   spec.add_development_dependency "rake", '~> 10.3'
   spec.add_development_dependency "pry", '~> 0.10'
-  # spec.add_development_dependency "pry-byebug", '~> 1.3'
+
 end

data/lib/ensembl/core/activerecord.rb

@@ -12,13 +12,7 @@ module Ensembl
 
       self.abstract_class = true
 
-      self.establish_connection :adapter  => "mysql2",
-                                :host     => Ensembl.host,
-                                :username => Ensembl.username,
-                                :password => Ensembl.password,
-                                :database => Ensembl.species+'_core_'+Ensembl.version+'_'+Ensembl.hg_version,
-                                :reconnect => true
-
+      self.establish_connection :core
     end
 
     class ModelBase < Connection

data/lib/ensembl/helpers/like_search.rb

@@ -0,0 +1,22 @@
+module ActiveRecord
+  module LikeSearch
+    def table
+      self.arel_table
+    end
+
+    def starts_with(attribute, string)
+      where(table[attribute].matches("#{string}%"))
+    end
+
+    def ends_with(attribute,string)
+      where(table[attribute].matches("%#{string}"))
+    end
+
+    def contains(attribute,string)
+      where(table[attribute].matches("%#{string}%"))
+    end
+
+  end
+
+  Base.extend LikeSearch
+end

data/lib/ensembl/helpers/variation_position.rb

@@ -0,0 +1,30 @@
+module Ensembl
+  module Helpers
+    #TODO: Implement phased
+    class Genotype
+      def initialize(attrib)
+        @left=attrib[1]
+        @right=attrib[2]
+      end
+    end
+
+    class VariationPosition
+      attr_reader :chromosome,:start_pos,:end_pos
+
+      def initialize(row)
+        @chromosome,@start_pos,@end_pos,@strand=row
+      end
+
+      def strand
+        if @strand==1
+          return 'forward'
+        end
+        'reverse'
+      end
+
+      def to_s
+        'Chromosome: ' + @chromosome + ' ' + @start_pos.to_s + ':' + @end_pos.to_s
+      end
+    end
+  end
+end

data/lib/ensembl/variation/activerecord.rb

@@ -7,12 +7,7 @@ module Ensembl
 
       self.abstract_class = true
 
-      self.establish_connection :adapter  => "mysql2",
-                                :host     => Ensembl.host,
-                                :username => Ensembl.username,
-                                :password => Ensembl.password,
-                                :database => Ensembl.species+'_variation_'+Ensembl.version+'_'+Ensembl.hg_version,
-                                :reconnect => true
+      self.establish_connection :variation
 
     end
 
@@ -42,7 +37,7 @@ module Ensembl
     end
 
     class Attrib < ModelBase
-      self.extend SearchByAttribute
+      # self.extend AttributeLike
 
       belongs_to :attrib_type
     end
@@ -52,10 +47,24 @@ module Ensembl
     end
 
     class AttribType < ModelBase
+
       has_many :attribs, class_name: 'Attrib'
       has_many :pheotype_feature_attrib
       has_many :phenotype_features, through: :phenotype_feature_attrib
 
+      scope :common_values, -> { where(attrib_type_id: self.mapping_hash.keys)}
+
+      def self.mapping_hash
+        @mapping_hash||={14=>:risk_allele,15=>:p_value,23=>:odds_ratio,24=>:beta}
+      end
+
+      def self.key(value)
+        mapping_hash.key(value)
+      end
+
+      def self.symbol(key)
+        mapping_hash[key]
+      end
     end
 
     class CompressedGenotypeRegion < Connection
@@ -74,11 +83,9 @@ module Ensembl
         # FIXME: Should be in GenotypeCodes class or should use caching
         allele_codes=GenotypeCode.eager_load(:allele_code).where(:genotype_code_id=>genotype_code_ids.uniq).inject({}){|hsh,gc|hsh[gc.genotype_code_id]=gc.allele_code.allele;hsh}
 
-            #genotype_code_ids.uniq.inject({}) { |hsh, gc_id | hsh[gc_id]=GenotypeCode.find(gc_id).allele_code.allele;hsh  }
-
         @igs||=unpacked_genotypes.map{|s|
           IndividualGenotype.new({ individual_id:  s[0],
-                                   genotype_code_id:s[1],
+                                   genotype_code_id: s[1],
                                    allele: allele_codes[s[1]] })}
       end
 
