sylfy 0.0.2 → 0.1.0

data/lib/sylfy/mathf.rb

@@ -0,0 +1,40 @@
+# 
+# UniSysDB library in Ruby
+# Copyright (C) 2012
+#
+# author: Natapol Pornputtapong <natapol@chalmers.se>
+#
+# Documentation: Natapol Pornputtapong (RDoc'd and embellished by William Webber)
+#
+
+require 'gsl'
+
+module Sylfy
+    # additional Mathematic function
+    module Mathf
+        
+        # calculate log factorial
+        # @param n [Int]
+        def self.lnfactorial(n)
+            return GSL::Sf::lngamma(n+1)
+        end
+        
+        # calculate log combinatorial
+        # @param n [Int] the population size
+        # @param r [Int] the number of draws
+        def self.lncombinatorial(n, r)
+            return lnfactorial(n) - lnfactorial(r) - lnfactorial(n-r)
+        end
+        
+        # hypergeometric function
+        # @param nn [Int] the population size
+        # @param kk [Int] the number of success states in the population
+        # @param n [Int] the number of draws
+        # @param k [Int] the number of successes
+        def self.hypergeometric(k, nn, kk, n)
+            log_h = lncombinatorial(kk, k) + lncombinatorial(nn-kk, n-k) - lncombinatorial(nn, n)
+            return (Math::E ** log_h)
+        end
+        
+    end
+end

data/lib/sylfy/pattern.rb

@@ -0,0 +1,14 @@
+# 
+# Sylfy library in Ruby
+# Copyright (C) 2012
+#
+# @author Natapol Pornputtapong <natapol@chalmers.se>
+#
+
+module Sylfy
+    module Pattern
+        INCHI = /^InChI\=1S\/[A-Za-z0-9\.\+]+(\/[cnpqbtmsih][A-Za-z0-9\-\+\(\)\,\?\*\;\.]+)*$/
+        SPINCHI = /^InChI\=1[S]?\/[A-Za-z0-9\.\+]+(\/[cnpqbtmsih][A-Za-z0-9\-\+\(\)\,\?\*\;\.]+)*$/
+        INCHIKEY = /^[A-Z]{14}\-[A-Z]{10}(\-[A-N])?/
+    end
+end

data/lib/sylfy/service.rb

@@ -0,0 +1,34 @@
+# 
+# UniSysDB library in Ruby
+# Copyright (C) 2012
+#
+# author: Natapol Pornputtapong <natapol.por@gmail.com>
+#
+# Documentation: Natapol Pornputtapong (RDoc'd and embellished by William Webber)
+#
+
+require 'rubabel'
+
+#require 'sylfy/service/keggrest'
+require 'sylfy/service/ensemblrest'
+require 'sylfy/service/pubchem'
+require 'sylfy/service/ebisoap'
+require 'sylfy/service/lipidmaprest'
+require 'sylfy/service/unichemrest'
+require 'sylfy/service/biocycrest'
+require 'sylfy/service/cactusncirest'
+require 'sylfy/service/rest'
+
+module Sylfy
+    # A module supports as interfaces for web services
+    module Service
+        # Raise when puts wrong parameter
+        class ParameterError < Exception
+            
+        end
+        # Raise when services return 404 error
+        class DataNotFound < Exception
+            
+        end
+    end
+end

data/lib/sylfy/service/biocycrest.rb

@@ -0,0 +1,41 @@
+# 
+# UniSysDB library in Ruby
+# Copyright (C) 2012
+#
+# @author Natapol Pornputtapong <natapol@chalmers.se>
+#
+require 'open-uri'
+require 'rexml/document'
+
+module Sylfy
+
+	module Service
+		
+		module Biocyc
+			module REST
+				@@baseuri = 'http://websvc.biocyc.org/apixml?'
+				#~ @@prefix = { 'CAS Registry Number'=>'cas:', 'KEGG COMPOUND'=>'kegg.compound:', 'KEGG DRUG'=>'kegg.drug:', 'DrugBank'=>'drugbank:', 'PDBeChem'=>'pdb-ccd:' }
+				
+				module_function
+				
+				def apixml(apifn, id, detail = 'none')
+					detail = 'none' if !['none', 'low', 'full'].include?(detail)
+					uri = "#{@@baseuri}fn=#{apifn}&id=#{id}&detail=#{detail}"
+					begin
+						doc = REXML::Document.new(URI.parse(uri).read().strip())
+					rescue OpenURI::HTTPError
+						raise Unisys::ServiceException, "Id #{id} not found."
+					end
+					
+					return doc
+					
+				end
+				
+				def reaction_reactants_and_products(id, detail = 'none')
+					RESTBioCyc.apixml('enzymes-of-reaction', id, detail)
+					#~ RESTBioCyc.apixml('reaction-reactants-and-products', id, detail)
+				end
+			end
+		end
+	end
+end

