mspire 0.7.5 → 0.7.6

data/VERSION

@@ -1 +1 @@
-0.7.5
+0.7.6

data/lib/cv/param.rb

@@ -1,37 +1,30 @@
-=begin
-# if you want to use Nokogiri as the builder, you need something like this
-# code:
-class XML::Nokogiri::Builder
-  def tag!(name, *data)
-    send(name, *data)
-  end
-end
-=end
-
 
 module CV
-  # the xml writer is written with the assumption that the object is a
-  # Builder::XmlMarkup object.  You can get away with using Nokogiri
-  class Param
 
-    attr_accessor :cv_ref, :accession, :name, :value
+  Param = Struct.new(:cv_ref, :accession, :name, :value, :unit)
 
-    # A valueless CV::Param object that describes the units being used
-    attr_accessor :unit
+  class Param
 
-    def initialize(cv_ref, accession, name, value=nil, unit=nil)
-      @cv_ref, @accession, @name, @value, @unit = cv_ref, accession, name, value, unit
+    # standard struct invocation.  Ensures that value is nil if an empty
+    # string is given.
+    def initialize(*args)
+      args[3] = nil if (args[3] == '')
+      super(*args)
     end
 
     # for now, assumes this is a Builder::XmlMarkup object.  
     # returns the xml builder object
     def to_xml(xml, name=:cvParam)
-      hash_to_send = {:cvRef => @cv_ref, :accession => @accession, :name => @name}
-      hash_to_send[:value] = @value if @value
+      hash_to_send = {:cvRef => self.cv_ref, :accession => self.accession, :name => self.name}
+      if v=self.value
+        hash_to_send[:value] = v
+      end
       if unit
-        hash_to_send.merge!( { :unitCvRef => unit.cv_ref, 
+        hash_to_send.merge!( 
+                            { :unitCvRef => unit.cv_ref, 
                             :unitAccession => unit.accession,
-                            :unitName => unit.name } )
+                            :unitName => unit.name } 
+                           )
       end
 
       # xml.send for builder results in tags with 'send' in the front
@@ -42,13 +35,26 @@ module CV
     end
 
     def ==(other)
-      if !other.nil? && other.is_a?(CV::Param)
+      if !other.nil?
         [:cv_ref, :accession, :name, :value, :unit].inject(true) do |bool, mthd|
           bool && (self.send(mthd) == other.send(mthd))
         end
-      else ; false
+      else 
+        false
       end
     end
   end
 end
 
+
+
+=begin
+# if you want to use Nokogiri as the builder, you need something like this
+# code:
+class XML::Nokogiri::Builder
+  def tag!(name, *data)
+    send(name, *data)
+  end
+end
+=end
+

data/lib/mspire/bin.rb

@@ -0,0 +1,67 @@
+
+module Mspire
+  class Bin < Range
+    attr_accessor :data
+
+    def initialize(*args)
+      super(*args)
+      @data = []
+    end
+
+    def inspect
+      "<(" + super + ") @data=#{data.inspect}>"
+    end
+
+    def <<(val)
+      @data << val
+    end
+
+    # O(m + n) speed to bin objects.
+    # bin objects must respond to === .
+    # the object to bin must be a value that is sortable (< > ==), or you can
+    # pass in a block to get the value.
+    # bins and objects must be accessible by index (e.g., bins[11]).
+    # if data_capture is given, it should be a parallel array to bins, and each
+    # object should respond to the '<<' method.  Otherwise, the bins themselves
+    # will be used to push data onto.
+    #
+    # Here's a simple example of binning x,y points where we want to bin the
+    # points based on the x value:
+    #
+    #     bins = (0...10).map {|i| Mspire::Bin.new(i, i+1, false) }
+    #     points = [[2.2, 100], [3.5, 200], [8.8, 150]]
+    #
+    #     Mspire::Bin.bin(bins, points) {|point| point.first }
+    #     # --or--:     Mspire::Bin.bin(bins, points, &:first)
+    #
+    # An example where we want to use a separate data store:
+    #
+    #
+    def self.bin(bins, objects, *data_capture_obj, &block)
+      obj_e = objects.each ; obj = obj_e.next  
+
+      data_capture = data_capture_obj.first || bins
+
+      bin_i = 0  # the bin index
+      cbin = bins[bin_i]  # the current bin
+      done = false
+      until done
+        value = (block.nil? ? obj : block.call(obj))
+        if cbin.begin <= value
+          until cbin === value && data_capture[bin_i] << obj
+            bin_i += 1
+            cbin=bins[bin_i] || (done=true && break)
+          end
+          obj=obj_e.next rescue done=true
+        else
+          while cbin.begin > value && !done
+            obj=obj_e.next rescue done=true && break
+            value = (block.nil? ? obj : block.call(obj))
+          end
+        end
+      end
+      data_capture
+    end
+
+  end
+end

data/lib/mspire/cv/param.rb

@@ -5,7 +5,7 @@ module Mspire
   module CV
 
     # a mass spec related CVParam.  
-    class Param < ::CV::Param
+    module Param
 
       # Takes one of these invocations:
       #
@@ -15,6 +15,8 @@ module Mspire
       # Where acc_num and unit_acc_num are strings containing valid accession
       # numbers (e.g., 'MS:1000514' or 'UO:0000108').  Note that sometimes units are
       # from obo's other than UO.
+      #
+      # returns a CV::Param object
       def self.[](*args)
         #puts "param args #{args.inspect}"
         unit = 
@@ -27,7 +29,7 @@ module Mspire
             Mspire::CV::Param[args.pop]
           end
         obo_type = args[0][/([A-Za-z]+):/,1]
-        self.new(obo_type, args[0], Mspire::CV::Obo[obo_type][args.first], args[1], unit)
+        ::CV::Param.new(obo_type, args[0], Mspire::CV::Obo[obo_type][args.first], args[1], unit)
       end
     end
   end

data/lib/mspire/cv/paramable.rb

@@ -1,4 +1,7 @@
+require 'cv/param'
+require 'mspire/user_param'
 require 'mspire/cv/param'
+require 'nokogiri'
 
 module Mspire
   module CV
@@ -7,26 +10,78 @@ module Mspire
       attr_accessor :params
 
       def initialize(opts={params: []})
-        describe!(*opts[:params])
+        describe_many!(opts[:params])
       end
 