@@ -127,9 +134,19 @@ module Ensembl
     end
 
     class GenotypeCode < ModelBase
-
       belongs_to :allele_code
 
+      def self.genotype_for(genotype_code_id)
+        joins(:allele_code).where(genotype_code_id: genotype_code_id).order(:haplotype_id).pluck('allele_code.allele').join('|')
+      end
+
+      def self.genotypes_for(genotype_code_ids)
+        includes(:allele_code).where(genotype_code_id: genotype_code_ids).pluck('genotype_code.genotype_code_id','genotype_code.haplotype_id','allele_code.allele').group_by{|r| r[0]}.map{|k,v| [k,v.sort_by{|f,s| f[1]<=>s[1]}.map{|v| v[2]}.join('|')]}
+      end
+
+      def self.genotypes_hash_for(genotype_code_ids)
+        genotypes_for(genotype_code_ids).to_h
+      end
     end
 
     class Individual < ModelBase
@@ -161,8 +178,12 @@ module Ensembl
       belongs_to :individual
       belongs_to :population
 
-      scope :displayable, -> { joins(:population).where(population: {display:true})}
-      scope :by_ids, ->(ids) { where(individual_id: ids) }
+      scope :displayable, -> { joins(:population).merge(Population.displayable) }
+      scope :thousand_genomes, -> { joins(:population).merge(Population.thousand_genomes)}
+
+      scope :by_individual_ids, ->(ids) { where(individual_id: ids) }
+
+
     end
 
     class IndividualSynonym < Connection
@@ -189,6 +210,17 @@ module Ensembl
 
     class Phenotype < ModelBase
       has_many :phenotype_features
+
+      def studies
+        ids=phenotype_features
+        .with_studies
+        .uniq
+        .pluck(:study_id)
+
+        return nil unless ids.size > 0 #
+
+        Study.where(study_id: ids)
+      end
     end
 
     class PhenotypeFeature < ModelBase
@@ -205,19 +237,53 @@ module Ensembl
       has_many :phenotype_feature_attribs
       has_many :attrib_types, through: :phenotype_feature_attribs
 
+      scope :significant, -> { where(is_significant: true )}
+      scope :with_studies, -> { where.not(study_id:nil)}
+
       def variation
         Variation.find_by name: object_id
       end
 
+      def risk_allele
+        pf=phenotype_feature_attribs.risk_alleles.first
+        pf.value unless pf.nil?
+      end
+
+      def p_value
+        pf=phenotype_feature_attribs.p_values.first
+        pf.value unless pf.nil?
+      end
+
+      def odds_ratio
+        pf=phenotype_feature_attribs.odds_ratios.first
+        pf.value unless pf.nil?
+      end
+
+      def description
+        phenotype.description
+      end
+
     end
 
     class PhenotypeFeatureAttrib < Connection
       belongs_to :attrib_type
       belongs_to :phenotype_feature
+
+      scope :risk_alleles, -> {
+        where(attrib_type_id: AttribType.key(:risk_allele)) }
+
+      scope :p_values, -> {
+        where(attrib_type_id: AttribType.key(:p_value)) }
+
+      scope :odds_ratios, -> {
+        where(attrib_type_id: AttribType.key(:odds_ratio))}
+
+      scope :betas, -> {
+        where(attrib_type_id: AttribType.key(:beta))}
     end
 
     class Population < ModelBase
-      self.extend Ensembl::SearchByName
+      # self.extend Ensembl::AttributeLike
 
       has_many :alleles
       has_many :population_synonyms
@@ -234,6 +300,7 @@ module Ensembl
       has_many :population_genotypes
 
       scope :displayable, -> { where(display:'LD')}
+      scope :thousand_genomes, -> { displayable.starts_with(:name,'1000GENOMES')}
 
       def all_individual_populations
         IndividualPopulation.where(population_id: sub_population_ids(self)<<id)
@@ -289,16 +356,8 @@ module Ensembl
       belongs_to :variation_feature
     end
 
