exodb 0.0.2

data/lib/exodb/datamodel/source.rb

@@ -0,0 +1,122 @@
+# 
+# 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
+	
+	class Source
+		include Mongoid::Document
+		#include Mongoid::Versioning
+		include Mongoid::Timestamps
+	end
+	
+	class Dataset < Source
+		
+		field :oid,					type: String
+		field :name,				type: String
+		field :experiment,			type: Array
+		
+		has_many :cells, autosave: true
+		
+		validates_uniqueness_of :oid, message: "Dataset oid of experiment is not unique"
+		index({oid: 1}, {unique: false})
+		index({name: 1})
+	end
+	
+	class Cell < Source
+		
+		field :oid,					type: String
+		field :patient,				type: String
+		field :age,					type: String
+		field :type,				type: String
+		field :typeid,				type: String
+		field :preferred,			type: Boolean
+		field :paired,				type: Boolean
+		field :purity,				type: Float
+		
+		default_scope ->{where(preferred: true)}
+		
+		belongs_to :dataset
+		has_many :variants, autosave: true, dependent: :delete
+		has_many :genes, autosave: true, dependent: :delete
+		
+		validates_uniqueness_of :oid, message: "Cell sample oid of experiment is not unique"
+		accepts_nested_attributes_for :variants
+		
+		index({oid: 1, type: 1, typeid: 1, paired: 1, purity: 1, preferred: 1})
+		
+		def self.translate!
+			self.where({}).each {|e| e.translate!}
+		end
+		
+		# get the start position of gene rely on the genome
+		# 
+		# @param [String] oid or 
+		def add_to_dataset(str)
+			
+			dataset = Dataset.where('$or' => [{'oid' => str}, {'name' => str}])
+			
+			if dataset.exists?
+				self.dataset = dataset.first()
+				#output.puts "#EXODUS:INFO This sample is added to #{dataset.first().name}." if $0 == 'pry'
+			else
+				#output.puts "#EXODUS:ERRO Cannot find dataset by #{str}."  if $0 == 'pry'
+			end
+		end
+		
+		# Translate variant
+		# 
+		# @param [Float] cutoff score 
+		def translate!(cutoff = 0)
+			self.variants.each do |variant|
+				
+				Generef.cover?(variant.location_str).each do |generef|
+					
+					if generef.can_translated?
+						
+						mainsplice = generef.longest_splice
+						position = mainsplice.get_prot_pos(variant.start)
+						
+						if !position.empty?
+							gene = self.genes.find_or_create_by({symbol: generef.symbol})
+							gene.generef = generef
+							aacid = gene.aacids.find_or_create_by({position: /\A[A-Z]#{position[0]}[A-Z]?\z/ =~ variant.aachange ? position[0] : variant.aachange.split(/(\d+)/)[1].to_i})
+							aacid.refcodon = mainsplice.get_codon(position[0]) if !aacid.refcodon
+							aacid.refaa = mainsplice.get_codon(position[0]).translate if !aacid.refaa
+							aacid.altcodon = {} if !aacid.altcodon
+							aacid.altcodon[position[1]]  = aacid.altcodon[position[1]] ? aacid.altcodon[position[:posincodon]] | variant.alternate : variant.alternate
+							aacid.isoform = [] if !aacid.isoform
+							aacid.variants.push(variant)
+							
+							generef.splices.each do |splice|
+								position = splice.get_prot_pos(variant.start)
+								aacid.isoform.push("#{generef.get_xref()}:p.#{mainsplice.get_codon(position[0]).translate}#{position[0]}") if !position.empty?
+							end
+							aacid.save!
+							gene.save!
+						end
+					end
+				end
+			end
+		end
+		
+	end
+	
+	class Tumor < Cell
+		
+		field :metastasis,			type: Boolean
+		
+	end
+	
+	class Normal < Cell
+		
+	end
+	
+end

