exodb 0.1.2 → 0.1.3

data/lib/exodb/datamodel/region.rb

@@ -22,7 +22,6 @@ module Exodb
 	class Gene < Region
 		
 		include Mongoid::Versioning
-		include Exodb::GenomeLocationField
 		
 		field :symbol,				type: String
 		field :loh,					type: Boolean
@@ -31,7 +30,8 @@ module Exodb
 		belongs_to :generef
 		belongs_to :cell
 		
-		index({'symbol' => 1, 'aacids.position' => 1}, background: true)
+		index({'symbol' => 1}, background: true)
+		
 	end
 	
 	class Change
@@ -47,12 +47,14 @@ module Exodb
 	class Onexon < Change
 		
 		field :aaposition,			type: Integer # position referenced to the first codon from the longest splice variant
-		field :refcodon,			type: String
+		field :refna,				type: String
 		field :refaa,				type: String
-		field :altcodon,			type: Hash
-		field :inhcodon,			type: Hash
+		field :altna,				type: Hash
+		field :inhna,				type: Hash
 		field :isoform,				type: Array
 		
+		index({'aaposition' => 1}, background: true)
+		
 	end
 	
 	class Aa < Onexon

data/lib/exodb/datamodel/source.rb

@@ -41,6 +41,7 @@ module Exodb
 		field :paired,				type: Boolean
 		field :purity,				type: Float
 		field :labels,				type: Hash
+		field :metastasis,			type: Boolean
 		
 		default_scope ->{where(preferred: true)}
 		
@@ -109,14 +110,4 @@ module Exodb
 		
 	end
 	
-	class Tumor < Cell
-		
-		field :metastasis,			type: Boolean
-		
-	end
-	
-	class Normal < Cell
-		
-	end
-	
 end

data/lib/exodb/datamodel/variant.rb

@@ -17,17 +17,14 @@ module Exodb
 		include Mongoid::Versioning
 		include Mongoid::Timestamps
 		
-		include Exodb::GenomeLocationField
+		include Exodb::VarLocationField
 		
 		#max_versions 5
 		
-		#PATTERN = /(?<gene>[A-Z0-9]+)-?(?<position>[0-9,]*|[is]?)(?<to>[A-Z=]*)/
-		#SILENTSIGN = '='
-		
 		field :oid,					type: String # chromosome:position..alternative:samplename
 		field :reference,			type: String #reference genotype
-		field :alternaten,			type: Array #alternate genotype from tumor cell
-		field :alternatet,			type: Array #alternate genotype from normal cell
+		field :genotypet,			type: Array # genotype from tumor cell
+		field :genotypen,			type: Array # genotype from normal cell
 		field :filter,				type: String
 		field :passfilter,			type: Boolean #, default: false
 		field :somstatus,			type: String #unknown, inherited, somatic
@@ -39,16 +36,17 @@ module Exodb
 		field :qualt,				type: String #pileup quality from tumor cell
 		field :qualn,				type: String #pileup quality from normal cell
 		field :predicted_damage,	type: Boolean #Temporaly field
-		field :temp,				type: Array #Temporaly field
+		field :evidence,			type: Array # experimental evidence of the variant
+		field :temp,				type: Hash #Temporaly field
 		
 		belongs_to :cell
 		belongs_to :aacid
 		
 		validates_uniqueness_of :oid, message: "Variant oid of experiment is not unique"
 		
-		def self.from_pileup(pileupline)
-			
-		end
+		index({oid: 1, reference: 1, alternatet: 1, passfilter: 1, somstatus: 1, somscore: 1, inhscore: 1, predicted_damage: 1}, background: true)
+		
+		before_save :update_oid
 		
 		# add this variant to original cell sample
 		# 
@@ -64,68 +62,10 @@ module Exodb
 				#output.puts "#EXODB:ERRO Cannot find dataset by #{str}." if $0 == 'pry'
 			end
 			