-    # class SeqRegion < Ensembl::Core::SeqRegion
-    #   belongs_to :coord_system
-    #   has_many :compressed_genotype_regions
-    #   has_many :phenotype_features
-    #   has_many :structureal_variation_features
-    # end
-
     class StrainGtypePoly < Connection
       belongs_to :variation
-
     end
 
     class StructuralVariation < ModelBase
@@ -347,15 +406,25 @@ module Ensembl
     end
 
     class Source < ModelBase
+      has_many :studies
+
+      scope :no_db_gap, -> { where.not(source_id: 46)}
     end
 
     class Study < ModelBase
+      # include AttributeLike
+
+      default_scope -> { includes(:source) }
+
+      belongs_to :source
+
       has_many :associate_studies, foreign_key: 'study1_id'
       has_many :associated_studies, through: :associate_studies, source: :associated_study
 
       # FIXME: No data in database
       has_many :study_variations
       has_many :variations, through: :study_variations
+
     end
 
     # FIXME: No data in database
@@ -394,7 +463,6 @@ module Ensembl
     end
 
     class Variation < ModelBase
-      self.extend Ensembl::SearchByName
 
       belongs_to :source
 
@@ -420,32 +488,91 @@ module Ensembl
         PhenotypeFeature.eager_load(:phenotype).where(object_id_column: name, type: 'Variation')
       end
 
-      def synonyms
-        variation_synonyms.map{ |vs| vs.name }
+      def all_phenotype_features
+        object_ids = synonyms
+        object_ids<<name
+        PhenotypeFeature.eager_load(:phenotype).where(object_id: object_ids, type: 'Variation')
+      end
+
+      # Made because of the need to cut down database queries
+      # @return
+      # { phenotype_feature_id =>
+      #   { :phenotype=> "Phenotype description" ,
+      #     :phenotype_id => _ ,
+      #     :p_value => _ ,
+      #     :odds_ratio => _,
+      #     :risk_allele => _ },
+      #  phenotype_feature_id =>
+      #   { :phenotype=> "Phenotype description" ,
+      #     :phenotype_id => _ ,
+      #     :p_value => _ ,
+      #     :odds_ratio => _,
+      #     :risk_allele => _ }}
+      def phenotype_features_hash
+
+        # Do enable two level inserts hsh[:first][:second]
+        hash=Hash.new{ |hsh,key| hsh[key] = Hash.new {} }
+
+        all_phenotype_features
+        .joins(:phenotype)
+        .pluck(
+            :phenotype_feature_id,
+            'phenotype.description',
+            :phenotype_id)
+        .each{ |r| hash[r[0]][:phenotype]=r[1]; hash[r[0]][:phenotype_id]=r[2]}
+
+        PhenotypeFeatureAttrib
+        .where(phenotype_feature_id: hash.keys)
+        .pluck(
+            'phenotype_feature_attrib.phenotype_feature_id',
+            'phenotype_feature_attrib.value',
+            'phenotype_feature_attrib.attrib_type_id')
+        .each{ |v| hash[v[0]][AttribType.symbol(v[2])]=v[1] }
+
+        hash
+      end
+
+      def synonym_names
+        variation_synonyms.map{|vs| vs.name}
       end
 
-      def genotype_frequencies
-        igs=compressed_genotype_vars
-        .map{|cgv| cgv.individual_genotypes }
-        .flatten
-        .each_with_object(Hash.new){ |o,hsh| hsh[o.individual_id] = o.allele;hsh}
+      # Genotype counts for each population
+      # @returns {"CSHL-HAPMAP:HapMap-CEU"=>{"C|T"=>59, "C|C"=>102, "T|T"=>12},
+      # "CSHL-HAPMAP:HapMap-YRI"=>{"C|C"=>172, "C|T"=>1}}
+      def genotype_counts
+        counts = Hash.new{ |hsh,k| hsh[k] = Hash.new 0 }
+
+        individual_populations.pluck('population.name',:individual_id).map{|ip| [ip[0],genotype_codes[individual_genotypes[ip[1]]]] }.each{|r| counts[r[0]][r[1]]+=1}
+
+        return counts
+      end
+
+      # Individual and genotype_code id's related to variation
+      # @returns
+      # Example:
+      # [[1,2],[2,3],[<individual_id>,<genotype_code_id>]]
+      def individual_genotypes
+        @individual_genotypes||=compressed_genotype_vars.map{|cgv| cgv.unpacked_genotypes }.flatten(1).to_h
+      end
 
-        counts=Hash.new 0
+      def individual_genotype_ids
+        individual_genotypes.keys
+      end
 