data/lib/exodb/datamodel/variant.rb

@@ -0,0 +1,89 @@
+# 
+# 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 Variant
+		
+		include Mongoid::Document
+		include Mongoid::Versioning
+		include Mongoid::Timestamps
+		
+		include Exodb::LocationField
+		
+		#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 :alternate,			type: Array #alternate genotype
+		field :quality,				type: String
+		field :filter,				type: String
+		field :somaticStatus,		type: String #unknown, inherited, somatic
+		field :somaticScore,		type: Float #Somatic score
+		field :inheritantScore,		type: Float #Inheritant score
+		#field :fdr,					type: Float #False discovery rate score 
+		field :ctrlread,			type: String
+		field :inhreads,			type: String #reads from normal cell
+		field :reads,				type: String
+		field :predicted_damage,	type: Boolean #Temporaly field
+		field :aachange,			type: String #Temporaly field
+		
+		belongs_to :cell
+		belongs_to :aacid
+		
+		validates_uniqueness_of :oid, message: "Variant oid of experiment is not unique"
+		
+		# add this variant to original cell sample
+		# 
+		# @param [String] oid
+		def add_to_sample(str)
+			
+			sample = Cell.where({'oid' => str})
+			
+			if sample.exists?
+				self.cell = sample.first()
+				#output.puts "#EXODB:INFO This sample is added to #{dataset.first().name}." if _pry_
+			else
+				#output.puts "#EXODB:ERRO Cannot find dataset by #{str}." if _pry_
+			end
+			
+			self.oid = "#{self.location_str}:#{sample.first().oid}"
+			
+		end
+		
+		alias_method :add_to_cell, :add_to_sample
+	end
+	
+	class SNV < Variant
+		
+		# add this variant to original cell sample
+		def calculate_score
+			
+		end
+		
+	end
+	
+	class SV < Variant #
+		
+	end
+	
+	class Indel < Variant
+		
+	end
+	
+	class CNV < Variant
+		
+	end
+	
+end

data/lib/exodb/datamodel/xrefsfield.rb

@@ -0,0 +1,42 @@
+# 
+# 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 XrefsField
+		
+		extend ActiveSupport::Concern
+		
+		included do
+			
+			field :xrefs,				type: Array
+			
+			index(xrefs: 1)
+			
+		end
+		
+		module ClassMethods
+			
+		end
+		
+		# convert genomic position to codon position
+		#
+		# @param [String] namespace
+		# @return [String] codon position
+		def get_xref(ns = 'urn:miriam:refseq')
+			xref = nil
+			self[:xrefs].each {|e| xref = e if "urn:miriam:#{e.namespace}" == ns}
+			return xref
+		end
+		
+	end
+	
+end

data/lib/exodb/datamodel.rb

@@ -0,0 +1,19 @@
+# 
+# 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"
+
+require 'mongoid'
+
+require 'exodb/datamodel/locationfield.rb'
+require 'exodb/datamodel/xrefsfield.rb'
+require 'exodb/datamodel/variant.rb'
+require 'exodb/datamodel/reference.rb'
+require 'exodb/datamodel/region.rb'
+require 'exodb/datamodel/source.rb'

data/lib/exodb/dbconnection.rb

@@ -0,0 +1,83 @@
+# 
+# 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_function
+	
+	# load session file
+	#
+	# @param [String] path to session file
+    def sessionload!(sessionfile = nil)
+		if sessionfile && File.exist?(sessionfile)
+			Mongoid.load!(sessionfile, :production)
+		else
+			Mongoid.load!("#{Dir.pwd}/session.yml", :production) if File.exist?("#{Dir.pwd}/session.yml")
+		end
+    end
+    
+    # connect to the database without authentication
+	#
+	# @param [Hash, String] Connection string or a Hash in format { database: 'exodus', hosts: ['localhost:27017'], username: 'xxx', options: {}}
+	def connect(connectionstr = {})
+		
+		settings = { database: 'exodus', hosts: ['localhost:27017'], options: {}}
+		
+		if connectionstr.is_a?(Hash)
+			connectionstr.each_pair {|k, v| settings[k.to_sym] = k.to_sym == :hosts ? [v].flatten : v}
+		elsif connectionstr.is_a?(String)
+			split1 = connectionstr.split('@')
+			settings[:username] = split1[0] if split1.length > 1
+			split2 = split1[-1].split('/')
+			settings[:database] = split2[1] if split2.length > 1
+			settings[:hosts] = [split2[0]].flatten
+		end
+		
+		password = ask("Password:  ") { |q| q.echo = "*" } if settings[:username] && !settings[:password]
+		
+		Mongoid::Sessions.disconnect
+		Mongoid::Sessions.clear
+		Mongoid.load_configuration({"sessions"=>{"default"=> password ? settings.merge({password: password}) : settings}})
+		
+		return "#EXODB:INFO Connection with #{settings}" if Pry.current
+	end
+	
+	# connect to the database with authentication
+	#
+	# @param [String] (see #connect)
+	#def connect!(db = 'exodus', hosts = ['localhost:27017'])
+	#	username = ask("Username:  ")
+	#	password = ask("Password:  ") { |q| q.echo = "*" }
+	#	settings = {}
+	#	if db
+	#		settings = {"database"=>db, "hosts"=>[hosts].flatten, "username"=>username}
+	#		Mongoid::Sessions.disconnect
+	#		Mongoid::Sessions.clear
+	#		Mongoid.load_configuration({"sessions"=>{"default"=>settings.merge({password: password})}})
+	#	end
+	#	return "#EXODUS:INFO Connection with #{settings} setting" if Pry.current
+	#end
+	
+	# Return the session setting
+	#
+	# @return [String] the session setting
+	def session
+		Mongoid.session(:default)
+	end
+	
+	# Return the current database name
+	#
+	# @return [String] the database name
+	def current_database
+		self.session.options[:database]
+	end
+	
+end