data/lib/sylfy/service/cactusncirest.rb

@@ -0,0 +1,64 @@
+# 
+# UniSysDB library in Ruby
+# Copyright (C) 2012
+#
+# @author Natapol Pornputtapong <natapol@chalmers.se>
+#
+require 'open-uri'
+require 'net/http'
+require 'json'
+
+module Sylfy
+	
+	module Service
+		
+		
+		module CIR
+            
+            module REST
+				@@baseuri = "http://cactus.nci.nih.gov/chemical/structure"
+				@@representation_type = [:stdinchi, :stdinchikey, :smiles, :ficts, :ficus, :uuuuu, :sdf, :names, :hashisy, :image, :iupac_name]
+				
+				module_function
+				
+				def perform(identifier, representation, xml = false)
+					if @@representation_type.include?(representation.to_sym)
+						uri = "#{@@baseuri}/#{URI.escape(identifier, "[] ")}/#{representation.to_s}"
+						
+						begin
+							return URI.parse(uri).read().strip().chomp.split(/\n/)
+						rescue OpenURI::HTTPError
+							return []
+						end
+						
+					else
+						raise Sylfy::Service::ParameterError, "Wrong representation type. Should be #{@@representation_type.join(', ')}"
+					end
+				end
+				
+            end
+            
+            module_function
+            
+			def getnames(inchi)
+				result = []
+				inchi = Rubabel[inchi, :inchi].to_s(:inchikey) if inchi =~ /^InChI\=1S\/[A-Za-z0-9]+(\/[cnpqbtmsih][A-Za-z0-9\-\+\(\)\,]+)+$/
+				
+				if inchi =~ /^[A-Z]{14}\-[A-Z]{10}(\-[A-N])?/
+					result += REST.perform(inchi, :iupac_name)
+					result += REST.perform(inchi, :names)
+				end
+				
+				return result.uniq[0..21]
+			end
+             
+		end
+	end
+end
+
+#require './unisys.rb'
+#
+#a = Unisys::Service::RESTCactusNCI
+#puts a.perform('Aspirin', :stdinchi)
+#puts a.perform('(2E)-eicosenoyl-CoA', :stdinchi)
+#puts a.perform('18-HEPE', :names)