-        IndividualPopulation
-        .displayable
-        .by_ids(igs.keys)
-        .map{|ip| [ip.population_id,igs[ip.individual_id]] }
-        .each{|pig| counts[pig]+=1 }
+      # IndividualPopulations from individual_genotypes
+      # @returns [IndividualPopulation,IndividualPopulation,...]
+      def individual_populations
+        IndividualPopulation.where(individual_id: individual_genotype_ids)
+      end
 
-        counts.group_by{|k,v| k[0]}
+      def genotype_code_ids
+        @genotype_code_ids||=individual_genotypes.values.uniq
       end
 
-      def individual_populations(individual_ids)
-        IndividualPopulation
-        .joins(:population)
-        .where(population: { display:true })
-        .where(individual_population: { individual_id: individual_ids })
+      # Unique genotype codes from individual_genotypes
+      # @returns [<genotype_code_id>=>'G|C',2=>'A|A']
+      def genotype_codes
+        @genotype_codes||=GenotypeCode.genotypes_hash_for(genotype_code_ids)
       end
 
       # Find Variation by also using VariationSynonyms
@@ -455,18 +582,27 @@ module Ensembl
         v  = self.find_by(name: name)
         return v unless v.nil?
         vs = VariationSynonym.eager_load(:variation).find_by(name: name)
-        vs.variation unless vs.nil?
+        return vs.variation unless vs.nil?
+        nil
       end
 
-      def all_phenotype_features
-        object_ids = variation_synonyms.pluck :name
-        object_ids<<name
-        PhenotypeFeature.where(object_id: object_ids, type: 'Variation')
+      def self.find_all_by_name(name)
+        v_ids = where(name: name).pluck(:variation_id)
+        v_ids = variation_synonyms.where(name: name).pluck(:variation_id) if v_ids.nil?
+
+        return nil if v_ids.nil?
+
+        where(variation_id: v_ids).order(:name)
+      end
+
+      def genes
+        variation_genenames.pluck(:gene_name)
+      end
+
+      def positions
+        variation_features.includes(:seq_region).pluck('seq_region.name',:seq_region_start,:seq_region_end,:seq_region_strand).map{|r| Ensembl::Helpers::VariationPosition.new(r)}
       end
 
-      # def population_genotypes
-      #   PopulationGenotype.where(variation_id: id)
-      # end
     end
 
     class VariationCitation < Connection
@@ -488,6 +624,15 @@ module Ensembl
         VariationSets.where[variation_set_id: [variation_set_id.split(',').map{|id| id.to_i }]] unless variation_set_id.nil?
       end
 
+      def strand_name(id)
+        case(id)
+          when 1
+            'forward'
+          else
+            'reverse'
+        end
+      end
+
       def class_type
         Attrib.find(class_attrib_id) unless class_attrib_id.nil?
       end
@@ -502,7 +647,7 @@ module Ensembl
     end
 
     class VariationSet < ModelBase
-      self.extend Ensembl::SearchByName
+      # self.extend Ensembl::SearchByName
 
       belongs_to :short_name, foreign_key: 'short_name_attrib_id', class_name: 'Attrib'
       has_many :structural_variations
@@ -536,6 +681,19 @@ module Ensembl
     class VariationSynonym < ModelBase
       belongs_to :variation
       belongs_to :source
+
+      scope :name_like, ->(name, search_type=:starts_with){
+        at=self.arel_table
+
+        if search_type == :ends_with
+          where(at[:name].matches("%#{name}"))
+        elsif search_type == :starts_with
+          where(at[:name].matches("#{name}%"))
+        else
+          where(at[:name].matches("%#{name}%"))
+        end
+
+      }
     end
   end
 end

data/lib/ensembl/variation/tableless.rb

@@ -3,6 +3,7 @@ require 'activerecord-tableless'
 
 module Ensembl
   module Variation
+
     class IndividualGenotype < ActiveRecord::Base
       has_no_table
 
@@ -13,12 +14,9 @@ module Ensembl
       belongs_to :individual
       belongs_to :genotype_code
 