data/lib/exodb/exception.rb

@@ -0,0 +1,18 @@
+# 
+# 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 CreateUserError < Exception
+		
+	end
+
+end

data/lib/exodb/usermanage.rb

@@ -0,0 +1,84 @@
+# 
+# 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_function
+	
+	# General command for creating a user
+	#
+	# @param [String] user name to be created
+	# @param [String] database that user created
+	# @param [Array] roles to be assigned to the user
+	def create_user(username, database, *roles)
+		
+		if database != 'admin'
+			
+			password = nil
+			confirmed = nil
+			
+			while !password || password != confirmed
+				password = ask("New password:  ") { |q| q.echo = "*" }
+				confirmed = ask("Confirm password:  ") { |q| q.echo = "*" }
+				
+				puts "Password not match!\n" if password != confirmed
+				
+			end
+			
+			if database
+				Mongoid.session(:default).with(database: database).command(
+					createUser: username,
+					pwd: password,
+					roles: roles.flatten
+				)
+			else
+				Mongoid.session(:default).command(
+					createUser: username,
+					pwd: password,
+					roles: roles.flatten
+				)
+			end
+		else
+			raise CreateUserError, 'Cannot create user on admin database'
+		end
+	end
+	
+	# To create an admin user
+	#
+	# @param [String] user name to be created
+	# @param [String] database that user created
+	def create_admin(username, database)
+		
+		create_user(username, database, "readWrite", "dbAdmin", "userAdmin")
+		
+	end
+	
+	# To create a rw user
+	#
+	# @param [String] user name to be created
+	# @param [String] database that user created
+	def create_rwuser(username)
+		
+		create_user(username, Mongoid.session.options[:database], "readWrite")
+		
+	end
+	
+	# To create a read-only user
+	#
+	# @param (see #create_rwuser)
+	def create_ruser(username)
+		
+		create_user(username, Mongoid.session.options[:database], "read")
+		
+	end
+	
+end

