ms-mascot 0.2.2 → 0.3.0

data/History

@@ -1,3 +1,11 @@
+== 0.3.0 / 2009-05-25
+
+Updated to utilize Tap-0.17.0
+
+* reworked format_mgf as a dump task (dump/mgf)
+* added a load/mgf task
+* added xml format for accessing results
+
 == 0.2.2 / 2009-03-31
 
 * updates to use latest tap

data/lib/ms/mascot/dump/mgf.rb

@@ -0,0 +1,64 @@
+require 'tap/tasks/dump'
+require 'ms/mascot/mgf/entry'
+
+module Ms
+  module Mascot
+    module Dump
+      # :startdoc::task dumps a fragment spectrum as mgf
+      #
+      # Formats the data produced by an Ms::Mascot::Fragment task as an mgf. The
+      # configurations specify various details of the dump, including the
+      # precision and default headers.
+      #
+      #   % tap run -- fragment TVQQEL --: dump/mgf
+      #
+      # :startdoc::task-
+      #
+      # Examples:
+      # 
+      #   # reformat an mgf file (inefficient but works)
+      #   % rap load/mgf --file OLD_FILE --:i dump/mgf --mz-precision 2 > NEW_FILE
+      #
+      class Mgf < Tap::Tasks::Dump
+      
+        config :default_headers, {}, &c.hash        # A hash of default headers
+        config :mz_precision, 6, &c.integer         # The precision of mzs
+        config :intensity_precision, 0, &c.integer  # The precision of intensities
+        config :pepmass_precision, 6, &c.integer    # The precision of peptide mass
+      
+        config :prefix, nil, &c.string_or_nil       # An optional prefix
+        config :suffix, "\n", &c.string_or_nil      # An optional suffix
+      
+        # Maps common variations of header keys (typically output
+        # by a fragment task) to Mgf::Entry header strings.
+        HEADER_MAP = {:parent_ion_mass => 'PEPMASS'}
+      
+        # Dumps the object to io as YAML.
+        def dump(obj, io)
+          unless obj.kind_of?(Ms::Mascot::Mgf::Entry)
+            data, headers = obj
+            mgf_headers = format_headers(headers)
+            obj = Ms::Mascot::Mgf::Entry.new(mgf_headers, data)
+          end
+          
+          io << prefix if prefix
+          obj.dump(io, config)
+          io << suffix if suffix
+        end
+        
+        protected
+      
+        # helper to format the headers properly for an mgf entry
+        def format_headers(headers) # :nodoc:
+          headers ||= {}
+          mgf_headers = {}
+          default_headers.merge(headers).each_pair do |key, value|
+            key = HEADER_MAP[key] || key.to_s.upcase
+            mgf_headers[key] = value
+          end
+          mgf_headers
+        end
+      end
+    end
+  end
+end

data/lib/ms/mascot/export.rb

@@ -3,7 +3,7 @@ require 'ms/mascot/validation'
 
 module Ms
   module Mascot
-    # :startdoc::manifest exports results from a search
+    # :startdoc::task exports results from a search
     class Export < Tap::Mechanize::Submit
       include Validation
       
@@ -15,7 +15,7 @@ module Ms
         config "pep_expect", "1", &MASCOT_SWITCH
         config "prot_mass", "1", &MASCOT_SWITCH
         config "protein_master", "1", &MASCOT_SWITCH
-        config "_server_mudpit_switch", 0.000000001, &c.num
+        config "_server_mudpit_switch", 0.000000001, &c.numeric
         config "pep_exp_mz", "1", &MASCOT_SWITCH
         config "do_export", "1", &MASCOT_SWITCH
         config "pep_delta", "1", &MASCOT_SWITCH
@@ -37,7 +37,7 @@ module Ms
         config "_showallfromerrortolerant", ""
         config "prot_hit_num", "1", &MASCOT_SWITCH
         config "search_master", "1", &MASCOT_SWITCH
-        config "_sigthreshold", 0.05, &c.num
+        config "_sigthreshold", 0.05, &c.numeric
         config "show_params", "1", &MASCOT_SWITCH
         config "show_mods", "1", &MASCOT_SWITCH
         config "show_header", "1", &MASCOT_SWITCH