-      #has_one :allele_code, through: :genotype_code
-
       delegate :individual_populations, to: :individual
       delegate :populations, to: :individual
-
-
     end
+
   end
 end

data/lib/ensembl/version.rb

@@ -1,3 +1,3 @@
 module Ensembl
-  VERSION = '0.0.6'
+  VERSION = '0.0.7'
 end

data/lib/ensembl.rb

@@ -1,42 +1,11 @@
 require "ensembl/version"
 require 'active_record'
+require 'yaml'
 require 'active_support/core_ext'
 
 module Ensembl
-
-  class << self
-
-    attr_accessor :host, :port, :database, :username, :password
-
-    def host
-      @host||='ensembldb.ensembl.org'
-    end
-
-    def port
-      @port||=3306
-    end
-
-    def username
-      @username||='anonymous'
-    end
-
-    def password
-      @password||=''
-    end
-
-    def hg_version
-      @hg_version||='37'
-    end
-
-    def version
-      @version||='75'
-    end
-
-    def species
-      @species||='homo_sapiens'
-    end
-
-  end
+  # Load configuration from database.yml
+  ActiveRecord::Base.configurations = YAML::load(IO.read('config/database.yml'))
 
   module TableNameOverrides
     def table_name
@@ -50,71 +19,22 @@ module Ensembl
     end
   end
 
-  module SearchByName
-    def search(name)
-      table=self.arel_table
-      self.where(table[:name].matches("%#{name}%"))
-    end
-  end
-
-  module SearchByAttribute
-    def search(attribute, string)
-      table=self.arel_table
-      self.where(table[attribute].matches("%#{string}%"))
-    end
-  end
-
-  # class BaseConnection < ActiveRecord::Base
-  #   self.extend TableNameOverrides
-  #   self.abstract_class = true
-  # end
-
-  # module Core
-  #   class Connection < ActiveRecord::Base
-  #     self.extend TableNameOverrides
-  #
-  #     self.abstract_class = true
-  #
-  #     self.establish_connection :adapter  => "mysql2",
-  #     :host     => Ensembl.host,
-  #     :username => Ensembl.username,
-  #     :password => Ensembl.password,
-  #     :database => Ensembl.species+'_core_'+Ensembl.version+'_'+Ensembl.hg_version,
-  #     :reconnect => true
-  #
-  #   end
-  #
-  #   class ModelBase < Connection
-  #     self.extend PrimaryKeyOverrides
-  #
-  #     self.abstract_class = true
+  # module AttributeLike
+  #   def a_like(attribute, string, search_type=:between)
+  #     at=self.arel_table
+  #     if search_type == :ends_with
+  #       where(at[attribute].matches("%#{string}"))
+  #     elsif search_type == :starts_with
+  #       where(at[attribute].matches("#{string}%"))
+  #     else
+  #       where(at[attribute].matches("%#{string}%"))
+  #     end
   #   end
   # end
-  #
-  # module Variation
-  #   class Connection < ActiveRecord::Base
-  #     self.extend TableNameOverrides
-  #
-  #     self.abstract_class = true
-  #
-  #     self.establish_connection :adapter  => "mysql2",
-  #                               :host     => Ensembl.host,
-  #                               :username => Ensembl.username,
-  #                               :password => Ensembl.password,
-  #                               :database => Ensembl.species+'_variation_'+Ensembl.version+'_'+Ensembl.hg_version,
-  #                               :reconnect => true
-  #
-  #   end
-  #
-  #   class ModelBase < Connection
-  #     self.extend PrimaryKeyOverrides
-  #
-  #     self.abstract_class = true
-  #   end
-  # end
-
 end
 
+require File.dirname(__FILE__) + '/ensembl/helpers/like_search.rb'
 require File.dirname(__FILE__) + '/ensembl/core/activerecord.rb'
+require File.dirname(__FILE__) + '/ensembl/helpers/variation_position.rb'
 require File.dirname(__FILE__) + '/ensembl/variation/activerecord.rb'
 require File.dirname(__FILE__) + '/ensembl/variation/tableless.rb'

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: ensembl
 version: !ruby/object:Gem::Version
-  version: 0.0.6
+  version: 0.0.7
 platform: ruby
 authors:
 - Kristjan Metsalu
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-07-28 00:00:00.000000000 Z
+date: 2014-08-12 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: mysql2
@@ -106,9 +106,12 @@ files:
 - LICENSE.txt
 - README.md
 - Rakefile
+- config/database.yml.example
 - ensembl.gemspec
 - lib/ensembl.rb
 - lib/ensembl/core/activerecord.rb
+- lib/ensembl/helpers/like_search.rb
+- lib/ensembl/helpers/variation_position.rb
 - lib/ensembl/variation/activerecord.rb
 - lib/ensembl/variation/tableless.rb
 - lib/ensembl/version.rb