data/lib/exodb/utils/upload_generef.rb

@@ -0,0 +1,163 @@
+# 
+# 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
+	
+	module Utils
+		
+		module_function
+		
+		# Upload gene information to database using gff3 and genome sequence fasta file
+		#
+		# @param [String] gff3 file
+		# @param [String] assembly name [default: gff file name]
+		def upload_generef_from_gff3(filename, assembly = nil)
+			
+			gff = Bio::GFF::GFF3.new(File.open(filename).read)
+			
+			processDbxref = lambda do |str|
+				case str
+				when /^GeneID/
+					return "urn:miriam:ncbigene:#{str.split(/:/)[1]}"
+				when /^HGNC/
+					return "urn:miriam:hgnc:#{str}"
+				when /^HPRD/
+					return "urn:miriam:hprd:#{str.split(/:/)[1]}"
+				when /^miRBase/
+					return "urn:miriam:mirbase:#{str.split(/:/)[1]}"
+				when /^Genbank/
+					return "urn:miriam:refseq:#{str.split(/:/)[1]}"
+				when /^CCDS/
+					return "urn:miriam:ccds:#{str.split(/:/)[1]}"
+				when /^MIM/
+					return "urn:miriam:omim:#{str.split(/:/)[1]}"
+				else
+					return str
+				end
+			end
+			
+			assembly = assembly ? assembly : File.basename(filename, '.gff3')
+			
+			regions = {}
+			genes = {}
+			seq = {}
+			
+			
+			gff.records.each do |e|
+				
+				case e.feature
+				when 'region'
+					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
+						end
+					end
+					
+					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}
+					end
+					
+				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: []}
+					
+					e.attributes.each do |attr|
+						case attr[0]
+						when 'Dbxref'
+							gene[:xrefs].push(processDbxref.call(attr[1]))
+						when 'Name'
+							gene[:xrefs].push("urn:miriam:hgnc.symbol:#{attr[1]}") if attr[1] !~ /^LOC\d+$/
+						when 'pseudo'
+							gene[:psuedo] = attr[1] == 'true' ? true : false
+						when 'ID'
+							gene[:id] = attr[1]
+						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]
+					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: []}
+					
+					e.attributes.each do |attr|
+						case attr[0]
+						when 'Dbxref'
+							rna[:xrefs].push(processDbxref.call(attr[1]))
+						when 'pseudo'
+							rna[:psuedo] = attr[1] == 'true' ? true : false
+						when 'ID'
+							rna[:id] = attr[1]
+						when 'Parent'
+							rna[:parent] = attr[1]
+						end
+					end
+					
+					genes[rna[:id]] = rna
+					genes[rna[:parent]][:childs].push(rna[:id]) if rna[:parent]
+					
+				when 'exon'
+					e.attributes.each do |attr|
+						case attr[0]
+						when 'Parent'
+							genes[attr[1]][:exon].push([e.start, e.end].sort)
+						end
+					end
+				when 'CDS'
+					e.attributes.each do |attr|
+						case attr[0]
+						when 'Parent'
+							genes[attr[1]][:cds].push([e.start, e.end].sort)
+						end
+					end
+				end
+			end
+			
+			genes.each_pair do |k, v|
+				if v[: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)
+					
+					v[:childs].each do |child|
+						
+						rna = Splice.new()
+						data = genes[child]
+						rna.xrefs = data[:xrefs]
+						rna.exon = data[:exon].sort
+						rna.cds = data[:cds].sort
+						
+						gene.splices.push(rna)
+						
+					end
+					
+					p gene.save!
+					
+				end
+				
+			end
+			
+			
+		end
+	end
+	
+end

data/lib/exodb/utils/upload_var.rb

@@ -0,0 +1,60 @@
+# 
+# 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 'csv'
+
+module Exodb
+	
+	module Utils
+		
+		module_function
+		
+		def load_variant_from_csv(csvfile)
+			
+			CSV.read(csvfile, col_sep: "\t", headers: true).each do |record|
+				
+				var = SNV.new()
+				var.parse_location("#{record["chromosome"]}:#{record["start position"]}")
+				var.reference = record["ref nucleotide"].split('/')[0]
+				var.alternate = record["var nucleotide"].split('/').uniq
+				var.somaticStatus = record["Somatic Status"]
+				var.reads = record["Reads"]
+				var.predicted_damage = record["PolyPhen"] =~ /probably_damaging/ || record["SIFT"] =~ /deleterious/i || record["PROVEAN"] =~ /deleterious/i ? true : false
+				var.aachange = record["AA Change"]
+				var.add_to_sample(record["cell"])
+				
+				p var.save!
+			end
+		end
+		
+		def load_sample_from_csv(csvfile)
+			CSV.read(csvfile, col_sep: "\t", headers: true).each do |record|
+				
+				sample = Tumor.new({oid: record["SampleFinal"],
+								type: record["Type"].downcase,
+								typeid: "urn:miriam:bioportal.meddra:#{record["Type"] =~ /^spitz /i ? '10041632' : record["Type"] =~ /^spitzoid /i ? '10072450' : '10028679'}",
+								patient: record["SampleFinal"].split('T')[0],
+								preferred: record["Preferred"] == 'Y' ? true : false,
+								paired: record["merge41final"] =~ /\Apaired\z/i ? true : false})
+				
+				sample.add_to_dataset('internal.ds:000001')
+				
+				p sample.save!
+				
+			end
+		end
+	end
+	
+end
+
+#Exodb::Utils.load_sample_from_csv('Samples_all.txt')
+#Exodb::Utils.load_variant_from_csv('NovelSNVs_13.txt')
+