data/lib/ms/mascot/formats/xml.rb

@@ -0,0 +1,66 @@
+require 'nokogiri'
+
+module Ms
+  module Mascot
+    module Formats
+      class Xml
+        module Utils
+          def hashify(nodes, cast=true)
+            nodes.inject({}) do |hash, node|
+              value = node.content
+              hash[node.name] = cast ? objectify(value) : value
+              hash
+            end
+          end
+
+          def objectify(value)
+            case value
+            when nil
+              nil
+            when /\A\d+(\.\d+)?\z/ 
+              $1 ? value.to_f : value.to_i
+            else
+              value
+            end
+          end
+        end
+        
+        attr_reader :doc
+        
+        def initialize(xml)
+          xml = xml.read if xml.respond_to?(:read)
+          xml = xml.sub(%q{xmlns="http://www.matrixscience.com/xmlns/schema/mascot_search_results_2"}, "")
+          @doc = Nokogiri::XML(xml)
+        end
+        
+        def header
+          doc.at("/mascot_search_results/header")
+        end
+        
+        def modifications
+          doc.xpath("/mascot_search_results/variable_mods/modification")
+        end
+        
+        def search_parameters
+          doc.xpath("/mascot_search_results/search_parameters")
+        end
+        
+        def format_parameters
+          doc.xpath("/mascot_search_results/format_parameters")
+        end
+        
+        def hits
+          doc.xpath("/mascot_search_results/hits/hit")
+        end
+        
+        def proteins(hit)
+          hit.xpath("protein")
+        end
+        
+        def peptides(protein)
+          protein.xpath("peptide")
+        end
+      end
+    end
+  end
+end

data/lib/ms/mascot/fragment.rb

@@ -4,7 +4,7 @@ require 'ms/mascot/spectrum'
 module Ms
   module Mascot
   
-    # :startdoc::manifest calculates a theoretical Mascot ms/ms spectrum
+    # :startdoc::task calculates a theoretical Mascot ms/ms spectrum
     #
     # Calculates the theoretical Mascot ms/ms spectrum for a peptide sequence.
     # A Mascot spectrum differs from the standard in-silico spectrum only in
@@ -23,7 +23,7 @@ module Ms
     # See Ms::Mascot::Spectrum for more details.
     class Fragment < InSilico::Fragment
       
-      config :intensity, 1, &c.num_or_nil  # a uniform intensity value
+      config :intensity, 1, &c.numeric_or_nil  # a uniform intensity value
       
       # Generates some MGF-specific headers.
       def headers(spec)

data/lib/ms/mascot/load/mgf.rb

@@ -0,0 +1,79 @@
+require 'tap/tasks/load'
+require 'ms/mascot/mgf/archive'
+
+module Ms
+  module Mascot
+    module Load
+      
+      # :startdoc::task loads entries from an mgf file
+      #
+      # Load entries from an mgf file.  A selector may be specified to select
+      # only a subset of entries; by default all entries in the mgf file are
+      # returned.
+      #
+      #   % tap run -- mgf/load --select 1 < MGF_FILE
+      #   % tap run -- mgf/load --select 1..10 < MGF_FILE
+      #
+      # Entries are always returned as an array, even when the selecton is
+      # for a single entry.
+      #
+      class Mgf < Tap::Tasks::Load
+        Archive = Ms::Mascot::Mgf::Archive
+        
+        config :select, nil do |input|    # An array selector for entries
+          if input.kind_of?(String)
+            input = "!ruby/range #{input}" if input =~ /\.{2,3}/
+            input = YAML.load(input)
+          end
+          c.validate(input, [nil, Integer, Range, Array])
+        end
+        
+        nest :filter do
+          config :title, nil, &c.regexp_or_nil
+          config :charge, nil, &c.range_or_nil
+          config :pepmass, nil, &c.range_or_nil
+          config :n_ions, nil, &c.range_or_nil
+          
+          def ok?(entry)
+            (!title   || entry.title =~ title) &&
+            (!charge  || charge.include?(entry.charge)) &&
+            (!pepmass || pepmass.include?(entry.pepmass)) &&
+            (!n_ions  || n_ions.include?(entry.data.length))
+          end
+          
+          def filter!(entries)
+            return entries unless title || charge || pepmass || n_ions
+            entries.select {|entry| ok?(entry) }
+          end
+        end
+        
+        def open_io(input)
+          if input.kind_of?(String)
+            Archive.open(input) {|io| yield(io) }
+          else
+            super(input) do |io|
+              arc = Archive.new(io)
+              result = yield(arc)
+              arc.close
+              result
+            end
+          end
+        end
+        
+        def load(arc)
+          case select
+          when Integer
+            entry = arc[select]
+            filter.ok?(entry) ? [entry] : []
+          when Range
+            filter.filter!(arc[select])
+          when Array 
+            filter.filter!(arc[*select])
+          when nil 
+            arc.select {|entry| filter.ok?(entry) }
+          end
+        end
+      end
+    end
+  end
+end