data/lib/sylfy/service/cbioportal.rb

@@ -0,0 +1,156 @@
+# @author Natapol Pornputtapong <natapol@chalmers.se>
+#
+require 'open-uri'
+require 'net/http'
+require 'rexml/document'
+require 'csv'
+
+module Sylfy
+  
+		module CBioPortal
+            
+			module REST
+				
+				BASEURI = URI.parse('http://www.cbioportal.org')
+				#BASEURI = URI.parse('http://www.cbioportal.org/public-portal/webservice.do?')
+				HTTP = Net::HTTP.new(BASEURI.host, BASEURI.port)
+				
+				module_function
+				# internal method supporting all ENSEMBLREST service
+				#
+				# @param get_path [String] service get_path
+				#
+				# @param option_type [Array] list of option type of service
+				#
+				# @param option [String] option of service please consult {http://beta.rest.ensembl.org/documentation/info/xref_id}
+				#	
+				# @return [array of Hash] results
+				#
+				def service(get_path, csv_opts = {})
+					request = Net::HTTP::Get.new(get_path)
+					response = HTTP.request(request)
+					
+					if response.code != "200"
+						raise Sylfy::Service::ParameterError, "Invalid response: #{response.code}"
+					else
+                        return CSV.parse(response.body, {col_sep: "\t", skip_lines: /^#/, headers: true}.merge(csv_opts))
+					end
+				end
+				
+				# internal method supporting all ENSEMBLREST service
+				#
+				# @param get_path [String] service get_path
+				#
+				# @param option_type [Array] list of option type of service
+				#
+				# @param option [String] option of service please consult {http://beta.rest.ensembl.org/documentation/info/xref_id}
+				#	
+				# @return [array of Hash] results
+				#
+				def getTypesOfCancer()
+					get_path = "/public-portal/webservice.do?cmd=getTypesOfCancer"
+					return Sylfy::CBioPortal::REST.service(get_path)
+				end
+				
+				# internal method supporting all ENSEMBLREST service
+				#
+				# @param get_path [String] service get_path
+				#
+				# @param option_type [Array] list of option type of service
+				#
+				# @param option [String] option of service please consult {http://beta.rest.ensembl.org/documentation/info/xref_id}
+				#	
+				# @return [array of Hash] results
+				#
+				def getCancerStudies()
+					get_path = "/public-portal/webservice.do?cmd=getCancerStudies"
+					return Sylfy::CBioPortal::REST.service(get_path)
+				end
+				
+				# internal method supporting all ENSEMBLREST service
+				#
+				# @param cancer_study_id [String] service get_path
+				#	
+				# @return [array of Hash] results
+				#
+				def getGeneticProfiles(cancer_study_id)
+					get_path = "/public-portal/webservice.do?cmd=getGeneticProfiles&cancer_study_id=#{cancer_study_id}"
+					return Sylfy::CBioPortal::REST.service(get_path)
+				end
+				
+				# internal method supporting all ENSEMBLREST service
+				#
+				# @param cancer_study_id [String] service get_path
+				#	
+				# @return [array of Hash] results
+				#
+				def getCaseLists(cancer_study_id)
+					get_path = "/public-portal/webservice.do?cmd=getCaseLists&cancer_study_id=#{cancer_study_id}"
+					return Sylfy::CBioPortal::REST.service(get_path)
+				end
+				
+				
+				# internal method supporting all ENSEMBLREST service
+				#
+				# @param case_set_id [String] case set ID (required)
+				#
+				# @param genetic_profile_id [String] one or more genetic profile IDs (required). Multiple genetic profile IDs must be separated by comma (,) characters, or URL encoded spaces, e.g. +
+				#
+				# @param gene_list [String] one or more genes, specified as HUGO Gene Symbols or Entrez Gene IDs (required). Multiple genes must be separated by comma (,) characters, or URL encoded spaces, e.g. +
+				#
+				# @return [array of Hash] results
+				#
+				def getProfileData(case_set_id, genetic_profile_id, gene_list)
+					get_path = "/public-portal/webservice.do?cmd=getProfileData&case_set_id=#{case_set_id}&genetic_profile_id=#{genetic_profile_id}&gene_list=#{gene_list}"
+					return Sylfy::CBioPortal::REST.service(get_path)
+				end
+				
+				# internal method supporting all ENSEMBLREST service
+				#
+				# @param case_set_id [String] case set ID (required)
+				#
+				# @param genetic_profile_id [String] one or more genetic profile IDs (required). Multiple genetic profile IDs must be separated by comma (,) characters, or URL encoded spaces, e.g. +
+				#
+				# @param gene_list [String] one or more genes, specified as HUGO Gene Symbols or Entrez Gene IDs (required). Multiple genes must be separated by comma (,) characters, or URL encoded spaces, e.g. +
+				#
+				# @return [array of Hash] results
+				#
+				def getMutationData(case_set_id, genetic_profile_id, gene_list)
+					get_path = "/public-portal/webservice.do?cmd=getMutationData&case_set_id=#{case_set_id}&genetic_profile_id=#{genetic_profile_id}&gene_list=#{gene_list}"
+					return Sylfy::CBioPortal::REST.service(get_path)
+				end
+				
+				# internal method supporting all ENSEMBLREST service
+				#
+				# @param case_set_id [String] case set ID (required)
+				#
+				# @return [array of Hash] results
+				#
+				def getClinicalData(case_set_id)
+					get_path = "/public-portal/webservice.do?cmd=getClinicalData&case_set_id=#{case_set_id}"
+					return Sylfy::CBioPortal::REST.service(get_path)
+				end
+				
+				# internal method supporting all ENSEMBLREST service
+				#
+				# @param case_set_id [String] case set ID (required)
+				#
+				# @param genetic_profile_id [String] one or more genetic profile IDs (required). Multiple genetic profile IDs must be separated by comma (,) characters, or URL encoded spaces, e.g. +
+				#
+				# @param gene_list [String] one or more genes, specified as HUGO Gene Symbols or Entrez Gene IDs (required). Multiple genes must be separated by comma (,) characters, or URL encoded spaces, e.g. +
+				#
+				# @return [array of Hash] results
+				#
+				def getMutationData(case_set_id, genetic_profile_id, gene_list)
+					get_path = "/public-portal/webservice.do?cmd=getMutationData&case_set_id=#{case_set_id}&genetic_profile_id=#{genetic_profile_id}&gene_list=#{gene_list}"
+					return Sylfy::CBioPortal::REST.service(get_path)
+				end
+			end
+			
+		end
+		
+	
+end
+
+
+Sylfy::CBioPortal::REST.getMutationData("BRAF", "skcm_tcga_mutations").each {|e| p e; exit}