-      # casts each string or array as a Param object (using Mspire::CV::Param[]),
-      # pushes it onto the params attribute and returns the growing params object
-      def describe!(*args)
-        @params ||= []
-        as_params = args.map do |arg|
+      # cast may be something like :to_i or :to_f
+      def find_param_value_by_accession(accession, cast=nil)
+        param = params.find {|v| v.accession == accession }
+        if param
+          val = param.value
+          cast ? (val && val.send(cast)) : val
+        end
+      end
+
+      # cast may be something like :to_i or :to_f
+      def find_param_by_accession(accession)
+        params.find {|v| v.accession == accession }
+      end
+
+
+      def param_exists_by_accession?(accession)
+        params.any? {|v| v.accession == accession }
+      end
+
+      # takes an array of values, each of which is fed into describe!
+      def describe_many!(array)
+        array.each do |arg|
           if arg.is_a?(Array)
-            Mspire::CV::Param[ *arg ]
-          elsif arg.is_a?(String)
-            Mspire::CV::Param[ arg ]
+            describe!(*arg)
           else
-            arg
+            describe!(arg)
+          end
+        end
+      end
+
+      # Expects arguments describing a single CV::Param or Mapire::UserParam.
+      # Will also accept an Nokogiri::XML::Node or Nokogiri::XML::NodeSet
+      #
+      #     obj.describe! 'MS:1000130'  # a positive scan
+      #     obj.describe! CV::Param['MS:1000130']  # same behavior
+      #
+      #     # base peak intensity, units=number of counts
+      #     obj.describe! "MS:1000505", 1524.5865478515625, 'MS:1000131'
+      #
+      #     # given an XML::NodeSet
+      #     obj.describe! xml_node.xpath('.//cvParam')
+      #
+      #     # given an XML
+      #     obj.describe! xml_node.xpath('.//cvParam').first
+      def describe!(*args)
+        @params ||= []
+        case (arg=args.first)
+        when String
+          @params << Mspire::CV::Param[ *args ]
+        when Nokogiri::XML::Node  # a nokogiri node in particular
+          param = 
+            case arg.name
+            when 'cvParam'
+              ::CV::Param.new(arg[:cvRef], arg[:accession], arg[:name], arg[:value])
+            when 'userParam'
+              Mspire::UserParam.new(arg[:name], arg[:value], arg[:type])
+            end
+          if (unit_acc = arg[:unitAccession])
+            param.unit = ::CV::Param.new(arg[:unitCvRef], unit_acc, arg[:unitName])
           end
+          @params << param
+        when Nokogiri::XML::NodeSet
+          arg.each {|node| describe!(node) }
+        else
+          (@params << arg) if arg
         end
-        @params.push(*as_params)
+        @params
       end
 
-      # if params respond_to?(:to_xml) then will call that, otherwise
       # iterates over @params and calls .to_xml on each object.
       def to_xml(xml)
         if @params

data/lib/mspire/isotope/distribution.rb

@@ -104,7 +104,6 @@ module Mspire
 
   end
 
-
   class Isotope
     module Distribution
       def self.calculate(molecular_formula_like, normalize=Mspire::Isotope::Distribution::NORMALIZE, percent_cutoff=nil)

data/lib/mspire/mzml.rb

@@ -152,7 +152,7 @@ module Mspire
         @io = arg
         @encoding = @io.bookmark(true) {|io| io.readline.match(/encoding=["'](.*?)["']/)[1] }
         @index_list = get_index_list
-        # TODO: and read in 'header' info (everything until 'run'
+        read_header!
       when Hash
         arg.each {|k,v| self.send("#{k}=", v) }
       end
@@ -161,6 +161,49 @@ module Mspire
       end
     end
 
+    def read_header!
+      @io.rewind
+      chunk_size = 2**12
+      loc = 0
+      string = ''
+      while chunk = @io.read(chunk_size)
+        string << chunk
+        start_looking = ((loc-20) < 0) ? 0 : (loc-20)
+        break if string[start_looking..-1] =~ /<(spectrum|chromatogram)/
+        loc += chunk_size
+      end
+      doc = Nokogiri::XML.parse(string, nil, @encoding, Parser::NOBLANKS)
+      mzml_n = doc.root
+      if mzml_n.name == 'indexedmzML'
+        mzml_n = mzml_n.child
+      end
+      cv_list_n = mzml_n.child
+      file_description_n = cv_list_n.next
+      self.cvs = cv_list_n.children.map do |cv_n|
+        Mspire::Mzml::CV.from_xml(cv_n)
+      end
+      self.file_description = Mspire::Mzml::FileDescription.from_xml(file_description_n)
+      next_n = file_description_n.next
+      loop do
+        case next_n.name
+        when 'referenceableParamGroupList'
+          # get a hash ready
+        when 'sampleList'
+          # set objects
+        when 'softwareList'  # required
+          # set objects
+        when 'instrumentConfigurationList'
+          # set objects
+        when 'dataProcessingList'
+          # set objects
+        when 'run'
+          # get defaults ready
+          break
+        end
+        next_n = next_n.next
+      end
+    end
+
     class << self
 