data/lib/ms/mascot/mgf/entry.rb

@@ -136,9 +136,14 @@ module Ms
           (options[:headers] || headers.keys).each do |key|
             target << "#{key.upcase}=#{headers[key]}\n"
           end
-            
-          target << "CHARGE=#{charge_to_s}\n"
-          target << "PEPMASS=#{format options[:pepmass_precision]}\n" % pepmass
+          
+          if charge
+            target << "CHARGE=#{charge_to_s}\n"
+          end
+          
+          if pepmass
+            target << "PEPMASS=#{format options[:pepmass_precision]}\n" % pepmass
+          end
           
           entry = data[0]
           data_format = case

data/lib/ms/mascot/submit.rb

@@ -3,7 +3,7 @@ require 'ms/mascot/validation'
 
 module Ms
   module Mascot
-    # :startdoc::manifest submits a PMF or MS/MS search to Mascot
+    # :startdoc::task submits a PMF or MS/MS search to Mascot
     #
     # Submits a search request to Mascot using the mgf file and the search
     # parameters in a static config file.  Correctly formatting search
@@ -46,37 +46,38 @@ module Ms
       config :uri, "http://www.matrixscience.com/cgi/nph-mascot.exe?1"  # The uri of the mascot search site
       
       # Parameters for MS/MS searching of a human sample digested with trypsin
-      nest :params do                                                   # The query parameters
-        config "ErrTolRepeat", 0, &MASCOT_SWITCH
-        config "PFA", 1, &MASCOT_SWITCH
-        config "INSTRUMENT", "Default", &c.string
-        config "REPTYPE", "peptide",  &c.string
+      nest :params do                                          # The query parameters
+        config "USERNAME", "Name",  &c.string
+        config "USEREMAIL", '', &c.string
         config "COM", "Search Title",  &c.string
+        config "INSTRUMENT", "Default", &c.string
         config "FORMAT", "Mascot generic",  &c.string
-        config "PEAK", "AUTO",  &c.string
         config "CHARGE", "+2"
-        config "INTERMEDIATE", "",  &c.string
-        config "SHOWALLMODS", "",  &c.string
-        config "PRECURSOR", "",  &c.string
-        config "USERNAME", "Name",  &c.string
         config "TOLU", "ppm",  &c.string
-        config "USEREMAIL", '', &c.string
         config "CLE", "Trypsin", &c.string
-        config "TOL", 100, &c.num
+        config "TOL", 100, &c.numeric
         config "ITOLU", "Da", &c.string
-        config "QUANTITATION", "None", &c.string
-        config "SEARCH", "MIS", &c.string
+        config "PFA", 1, &MASCOT_SWITCH
         config "DB", "SwissProt", &c.string
-        config "PEP_ISOTOPE_ERROR", 0, &c.num
         config "ITOL", 0.6, &c.float
-        config "FORMVER", 1.01, &c.float
         config "IT_MODS", [
           "Acetyl (Protein N-term)",
           "Gln->pyro-Glu (N-term Q)",
-          "Oxidation (M)"], &c.list
+          "Oxidation (M)"
+        ], &c.list
         config "MASS", "Monoisotopic", &c.string
         config "REPORT", "AUTO", &c.string
         config "TAXONOMY", ". . . . . . . . . . . . . . . . Homo sapiens (human)", &c.string
+        config "INTERMEDIATE", "",  &c.string
+        config "PRECURSOR", "",  &c.string
+        config "QUANTITATION", "None", &c.string
+        config "PEP_ISOTOPE_ERROR", 0, &c.numeric
+        config "SEARCH", "MIS", :type => :hidden
+        config "PEAK", "AUTO", :type => :hidden
+        config "SHOWALLMODS", "", :type => :hidden
+        config "ErrTolRepeat", 0, :type => :hidden
+        config "REPTYPE", "peptide", :type => :hidden
+        config "FORMVER", 1.01, :type => :hidden
       end
       
       def process(mgf_file)