data/lib/sylfy/service/chembl.rb

@@ -0,0 +1,106 @@
+# 
+# UniSysDB library in Ruby
+# Copyright (C) 2012
+#
+# @author Natapol Pornputtapong <natapol@chalmers.se>
+#
+require 'open-uri'
+require 'net/http'
+require 'json'
+
+module Sylfy
+	
+	module Service
+		
+		
+		module ChEMBL
+			module REST
+			
+			
+				@@baseuri = "https://www.ebi.ac.uk/chemblws"
+				
+				module_function
+				
+				def perform(input)
+					
+					uri = "#{@@baseuri}/#{input}"
+					p uri
+					begin
+						return URI.parse(uri).read().strip()
+					rescue OpenURI::HTTPError => e
+						raise ServiceException, e
+					end
+				end
+				
+				private :perform
+				# Method for setting inchi data member
+				#
+				# == Parameters:
+				# id::
+				#   LipidMap id to search
+				# :id, :dataPrimarySource, :xrefs, :relations :inchi, :formula, :smiles, :inchiKey :names
+				
+				def compounds(id, domain = '', target = '', json = true)
+					input = 'compounds'
+					if [:stdinchikey, :smiles, :substructure].include?(domain.to_sym)
+						input += "/#{domain}/#{id}"
+						input += ".json" if json
+					elsif domain.to_sym == :similaritiy 
+						input += target.class == Integer ? "/#{domain}/#{id}/#{target}" : "/#{domain}/#{id}"
+						input += ".json" if json
+					elsif target.to_sym == :bioactivities
+						input += "/#{id}/bioactivities"
+						input += ".json" if json
+					elsif target == :image || target =~ /^image?dimensions=(\d+)$/ || target =~ /^image=(\d+)$/
+						if $1
+							input += "/#{id}/image?dimensions=#{$1}"
+						else
+							input += "/#{id}/image?dimensions=300"
+						end
+					else
+						input += "/#{id}"
+						input += ".json" if json
+					end
+					
+					return perform(input)
+					
+				end
+				
+				def targets(id, domain = '', bioactiv = false, json = true)
+					input = 'targets'
+					if id == ''
+						
+					elsif [:uniprot, :refseq].include?(domain.to_sym)
+						input += "/#{domain}/#{id}"
+						
+					elsif bioactiv
+						input += "/#{id}/bioactivities"
+					else
+						input += "/#{id}"
+					end
+					
+					input += ".json" if json
+					return perform(input)
+					
+				end
+				
+				def assays(id, bioactiv = false, json = true)
+					input = "assays/#{id}"
+					input += "/bioactivities" if bioactiv
+					input += ".json" if json
+					return perform(input)
+					
+				end
+			end
+		end
+		
+	end
+	
+end
+
+#require 'json'
+#require './unisys.rb'
+#
+#p JSON.load(Unisys::Service::RESTChEMBL.compounds('CHEMBL1'))
+#p JSON.load(Unisys::Service::RESTChEMBL.targets('CHEMBL2477'))
+#p JSON.load(Unisys::Service::RESTChEMBL.assays('CHEMBL1217643'))