       # read-only right now
@@ -212,6 +255,11 @@ module Mspire
 
     alias_method :each, :each_spectrum
 
+    # returns the nokogiri xml node for the spectrum at that index
+    def spectrum_node(index)
+      spectrum_node_from_start_byte(@index_list[:spectrum][index])
+    end
+
     def spectrum_node_from_start_byte(start_byte)
       xml = get_xml_string(start_byte, :spectrum)
       doc = Nokogiri::XML.parse(xml, nil, @encoding, Parser::NOBLANKS)

data/lib/mspire/mzml/activation.rb

@@ -28,6 +28,11 @@ module Mspire
     #     et al.
     class Activation
       include Mspire::CV::Paramable
+      def self.from_xml(xml)
+        obj = self.new
+        [:cvParam, :userParam].each {|v| obj.describe! xml.xpath("./#{v}") }
+        obj
+      end
     end
   end
 end

data/lib/mspire/mzml/cv.rb

@@ -26,12 +26,17 @@ module Mspire
       end
 
       def self.list_xml(objs, builder)
+        # we don't extend Mzml::List because of custom name below
         builder.cvList(count: objs.size) do |cvl_n|
           objs.each {|obj| obj.to_xml(cvl_n) }
         end
         builder
       end
 
+      def self.from_xml(xml)
+        self.new(xml[:id], xml[:fullName], xml[:URI], xml[:version])
+      end
+
       # These are derived by looking in the obo folder at the top of mspire
       IMS = self.new("IMS",  "Imaging MS Ontology", "http://www.maldi-msi.org/download/imzml/imagingMS.obo", "0.9.1")
       MS = self.new('MS', "Proteomics Standards Initiative Mass Spectrometry Ontology", "http://psidev.cvs.sourceforge.net/*checkout*/psidev/psi/psi-ms/mzML/controlledVocabulary/psi-ms.obo", "3.18.0")

data/lib/mspire/mzml/data_array_container_like.rb

@@ -28,7 +28,7 @@ module Mspire
 
       def initialize(id, opts={params: []})
         @id = id
-        describe!(*opts[:params])
+        describe_many!(opts[:params])
       end
 
       def default_array_length

data/lib/mspire/mzml/file_description.rb

@@ -24,7 +24,11 @@ module Mspire
       def initialize(file_content=nil, source_files=[], contacts=[], &block)
         @file_content, @source_files, @contacts = file_content, source_files, contacts
         block.call(self) if block
-        raise ArgumentError, "FileDescription must have file_content" unless @file_content
+        #raise ArgumentError, "FileDescription must have file_content" unless @file_content
+      end
+
+      def self.from_xml(xml)
+        self.new
       end
 
       def to_xml(builder)

data/lib/mspire/mzml/instrument_configuration.rb

@@ -17,7 +17,7 @@ module Mspire
       attr_accessor :software
 
       def initialize(id, components=[], opts={params: []})
-        describe!(*opts[:params])
+        describe_many!(opts[:params])
         @id = id
         @components = components
       end

data/lib/mspire/mzml/isolation_window.rb

@@ -16,6 +16,11 @@ module Mspire
     #     e.g.: MS:1000829 (isolation window upper offset)
     class IsolationWindow
       include Mspire::CV::Paramable
+      def self.from_xml(xml)
+        obj = self.new
+        [:cvParam, :userParam].each {|v| obj.describe! xml.xpath("./#{v}") }
+        obj
+      end
     end
   end
 end

data/lib/mspire/mzml/precursor.rb

@@ -1,5 +1,7 @@
 require 'mspire/mzml/list'
 require 'mspire/mzml/selected_ion'
+require 'mspire/mzml/isolation_window'
+require 'mspire/mzml/activation'
 
 module Mspire
   class Mzml
@@ -25,6 +27,19 @@ module Mspire
         @spectrum=spectrum_derived_from
       end
 
+      def self.from_xml(xml)
+        obj = self.new
+        %w(isolationWindow activation).each do |el|
+          sub_node = xml.xpath("./#{el}").first
+          el[0] = el[0].capitalize
+          Mspire::Mzml.const_get(el).from_xml(sub_node) if sub_node
+        end
+        obj.selected_ions = xml.xpath('./selectedIonList/selectedIon').map do |si_n|
+          Mspire::Mzml::SelectedIon.from_xml(si_n)
+        end
+        obj
+      end
+
       def to_xml(builder)
         atts = {}
         if @from_external_source_file

data/lib/mspire/mzml/processing_method.rb

@@ -9,7 +9,7 @@ module Mspire
 
       def initialize(order, software, opts={params: []}, &block)
         @order, @software = order, software
-        describe!(*opts[:params])
+        describe_many!(opts[:params])
         block.call(self) if block
       end
 

data/lib/mspire/mzml/referenceable_param_group.rb

@@ -14,7 +14,7 @@ module Mspire
 
       def initialize(id, opts={params: []} )
         @id = id
-        describe!(*opts[:params])
+        describe_many!(opts[:params])
       end
 
       def to_xml(builder)

data/lib/mspire/mzml/run.rb

@@ -30,7 +30,7 @@ module Mspire
       def initialize(id, default_instrument_configuration, opts={params: []}, &block)
         @id = id
         @default_instrument_configuration = default_instrument_configuration
-        describe!(*opts[:params])
+        describe_many!(opts[:params])
         block.call(self) if block
       end
 

data/lib/mspire/mzml/sample.rb

@@ -10,7 +10,7 @@ module Mspire
 
       def initialize(id, name, opts={params: []}, &block)
         @id, @name = id, name
-        describe!(*opts[:params])
+        describe_many!(opts[:params])
         block.call(self) if block
       end
 

data/lib/mspire/mzml/scan.rb

@@ -1,4 +1,5 @@
 require 'mspire/cv/paramable'
+require 'mspire/mzml/scan_window'
 
 module Mspire
   class Mzml
@@ -20,7 +21,7 @@ module Mspire
       attr_accessor :scan_windows
 
       def initialize(opts={params: []}, &block)
-        describe!(*opts[:params])
+        describe_many!(opts[:params])
         block.call(self) if block
       end
 