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: ms-mascot
 version: !ruby/object:Gem::Version 
-  version: 0.2.2
+  version: 0.3.0
 platform: ruby
 authors: 
 - Simon Chiang
@@ -10,31 +10,31 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2009-03-31 00:00:00 -06:00
+date: 2009-05-25 00:00:00 -06:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
-  name: tap
+  name: tap-mechanize
   type: :runtime
   version_requirement: 
   version_requirements: !ruby/object:Gem::Requirement 
     requirements: 
     - - ">="
       - !ruby/object:Gem::Version 
-        version: 0.12.4
+        version: 0.6.0
     version: 
 - !ruby/object:Gem::Dependency 
-  name: tap-mechanize
+  name: external
   type: :runtime
   version_requirement: 
   version_requirements: !ruby/object:Gem::Requirement 
     requirements: 
     - - ">="
       - !ruby/object:Gem::Version 
-        version: 0.5.1
+        version: 0.3.0
     version: 
 - !ruby/object:Gem::Dependency 
-  name: external
+  name: ms-in_silico
   type: :runtime
   version_requirement: 
   version_requirements: !ruby/object:Gem::Requirement 
@@ -44,14 +44,24 @@ dependencies:
         version: 0.3.0
     version: 
 - !ruby/object:Gem::Dependency 
-  name: ms-in_silico
-  type: :runtime
+  name: minitest
+  type: :development
+  version_requirement: 
+  version_requirements: !ruby/object:Gem::Requirement 
+    requirements: 
+    - - ">="
+      - !ruby/object:Gem::Version 
+        version: 1.3.1
+    version: 
+- !ruby/object:Gem::Dependency 
+  name: tap-test
+  type: :development
   version_requirement: 
   version_requirements: !ruby/object:Gem::Requirement 
     requirements: 
     - - ">="
       - !ruby/object:Gem::Version 
-        version: 0.2.3
+        version: 0.1.0
     version: 
 description: 
 email: simon.a.chiang@gmail.com
@@ -77,9 +87,11 @@ files:
 - lib/ms/mascot/dat/section.rb
 - lib/ms/mascot/dat/summary.rb
 - lib/ms/mascot/dat/summary/id.rb
+- lib/ms/mascot/dump/mgf.rb
 - lib/ms/mascot/export.rb
-- lib/ms/mascot/format_mgf.rb
+- lib/ms/mascot/formats/xml.rb
 - lib/ms/mascot/fragment.rb
+- lib/ms/mascot/load/mgf.rb
 - lib/ms/mascot/mgf.rb
 - lib/ms/mascot/mgf/archive.rb
 - lib/ms/mascot/mgf/entry.rb

data/lib/ms/mascot/format_mgf.rb

@@ -1,53 +0,0 @@
-require 'ms/mascot/mgf/entry'
-
-module Ms
-  module Mascot
-    # :startdoc::manifest formats an fragment spectrum as mgf
-    #
-    # Formats the data produced by an Ms::Mascot::Fragment task as an mgf. The
-    # configurations specify various details of the dump, including the
-    # precision and default headers.
-    #
-    #   % tap run -- fragment TVQQEL --:s format_mgf
-    #
-    # (note: be sure to use the splat option on the join)
-    #
-    class FormatMgf < Tap::Task
-      
-      config :default_headers, {}, &c.hash        # A hash of default headers
-      config :mz_precision, 6, &c.integer         # The precision of mzs
-      config :intensity_precision, 0, &c.integer  # The precision of intensities
-      config :pepmass_precision, 6, &c.integer    # The precision of peptide mass
-      
-      config :prefix, nil, &c.string_or_nil       # An optional prefix
-      config :suffix, "\n", &c.string_or_nil      # An optional suffix
-      
-      # Maps header keys (typically output by a fragment task)
-      # to Mgf::Entry header strings.
-      HEADER_MAP = {:parent_ion_mass => 'PEPMASS'}
-      
-      def process(data, headers)
-        lines = []
-        lines << prefix if prefix
-        
-        mgf_headers = format_headers(headers)
-        Ms::Mascot::Mgf::Entry.new(mgf_headers, data).dump(lines, config)
-        
-        lines << suffix if suffix
-        lines.join("")
-      end
-      
-      protected
-      
-      # helper to format the headers properly for an mgf entry
-      def format_headers(headers) # :nodoc:
-        mgf_headers = {}
-        default_headers.merge(headers).each_pair do |key, value|
-          key = HEADER_MAP[key] || key.to_s.upcase
-          mgf_headers[key] = value
-        end
-        mgf_headers
-      end
-    end
-  end
-end