-			self.oid = "#{self.location_str}:#{sample.first().oid}"
-			
 		end
 		
 		alias_method :add_to_cell, :add_to_sample
 		
-		## overwrite default assignment of pileuplinet
-		## automatically update :reference, :alternatet, and location
-		## 
-		## @param [String] pile-up line from mpileup
-		#def pileuplinet=(pileupline)
-		#	begin
-		#		if pileupline.is_a?(Bio::DB::Pileup)
-		#			pile = pileupline
-		#		else
-		#			pile = Bio::DB::Pileup.new(pileupline.chomp)
-		#		end
-		#		
-		#		self[:reference] = pile.ref_base
-		#		self.parse_location("#{pile.ref_name}:#{pile.pos}..#{pile.pos}")
-		#		self[:alternatet] = pile.genotype_list
-		#		self[:pileuplinet] = pileupline
-		#	rescue
-		#		self[:pileuplinet] = pileupline
-		#	end
-		#end
-		#
-		## overwrite default assignment of pileuplinen
-		## automatically update :alternaten
-		## 
-		## @param [String] pile-up line from mpileup
-		#def pileuplinen=(pileupline)
-		#	begin
-		#		if pileupline.is_a?(Bio::DB::Pileup)
-		#			pile = pileupline
-		#		else
-		#			pile = Bio::DB::Pileup.new(pileupline.chomp)
-		#		end
-		#		
-		#		if self[:reference] == pile.ref_base
-		#			self.parse_location("#{pile.ref_name}:#{pile.pos}..#{pile.pos}")
-		#			self[:alternaten] = pile.genotype_list
-		#			self[:pileuplinen] = pileupline
-		#		end
-		#	rescue
-		#		self[:pileuplinen] = pileupline
-		#	end
-		#end
-		#
-		## overwrite default read of pileuplinet
-		## 
-		## @return [Bio::DB::Pileup] pile-up object of tumor sample
-		#def pileuplinet()
-		#	return Bio::DB::Pileup.new(self[:pileuplinet])
-		#end
-		#
-		## overwrite default read of pileuplinet
-		## 
-		## @return [Bio::DB::Pileup] pile-up object of normal sample
-		#def pileuplinen()
-		#	return Bio::DB::Pileup.new(self[:pileuplinen])
-		#end
 		
 		# apply filter to the variant
 		# the result is kelp in passfilter
@@ -134,23 +74,16 @@ module Exodb
 		def apply_filter(filter)
 			self[:filter] = filter
 		end
-	end
-	
-	class SNV < Variant
 		
-		# add this variant to original cell sample
-		def calculate_score
-			
+		def alternate
+			return self[:genotypet] - [self[:reference]]
 		end
 		
-	end
-	
-	class Indel < Variant #
-		
-	end
-	
-	class CNV < Variant
+		protected
 		
+		def update_oid
+			self.oid = "#{self.location_str}:#{self.location['assembly']}:#{self.cell.oid}"
+		end
 	end
 	
 end

data/lib/exodb/datamodel/varlocfield.rb

@@ -0,0 +1,106 @@
+# 
+# Exodb
+# Copyright (C) 2014
+#
+# author: Natapol Pornputtapong <natapol.por@gmail.com>
+#
+# Documentation: Natapol Pornputtapong (RDoc'd and embellished by William Webber)
+#
+
+# raise "Please, use ruby 1.9.0 or later." if RUBY_VERSION < "1.9.0"
+
+module Exodb
+    
+	module VarLocationField
+		
+		extend ActiveSupport::Concern
+		
+		included do
+			field :location,		type: Hash #{chr: '', position: x, assembly: x}
+			field :convlocation,	type: Array
+			
+			validates_presence_of :location
+			
+			index({location: 1}, background: true)
+			
+		end
+		
+		module ClassMethods
+			
+			
+		end
+		
+		# get the chromosome
+		#
+		# @return [Integer] chromosome
+		def chromosome
+			self[:location]['chr']
+		end
+		
+		# Assign gene location in format of chromosome_number:pos;build
+		#
+		# @param [String] gene location in format of chromosome_number:start..stop
+		def parse_locstr(loc_str)
+			
+			dat = loc_str.split(/:/)
+			
+			return {'chr' => dat[0],
+				'position' => dat[1].to_i,
+				'assembly' => dat[2] ? Exodb::ASSEMBLY.has_key?(dat[2]) ? Exodb::ASSEMBLY[dat[2]] : dat[2] : Exodb::DEFAULTASSEMBLY
+			}
+			
+		end
+		
+		# Return location from specific genome assembly
+		#
+		# @param [String] assembly version
+		#
+		# @return [String] location string in chromosome:position
+		def location_str(assembly = nil)
+			
+			result = nil
+			
+			if assembly == nil || Exodb::ASSEMBLY[assembly] == self[:location]['assembly']
+				result = "#{self[:location]['chr']}:#{self[:location]['position']}:#{self[:location]['assembly']}"
+			else
+				self[:convlocation].each {|e| result = "#{[e['chr'], e['position'], e['assembly']].join(':')}" if e['assembly'] == Exodb::ASSEMBLY[assembly]}
+			end
+			
+			return result
+		end
+		
+		# Assign location
+		#
+		# @param [String, Hash] location string in chromosome:start..stop or chromosome:start-stop format
+		def location=(loc)
+			
+			if loc.is_a?(String)
+				
+				begin
+					self[:location] = parse_locstr(loc).delete_if {|k, v| ['start', 'stop'].include?(k)}
+				rescue
+					
+				end
+				
+			end
+		end
+		
+		# Assign location
+		#
+		# @param [String, Hash] location string in chromosome:start..stop or chromosome:start-stop format
+		def convlocation=(loc)
+			
+			if loc.is_a?(String)
+				
+				begin
+					self[:convlocation].push(parse_locstr(loc).delete_if {|k, v| ['start', 'stop'].include?(k)})
+				rescue
+					
+				end
+				
+			end
+			
+		end
+	end
+    
+end

data/lib/exodb/datamodel/xrefsfield.rb

@@ -25,6 +25,10 @@ module Exodb
 		
 		module ClassMethods
 			
+			def where_xrefs(str)
+				return self.where({xrefs: str})
+			end
+			
 		end
 		