@@ -28,6 +29,15 @@ module Mspire
       #def self.from_xml(xml)
       #end
 
+      def self.from_xml(xml)
+        scan = self.new
+        [:cvParam, :userParam].each {|v| scan.describe! xml.xpath("./#{v}") }
+        scan.scan_windows = xml.xpath('./scanWindowList/scanWindow').map do |scan_window_n|
+          Mspire::Mzml::ScanWindow.from_xml(scan_window_n)
+        end
+        scan
+      end
+
       def to_xml(builder)
         atts = {}
         if @from_external_source_file

data/lib/mspire/mzml/scan_list.rb

@@ -13,7 +13,7 @@ module Mspire
       include Mspire::CV::Paramable
 
       def initialize(opts={params: []}, &block)
-        describe!(*opts[:params])
+        describe_many!(opts[:params])
         block.call(self) if block
       end
 

data/lib/mspire/mzml/scan_settings.rb

@@ -10,7 +10,7 @@ module Mspire
 
       def initialize(id, opts={params: []}, &block)
         @id = id
-        describe!(*opts[:params])
+        describe_many!(opts[:params])
         block.call(self) if block
       end
 

data/lib/mspire/mzml/scan_window.rb

@@ -0,0 +1,19 @@
+require 'mspire/cv/paramable'
+
+module Mspire
+  class Mzml
+    # Typical params might be like:
+    #
+    #     accession="MS:1000501" name="scan window lower limit" value="400"
+    #     accession="MS:1000500" name="scan window upper limit" value="1800"
+    class ScanWindow
+      include Mspire::CV::Paramable
+
+      def self.from_xml(xml)
+        obj = self.new
+        [:cvParam, :userParam].each {|v| obj.describe! xml.xpath("./#{v}") }
+        obj
+      end
+    end
+  end
+end

data/lib/mspire/mzml/selected_ion.rb

@@ -13,6 +13,12 @@ module Mspire
     class SelectedIon
       include Mspire::CV::Paramable
       extend(Mspire::Mzml::List)
+      
+      def self.from_xml(xml)
+        obj = self.new
+        [:cvParam, :userParam].each {|v| obj.describe! xml.xpath("./#{v}") }
+        obj
+      end
     end
   end
 end

data/lib/mspire/mzml/software.rb

@@ -11,7 +11,7 @@ module Mspire
 
       def initialize(id='mspire', version=Mspire::VERSION, opts={params: []}, &block)
         @id, @version = id, version
-        describe!(*opts[:params])
+        describe_many!(opts[:params])
         block.call(self) if block
       end
 

data/lib/mspire/mzml/source_file.rb

@@ -33,7 +33,7 @@ module Mspire
 
       def initialize(id="sourcefile1", name="mspire-simulated", location='file://', opts={params: []}, &block)
         @id, @name, @location = id, name, location
-        describe!(*opts[:params])
+        describe_many!(opts[:params])
         block.call(self) if block
       end
 

data/lib/mspire/mzml/spectrum.rb

@@ -5,6 +5,8 @@ require 'mspire/mzml/scan_list'
 require 'mspire/mzml/precursor'
 require 'mspire/mzml/product'
 
+require 'andand'
+
 module Mspire
   class Mzml
 
@@ -64,62 +66,73 @@ module Mspire
       # currently being described, ordered.
       attr_accessor :products
 
-      # retention time in seconds
-      attr_accessor :retention_time
-      # when properly implemented, this will access the first scan and the
-      # 'scan start time' cv element.
-
-      # takes a Nokogiri node and sets relevant properties
-      def self.from_xml(xml)
-        spec = Mspire::Mzml::Spectrum.new(xml[:id])
-
-        params = {}
-        xml.xpath("./cvParam").each do |cvparam|
-          params[cvparam[:accession]] = cvparam[:value]
-        end
-        spec.ms_level = params['MS:1000511'].to_i
-        # we assume centroid if they don't tell us profile
-        spec.centroided = !params.key?("MS:1000128") # profile spectrum 
-        # centroid -> "MS:1000127"
-
-        # this is a quick hack to get retention time, implement fully as shown
-        # below!
-        cv_param = xml.xpath("./scanList/scan/cvParam[@accession='MS:1000016']").first
-        if cv_param
-          retention_time = cv_param['value'].to_f
-          units = cv_param['unitAccession']
+      # returns the retention time of the first scan object in the scan list
+      # *in seconds*!
+      def retention_time
+        rt_param = scan_list.first.find_param_by_accession('MS:1000016')
+        if rt_param
           multiplier = 
-            case units
+            case rt_param.unit.accession
             when 'UO:0000010' ; 1  # second
             when 'UO:0000031' ; 60 # minute
             when 'UO:0000032' ; 3600 # hour
             when 'UO:0000028' ; 0.001 # millisecond
             else raise 'unsupported units'
             end
-          retention_time *= multiplier
+          rt_param.value.to_f * multiplier
         end
+      end
+
+      # returns the ms_level as an Integer
+      def ms_level
+        find_param_value_by_accession('MS:1000511', :to_i)
+      end
+
+      def centroided?
+        param_exists_by_accession?('MS:1000127')
+      end
+
+      def profile?
+        param_exists_by_accession?('MS:1000128')
+      end
+
+      # returns the charge state of the first precursor as an integer
+      def precursor_charge
+        precursors.andand.first.andand.selected_ions.andand.first.andand.find_param_value_by_accession('MS:1000041', :to_i)
+      end
 
-        # this is roughly how the scan list stuff should be implemented:
-=begin
-        sl_obj = Mspire::Mzml::ScanList.new
+      def precursor_mz
+        precursors.andand.first.andand.selected_ions.andand.first.andand.find_param_value_by_accession('MS:1000744', :to_f)
+      end
 
-        # TODO: need to slot in all the other info in reasonable ways
-        # TODO: need to make sure we deal with referencable params
-        scan_list = xml.xpath('.scanList/scan').each do |scan_n|
-          sl_obj << Mspire::Mzml::Scan.from_xml(scan_n)
+      # takes a Nokogiri node and sets relevant properties
+      def self.from_xml(xml)
+        spec = Mspire::Mzml::Spectrum.new(xml[:id])
+
+        spec.spot_id = xml[:spotID]
+
+        [:cvParam, :userParam].each {|v| spec.describe! xml.xpath("./#{v}") }
+
+        scan_list = Mspire::Mzml::ScanList.new
+        xml.xpath('./scanList/scan').each do |scan_n|
+          scan_list << Mspire::Mzml::Scan.from_xml(scan_n)
+        end
+        spec.scan_list = scan_list
+
+        spec.precursors = xml.xpath('./precursorList/precursor').map do |prec_n|
+          Mspire::Mzml::Precursor.from_xml(prec_n)
         end
-=end
 
         data_arrays = xml.xpath('./binaryDataArrayList/binaryDataArray').map do |binary_data_array_n|
           accessions = binary_data_array_n.xpath('./cvParam').map {|node| node['accession'] }
           base64 = binary_data_array_n.xpath('./binary').text
           Mspire::Mzml::DataArray.from_binary(base64, accessions)
         end
+
         # if there is no spectrum, we will still return a spectrum object, it
         # just has no mzs or intensities
         data_arrays = [Mspire::Mzml::DataArray.new, Mspire::Mzml::DataArray.new] if data_arrays.size == 0
         spec.data_arrays = data_arrays
-        spec.retention_time = retention_time
         spec
       end
 
@@ -131,7 +144,7 @@ module Mspire
       #
       def initialize(id, opts={params: []}, &block)
         @id = id
-        describe! *opts[:params]
+        describe_many! opts[:params]
         block.call(self) if block
       end
 

data/lib/mspire/spectrum.rb

@@ -1,6 +1,6 @@
 require 'mspire/spectrum_like'
 require 'bsearch'
-require 'bin'
+require 'mspire/bin'
 require 'mspire/peak'
 
 module Mspire

data/lib/mspire/user_param.rb

@@ -1,19 +1,15 @@
 
 module Mspire
 
-  class UserParam
-
-    # (optional) a CV::Param object
-    attr_accessor :unit
-
-    # (required)
-    attr_accessor :name
+  UserParam = Struct.new(:name, :value, :type, :unit)
 
-    # (optional)
-    attr_accessor :value
+  class UserParam
 
-    # (optional) e.g. 'xsd:float'
-    attr_accessor :type
+    # returns nil
+    def accession
+      # that way all params can be queried by accession and not raise error
+      nil
+    end
 
     # takes a few different incantations:
     #
@@ -21,23 +17,13 @@ module Mspire
     #     name, value, unit_acc# or CV::Param object
     #     name, value, type, unit_acc# or CV::Param object
     def initialize(*args)
-      @unit = 
+      self.unit = 
         if args.size > 1 && ((args.last.is_a?(::CV::Param) || args.last =~ /^[A-Za-z]+:\d+$/))
           unit_arg = args.pop
           unit_arg.is_a?(::CV::Param) ? unit_arg : Mspire::CV::Param[unit_arg]
         end
-      @name, @value, @type = args
+      self.name, self.value, self.type = args
     end
 
   end
 end
-
-# A UserParam
-# (has no accession)
-# (has no cvRef)
-# name (required)
-# type
-# unitAccession
-# unitCvRef
-# unitName
-# value