 		# convert genomic position to codon position

data/lib/exodb/extra.rb

@@ -0,0 +1,17 @@
+# 
+# Exodb
+# Copyright (C) 2014
+#
+# author: Natapol Pornputtapong <natapol.por@gmail.com>
+#
+# Documentation: Natapol Pornputtapong (RDoc'd and embellished by William Webber)
+#
+
+# raise "Please, use ruby 1.9.0 or later." if RUBY_VERSION < "1.9.0"
+
+require 'exodb/extra/upload_generef.rb'
+require 'exodb/extra/upload.rb'
+
+module Exodb
+
+end

data/lib/exodb/extra/upload.rb

@@ -0,0 +1,43 @@
+# 
+# Exodus
+# Copyright (C) 2014
+#
+# author: Natapol Pornputtapong <natapol.por@gmail.com>
+#
+# Documentation: Natapol Pornputtapong (RDoc'd and embellished by William Webber)
+#
+
+# raise "Please, use ruby 1.9.0 or later." if RUBY_VERSION < "1.9.0"
+
+
+module Exodb
+
+	class Mapping
+		
+		def self.load_assembly_mapping!(from, to)
+			
+			self.where(from: Exodb::assembly(from), to: Exodb::assembly(to))
+			
+			Exodb::Chrref.where({:assembly => Exodb::assembly(from), 'location.chr' => /\Achr/}).each do |chr|
+				
+				Exodb::Ensembl::REST.assembly_map(chr.locstr =~ /\Achr/ ? chr.locstr[3..-1] : chr.locstr,  Exodb::assembly(to),  Exodb::assembly(from))["mappings"].each do |map|
+					
+					#map = {"original"=>{"seq_region_name"=>"1", "strand"=>1, "coordinate_system"=>"chromosome", "end"=>235674528, "start"=>235669016, "assembly"=>"GRCh37"}, "mapped"=>{"seq_region_name"=>"1", "strand"=>1, "coordinate_system"=>"chromosome", "end"=>235511225, "start"=>235505713, "assembly"=>"GRCh38"}}
+					
+					mapping = self.new()
+					mapping.chr = "#{map['original']["coordinate_system"] == "chromosome" ? 'chr' : ''}#{map['original']["seq_region_name"]}"
+					mapping.start = map['original']["start"]
+					mapping.stop = map['original']["end"]
+					mapping.from = map['original']["assembly"]
+					mapping.tchr = "#{map['mapped']["coordinate_system"] == "chromosome" ? 'chr' : ''}#{map['mapped']["seq_region_name"]}"
+					mapping.tstart = map['mapped']["start"]
+					mapping.tstop = map['mapped']["end"]
+					mapping.coeff = map['mapped']["strand"]
+					mapping.to = map['mapped']["assembly"]
+					
+					p mapping.save!
+				end
+			end
+		end
+	end
+end

data/lib/exodb/{utils → extra}/upload_generef.rb

@@ -17,10 +17,11 @@ module Exodb
 		module_function
 		
 		# Upload gene information to database using gff3 and genome sequence fasta file
+		# Exodb::Utils.upload_generef_from_gff3('ref_GRCh37.p5_top_level.gff3')
 		#
 		# @param [String] gff3 file
 		# @param [String] assembly name [default: gff file name]
-		def upload_generef_from_gff3(filename, assembly = nil)
+		def upload_generef_from_gff3(filename, assembly = Exodb::DEFAULTASSEMBLY)
 			
 			gff = Bio::GFF::GFF3.new(File.open(filename).read)
 			
@@ -45,12 +46,12 @@ module Exodb
 				end
 			end
 			
-			assembly = assembly ? assembly : File.basename(filename, '.gff3')
+			assembly = assembly.blank? ? Exodb::DEFAULTASSEMBLY : Exodb::ASSEMBLY[assembly.downcase]
 			
 			regions = {}
 			genes = {}
 			seq = {}
-			
+			regions.default='chr0'
 			
 			gff.records.each do |e|
 				
@@ -59,10 +60,16 @@ module Exodb
 					e.attributes.each do |attr|
 						case attr[0]
 						when 'chromosome'
-							regions[e.seqname] = attr[1] == 'X' ? 23 : attr[1] == 'Y' ? 24 : attr[1].to_i
+							regions[e.seqname] = e.seqname =~ /\ANC_/ ? "chr#{attr[1]}" : e.seqname
 						end
 					end
 					