data/mspire.gemspec

@@ -5,11 +5,11 @@
 
 Gem::Specification.new do |s|
   s.name = "mspire"
-  s.version = "0.7.5"
+  s.version = "0.7.6"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["John T. Prince", "Simon Chiang"]
-  s.date = "2012-03-23"
+  s.date = "2012-03-27"
   s.description = "mass spectrometry proteomics, lipidomics, and tools, a rewrite of mspire, merging of ms-* gems"
   s.email = "jtprince@gmail.com"
   s.extra_rdoc_files = [
@@ -21,7 +21,6 @@ Gem::Specification.new do |s|
     "README.rdoc",
     "Rakefile",
     "VERSION",
-    "lib/bin.rb",
     "lib/core_ext/array/in_groups.rb",
     "lib/cv.rb",
     "lib/cv/param.rb",
@@ -29,6 +28,7 @@ Gem::Specification.new do |s|
     "lib/io/bookmark.rb",
     "lib/merge.rb",
     "lib/mspire.rb",
+    "lib/mspire/bin.rb",
     "lib/mspire/cv.rb",
     "lib/mspire/cv/param.rb",
     "lib/mspire/cv/paramable.rb",
@@ -90,6 +90,7 @@ Gem::Specification.new do |s|
     "lib/mspire/mzml/scan.rb",
     "lib/mspire/mzml/scan_list.rb",
     "lib/mspire/mzml/scan_settings.rb",
+    "lib/mspire/mzml/scan_window.rb",
     "lib/mspire/mzml/selected_ion.rb",
     "lib/mspire/mzml/software.rb",
     "lib/mspire/mzml/source_file.rb",
@@ -116,7 +117,8 @@ Gem::Specification.new do |s|
     "obo/ms.obo",
     "obo/unit.obo",
     "script/mzml_read_binary.rb",
-    "spec/bin_spec.rb",
+    "spec/cv/param_spec.rb",
+    "spec/mspire/bin_spec.rb",
     "spec/mspire/cv/param_spec.rb",
     "spec/mspire/digester_spec.rb",
     "spec/mspire/error_rate/qvalue_spec.rb",
@@ -138,6 +140,8 @@ Gem::Specification.new do |s|
     "spec/mspire/mzml/index_list_spec.rb",
     "spec/mspire/mzml/plms1_spec.rb",
     "spec/mspire/mzml/referenceable_param_group_spec.rb",
+    "spec/mspire/mzml/source_file_spec.rb",
+    "spec/mspire/mzml/spectrum_spec.rb",
     "spec/mspire/mzml_spec.rb",
     "spec/mspire/peak_spec.rb",
     "spec/mspire/plms1_spec.rb",

data/spec/cv/param_spec.rb

@@ -0,0 +1,21 @@
+
+require 'spec_helper'
+
+require 'cv/param'
+
+describe ::CV::Param do
+  describe 'object creation from class methods' do
+
+    it '::new allows full description' do
+      param1 = ::CV::Param.new('MS', 'MS:1000052', 'suspension')
+      param1.value.should be_nil
+      # just nonsense: 32 ng suspensions
+      param2 = ::CV::Param.new('MS', 'MS:1000052', 'suspension', 32, ::CV::Param.new('UO', 'UO:0000024', 'nanogram'))
+      param2.cv_ref.should == 'MS'
+      param2.value.should == 32
+      param2.unit.accession.should == 'UO:0000024'
+    end
+  end
+end
+
+

data/spec/{bin_spec.rb → mspire/bin_spec.rb}

@@ -1,8 +1,8 @@
 require 'spec_helper'
 
-require 'bin'
+require 'mspire/bin'
 
-describe Bin do
+describe Mspire::Bin do
 
   describe 'putting data into bins' do
 
@@ -18,7 +18,7 @@ describe Bin do
     end
 
     def self.make_ranges(range, use_bin=false)
-      klass = use_bin ? Bin : Range
+      klass = use_bin ? Mspire::Bin : Range
       range.map {|i| klass.new(i.to_f, (i+1).to_f, true) }
     end
 
@@ -27,7 +27,7 @@ describe Bin do
     end
 
     def ranges_to_bins(ranges)
-      ranges.map {|range| Bin.new(range.begin, range.end, true) }
+      ranges.map {|range| Mspire::Bin.new(range.begin, range.end, true) }
     end
 
     data = {
@@ -62,7 +62,7 @@ describe Bin do
 
     data.each do |type, init|
       it "works for bins to data #{type.to_s.gsub('_',' ')}" do
-        rbins = Bin.bin(ranges_to_bins(init[:ranges]), init[:peaks], &:first)
+        rbins = Mspire::Bin.bin(ranges_to_bins(init[:ranges]), init[:peaks], &:first)
         matching(rbins, init[:peaks], init[:bin_to_peak_index_pairs])
       end
     end
@@ -70,7 +70,7 @@ describe Bin do
     data.each do |type, init|
       it "works for ranges to data #{type.to_s.gsub('_',' ')}" do
         custom_data_store = (0...init[:ranges].size).map { [] }
-        rbins = Bin.bin(init[:ranges], init[:peaks], custom_data_store, &:first)
+        rbins = Mspire::Bin.bin(init[:ranges], init[:peaks], custom_data_store, &:first)
         matching(rbins, init[:peaks], init[:bin_to_peak_index_pairs])
       end
     end

data/spec/mspire/cv/param_spec.rb

@@ -4,19 +4,9 @@ require 'mspire/cv/param'
 require 'cv/param'
 
 describe Mspire::CV::Param do
-  describe 'object creation from class methods' do
+  describe 'object creation from class method' do
 
-    it '::new allows full description' do
-      param1 = Mspire::CV::Param.new('MS', 'MS:1000052', 'suspension')
-      param1.value.should be_nil
-      # just nonsense: 32 ng suspensions
-      param2 = Mspire::CV::Param.new('MS', 'MS:1000052', 'suspension', 32, ::CV::Param.new('UO', 'UO:0000024', 'nanogram'))
-      param2.cv_ref.should == 'MS'
-      param2.value.should == 32
-      param2.unit.accession.should == 'UO:0000024'
-    end
-
-    it '::[] requires shortcut accession strings' do
+    it '::[] expects shortcut accession strings' do
       param1 = Mspire::CV::Param['MS:1000052']
       param1.cv_ref.should == 'MS'
       param1.value.should be_nil

data/spec/mspire/mzml/file_content_spec.rb

@@ -10,7 +10,7 @@ describe Mspire::Mzml::FileContent do
 
     desc = filecontent.params
     desc.size.should == 2
-    desc.all? {|par| par.class == Mspire::CV::Param }.should be_true
+    desc.all? {|par| par.class == ::CV::Param }.should be_true
     b = Builder::XmlMarkup.new
     filecontent.to_xml(b)
     xml = b.to_xml
@@ -19,6 +19,7 @@ describe Mspire::Mzml::FileContent do
     end
   end
 
+  it 'can be created from xml'
 
 end
 

data/spec/mspire/mzml/file_description_spec.rb

@@ -3,32 +3,10 @@ require 'mspire/mzml/file_description'
 require 'builder'
 
 describe 'creating mzml xml' do
-  describe 'making fileContent' do
-
-    end
-
-  describe 'making a SourceFile' do
-    it 'can be generated with params and a block' do
-      source_file = Mspire::Mzml::SourceFile.new("someFileID", "filename.mzML", "/home/jtprince/tmp", params: ['MS:1000584'])
-
-      params = source_file.params
-      params.size.should == 1
-      params.all? {|par| par.class == Mspire::CV::Param }.should be_true
-      b = Builder::XmlMarkup.new(:indent => 2)
-      source_file.to_xml(b)
-      xml = b.to_xml
-      [/<sourceFile/, /id="some/, /name="filen/, /location="\/home/, /cvRef="MS"/].each do |regexp|
-        xml.should match(regexp)
-      end
-    end
-
-  end
-
   describe Mspire::Mzml::FileDescription do
 
-    it 'creates valid xml' do
-      #MS::Mzml::FileDescription
+    it 'creates valid xml'
 
-    end
+    it 'can be set from xml'
   end
 end

data/spec/mspire/mzml/source_file_spec.rb

@@ -0,0 +1,24 @@
+require 'spec_helper'
+
+require 'mspire/mzml/source_file'
+require 'builder'
+
+describe 'making a SourceFile' do
+  it 'can be generated with params and a block' do
+    source_file = Mspire::Mzml::SourceFile.new("someFileID", "filename.mzML", "/home/jtprince/tmp", params: ['MS:1000584'])
+
+    params = source_file.params
+    params.size.should == 1
+    params.all? {|par| par.class == ::CV::Param }.should be_true
+    b = Builder::XmlMarkup.new(:indent => 2)
+    source_file.to_xml(b)
+    xml = b.to_xml
+    [/<sourceFile/, /id="some/, /name="filen/, /location="\/home/, /cvRef="MS"/].each do |regexp|
+      xml.should match(regexp)
+    end
+  end
+
+  it 'can be created from xml'
+end
+
+

data/spec/mspire/mzml/spectrum_spec.rb

@@ -0,0 +1,61 @@
+require 'spec_helper'
+
+require 'mspire/mzml'
+require 'mspire/mzml/spectrum'
+
+describe Mspire::Mzml::Spectrum do
+
+  describe 'creating an ms1 spectrum from xml' do
+    before(:all) do
+      @io = File.open(TESTFILES + "/mspire/mzml/j24z.idx_comp.3.mzML")
+      @mzml = Mspire::Mzml.new(@io)
+    end
+
+    after(:all) do
+      @io.close
+    end
+
+    it 'gathers key data for ms1 spectrum' do
+      xml_node = @mzml.spectrum_node(0)
+      spec = Mspire::Mzml::Spectrum.from_xml(xml_node)
+
+      # convenient access to common attributes 
+      spec.retention_time.should == 1981.5726
+      spec.ms_level.should == 1
+
+      spec.precursor_mz.should be_nil
+      spec.precursor_charge.should be_nil
+
+      # array info
+      spec.mzs.size.should == 20168
+      spec.intensities.size.should == 20168
+
+      params = spec.params
+      params.size.should == 9
+      params.first.should be_a(::CV::Param)
+    end
+
+    it 'gathers key data for ms2 spectrum' do
+      xml_node = @mzml.spectrum_node(1)
+      spec = Mspire::Mzml::Spectrum.from_xml(xml_node)
+
+      # convenient access to common attributes 
+      spec.retention_time.should == 1982.1077
+      spec.ms_level.should == 2
+
+      spec.precursor_mz.should == 479.7644958496094
+      spec.precursor_charge.should == 2 
+
+      # array info
+      spec.mzs.size.should == 315
+      spec.intensities.size.should == 315
+
+      params = spec.params
+      params.size.should == 9
+      params.first.should be_a(::CV::Param)
+    end
+
+  end
+
+end
+

data/spec/mspire/mzml_spec.rb

@@ -38,12 +38,6 @@ describe Mspire::Mzml do
         spectrum.should be_a(Mspire::Mzml::Spectrum)
       end
 
-      it 'each spectrum knows its retention_time' do
-        spec = @mzml[1]
-        rt = @mzml[1].retention_time
-        rt.should == 1982.1077
-      end
-
       it 'goes through spectrum with #each or #each_spectrum' do
         mz_sizes = [20168, 315, 634]
         centroided = [false, true, true]
@@ -67,6 +61,14 @@ describe Mspire::Mzml do
         lambda {iter.next}.should raise_error
       end
 