+					chr = Exodb::Chrref.new()
+					chr.location=("#{regions.has_key?(e.seqname) ? regions[e.seqname] : e.seqname}:#{e.start}..#{e.end}:#{assembly}")
+					chr.oid = "#{chr.chr}:#{assembly}"
+					chr.add_to_set(:xrefs, guess_miriam(e.seqname))
+					chr.save!
+					
 					if File.exist?("./genome/#{e.seqname}.fa")
 						seq = {}
 						Bio::FlatFile.open(Bio::FastaFormat, "./genome/#{e.seqname}.fa").each {|fasta| seq[fasta.acc_version] = fasta.to_seq}
@@ -70,7 +77,7 @@ module Exodb
 					
 				when 'gene', 'tRNA'
 					
-					gene = {type: 'gene', xrefs: [], strand: e.strand, chrrefseq: "#{guess_miriam(e.seqname)}", location: "#{e.seqname =~ /\ANC_/ ? regions[e.seqname] : e.seqname}:#{e.start}..#{e.end}", childs: [], exon: [], cds: []}
+					gene = {type: 'gene', xrefs: [], strand: e.strand, chrrefseq: "#{guess_miriam(e.seqname)}", location: "#{regions[e.seqname]}:#{e.start}..#{e.end}:#{assembly}", childs: [], exon: [], cds: []}
 					
 					e.attributes.each do |attr|
 						case attr[0]
@@ -85,12 +92,14 @@ module Exodb
 						end
 					end
 					
-					gene[:sequence] = seq[e.seqname].subseq(e.start.to_i, e.end.to_i).to_s if seq.has_key?(e.seqname)
-					gene[:oid] = gene[:location]
+					gene[:sequence] = seq[e.seqname].subseq(e.start.to_i - Exodb::Generef.expanding, e.end.to_i + Exodb::Generef.expanding).to_s if seq.has_key?(e.seqname)
+					gene[:seqstart] = e.start.to_i - Exodb::Generef.expanding
+					gene[:seqstop] = e.end.to_i + Exodb::Generef.expanding
+					gene[:oid] = "#{gene[:location]}:#{assembly}"
 					genes[gene[:id]] = gene
 					
 				when /\A(transcript|[^t]*RNA)/
-					rna = {type: 'rna', xrefs: [], strand: e.strand, chr: regions[e.seqname], location: "#{regions[e.seqname]}:#{e.start}..#{e.end}", exon: [], cds: []}
+					rna = {type: 'rna', xrefs: [], strand: e.strand, chr: regions[e.seqname], location: "#{regions[e.seqname]}:#{e.start}..#{e.end}:#{assembly}", exon: [], cds: []}
 					
 					e.attributes.each do |attr|
 						case attr[0]
@@ -125,20 +134,18 @@ module Exodb
 				end
 			end
 			
-			genes.each_pair do |k, v|
-				if v[:type] == 'gene'
+			count = {succ: 0, fail: 0}
+			
+			genes.each_pair do |id, entry|
+				if entry[:type] == 'gene'
 					
-					gene = Generef.new()
-					gene.oid = v[:oid] if v.has_key?(:oid)
-					gene.xrefs = v[:xrefs]
-					gene.parse_location(v[:location])
-					gene.chrrefseq = v[:chrrefseq]
-					gene.strand = v[:strand]
-					gene.psuedo = v[:psuedo] if v[:psuedo]
-					gene.genomeref = assembly
-					gene.sequence = v[:sequence] if v.has_key?(:sequence)
+					gene = Exodb::Generef.new()
 					
-					v[:childs].each do |child|
+					entry.each_pair do |k, v|
+						gene.method(:"#{k}=").call(v) if ![:type, :childs, :exon, :cds].include?(k)
+					end
+					
+					entry[:childs].each do |child|
 						
 						rna = Isoform.new()
 						data = genes[child]
@@ -150,12 +157,19 @@ module Exodb
 						
 					end
 					
-					puts "STATUS: #{gene.save! ? "SUCCESS" : "FAIL"}: Deposit Gene reference #{gene.xrefs[0]}"
+					if gene.save!
+						count[:succ] += 1
+						Exodb::putstv "Deposit Gene reference #{gene.xrefs[0]}"
+					else
+						count[:fail] += 1
+						Exodb::putstv "Deposit Gene reference #{gene.xrefs[0]}"
+					end
 					
 				end
 				
 			end
 			
+			Exodb::putst "SUCCESS: #{count[:succ]} , FAIL: #{count[:fail]}"
 			
 		end
 	end