+      # not quite ready for this one yet
+      xit 'contains scans linked to their instrument config objects' do
+        instr_config_first = @mzml.file_description.instrument_configurations[0]
+        instr_config_last = @mzml.file_description.instrument_configurations[1]
+        @mzml[0].scan_list.first.instrument_configuration.should == instr_config_first 
+        @mzml[1].scan_list.first.instrument_configuration.should == instr_config_last
+      end
+
       it 'can gracefully determine the m/z with highest peak in select scans' do
         highest_mzs = Mspire::Mzml.foreach(@file).select {|v| v.ms_level > 1 }.map do |spec|
           spec.points.sort_by(&:last).first.first
@@ -89,7 +91,7 @@ describe Mspire::Mzml do
         spec.scan_list = Mspire::Mzml::ScanList.new do |sl|
           scan = Mspire::Mzml::Scan.new do |scan|
             # retention time of 42 seconds
-            scan.describe! ['MS:1000016', 40.0, 'UO:0000010']
+            scan.describe! 'MS:1000016', 40.0, 'UO:0000010'
           end
           sl << scan
         end
@@ -103,18 +105,18 @@ describe Mspire::Mzml do
         spec.scan_list = Mspire::Mzml::ScanList.new do |sl|
           scan = Mspire::Mzml::Scan.new do |scan|
             # retention time of 42 seconds
-            scan.describe! ['MS:1000016', 45.0, 'UO:0000010']
+            scan.describe! 'MS:1000016', 45.0, 'UO:0000010'
           end
           sl << scan
         end
         precursor = Mspire::Mzml::Precursor.new( spec1 )
         si = Mspire::Mzml::SelectedIon.new
         # the selected ion m/z:
-        si.describe! ["MS:1000744", 2.0]
+        si.describe! "MS:1000744", 2.0
         # the selected ion charge state
-        si.describe! ["MS:1000041", 2]
+        si.describe! "MS:1000041", 2
         # the selected ion intensity
-        si.describe! ["MS:1000042", 5]
+        si.describe! "MS:1000042", 5
         precursor.selected_ions = [si]
         spec.precursors = [precursor]
       end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: mspire
 version: !ruby/object:Gem::Version
-  version: 0.7.5
+  version: 0.7.6
   prerelease: 
 platform: ruby
 authors:
@@ -10,7 +10,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-03-23 00:00:00.000000000 Z
+date: 2012-03-27 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: nokogiri
@@ -185,7 +185,6 @@ files:
 - README.rdoc
 - Rakefile
 - VERSION
-- lib/bin.rb
 - lib/core_ext/array/in_groups.rb
 - lib/cv.rb
 - lib/cv/param.rb
@@ -193,6 +192,7 @@ files:
 - lib/io/bookmark.rb
 - lib/merge.rb
 - lib/mspire.rb
+- lib/mspire/bin.rb
 - lib/mspire/cv.rb
 - lib/mspire/cv/param.rb
 - lib/mspire/cv/paramable.rb
@@ -254,6 +254,7 @@ files:
 - lib/mspire/mzml/scan.rb
 - lib/mspire/mzml/scan_list.rb
 - lib/mspire/mzml/scan_settings.rb
+- lib/mspire/mzml/scan_window.rb
 - lib/mspire/mzml/selected_ion.rb
 - lib/mspire/mzml/software.rb
 - lib/mspire/mzml/source_file.rb
@@ -280,7 +281,8 @@ files:
 - obo/ms.obo
 - obo/unit.obo
 - script/mzml_read_binary.rb
-- spec/bin_spec.rb
+- spec/cv/param_spec.rb
+- spec/mspire/bin_spec.rb
 - spec/mspire/cv/param_spec.rb
 - spec/mspire/digester_spec.rb
 - spec/mspire/error_rate/qvalue_spec.rb
@@ -302,6 +304,8 @@ files:
 - spec/mspire/mzml/index_list_spec.rb
 - spec/mspire/mzml/plms1_spec.rb
 - spec/mspire/mzml/referenceable_param_group_spec.rb
+- spec/mspire/mzml/source_file_spec.rb
+- spec/mspire/mzml/spectrum_spec.rb
 - spec/mspire/mzml_spec.rb
 - spec/mspire/peak_spec.rb
 - spec/mspire/plms1_spec.rb

data/lib/bin.rb

@@ -1,65 +0,0 @@
-
-class Bin < Range
-  attr_accessor :data
-
-  def initialize(*args)
-    super(*args)
-    @data = []
-  end
-
-  def inspect
-    "<(" + super + ") @data=#{data.inspect}>"
-  end
-
-  def <<(val)
-    @data << val
-  end
-
-  # O(m + n) speed to bin objects.
-  # bin objects must respond to === .
-  # the object to bin must be a value that is sortable (< > ==), or you can
-  # pass in a block to get the value.
-  # bins and objects must be accessible by index (e.g., bins[11]).
-  # if data_capture is given, it should be a parallel array to bins, and each
-  # object should respond to the '<<' method.  Otherwise, the bins themselves
-  # will be used to push data onto.
-  #
-  # Here's a simple example of binning x,y points where we want to bin the
-  # points based on the x value:
-  #
-  #     bins = (0...10).map {|i| Bin.new(i, i+1, false) }
-  #     points = [[2.2, 100], [3.5, 200], [8.8, 150]]
-  #
-  #     Bin.bin!(bins, points) {|point| point.first }
-  #     # --or--:     Bin.bin!(bins, points, &:first)
-  #
-  # An example where we want to use a separate data store:
-  #
-  #
-  def self.bin(bins, objects, *data_capture_obj, &block)
-    obj_e = objects.each ; obj = obj_e.next  
-
-    data_capture = data_capture_obj.first || bins
-
-    bin_i = 0  # the bin index
-    cbin = bins[bin_i]  # the current bin
-    done = false
-    until done
-      value = (block.nil? ? obj : block.call(obj))
-      if cbin.begin <= value
-        until cbin === value && data_capture[bin_i] << obj
-          bin_i += 1
-          cbin=bins[bin_i] || (done=true && break)
-        end
-        obj=obj_e.next rescue done=true
-      else
-        while cbin.begin > value && !done
-          obj=obj_e.next rescue done=true && break
-          value = (block.nil? ? obj : block.call(obj))
-        end
-      end
-    end
-    data_capture
-  end
-
-end