mspire 0.7.6 → 0.7.7

data/VERSION

@@ -1 +1 @@
-0.7.6
+0.7.7

data/bin/mzml_to_imzml

@@ -0,0 +1,9 @@
+#!/usr/bin/env ruby
+# encoding: UTF-8
+
+require 'mspire/imzml/writer/commandline'
+
+argv = ARGV.dup
+ARGV.clear
+Mspire::Imzml::Writer::Commandline.run(argv)
+

data/lib/mspire/cv/paramable.rb

@@ -7,29 +7,48 @@ module Mspire
   module CV
     module Paramable
 
-      attr_accessor :params
+      attr_accessor :cv_params
+      attr_accessor :user_params
+      attr_accessor :ref_param_groups
+
+      def params
+        cv_params + ref_param_groups.flat_map(&:params) + user_params 
+      end
+
+      def accessionable_params
+        cv_params + ref_param_groups.flat_map(&:params)
+      end
+
+      def params_by_name
+        params.index_by &:name
+      end
+
+      def params_by_accession
+        accessionable_params.index_by &:accession
+      end
 
       def initialize(opts={params: []})
+        @cv_params = []
+        @user_params = []
+        @ref_param_groups = []
         describe_many!(opts[:params])
       end
 
       # 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 }
+        param = accessionable_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 }
+        accessionable_params.find {|v| v.accession == accession }
       end
 
-
       def param_exists_by_accession?(accession)
-        params.any? {|v| v.accession == accession }
+        accessionable_params.any? {|v| v.accession == accession }
       end
 
       # takes an array of values, each of which is fed into describe!
@@ -57,11 +76,15 @@ module Mspire
       #
       #     # given an XML
       #     obj.describe! xml_node.xpath('.//cvParam').first
+      #
+      # returns self
       def describe!(*args)
-        @params ||= []
+        return self if args.first.nil?
         case (arg=args.first)
         when String
-          @params << Mspire::CV::Param[ *args ]
+          @cv_params << Mspire::CV::Param[ *args ]
+        when Mspire::Mzml::ReferenceableParamGroup
+          @ref_param_groups << arg
         when Nokogiri::XML::Node  # a nokogiri node in particular
           param = 
             case arg.name
@@ -73,25 +96,29 @@ module Mspire
           if (unit_acc = arg[:unitAccession])
             param.unit = ::CV::Param.new(arg[:unitCvRef], unit_acc, arg[:unitName])
           end
-          @params << param
+          @cv_params << param
         when Nokogiri::XML::NodeSet
           arg.each {|node| describe!(node) }
         else
-          (@params << arg) if arg
+          if arg.is_a?(Mspire::UserParam)
+            @user_params << arg
+          else
+            (@cv_params << arg) if arg
+          end
         end
-        @params
+        self
       end
 
-      # iterates over @params and calls .to_xml on each object.
-      def to_xml(xml)
-        if @params
-          @params.each do |el|
-            el.to_xml(xml)
+        # iterates over @params and calls .to_xml on each object.
+        def to_xml(xml)
+          [:ref_param_groups, :cv_params, :user_params].each do |kind|
+            self.send(kind).each do |obj|
+              obj.to_xml(xml)
+            end
           end
+          xml
         end
-        xml
-      end
 
+      end
     end
   end
-end

data/lib/mspire/imzml/writer.rb

@@ -0,0 +1,404 @@
+
+require 'nokogiri'
+require 'uuid'
+require 'cv'
+require 'mspire/mzml'
+require 'mspire/mzml/spectrum'
+require 'pathname'
+require 'digest/sha1'
+
+module Mspire end
+
+module Mspire::Imzml
+  # array length (number of values), the offset (start in file in bytes),
+  # and the encoded length (in bytes).
+  DataArrayInfo = Struct.new :array_length, :offset, :encoded_length
+
+  class Writer
+
+    DEFAULTS = {
+      mz_data_type: :float,
+      mz_precision: 64,
+      intensity_data_type: :float, 
+      intensity_precision: 32,
+    }
+
+    # Integer values (dtype=:int) may be stored with precision = 8, 16, 32,
+    # 64 . Floating point values may be stored as 32 or 64 bit precision.
+    # The byte order is always little endian (intel style).
+    #
+    # NOTE: the documentation is not clear whether they want signed or
+    # unsigned integers!  Right now this outputs signed integers (my
+    # educated guess as to what is required)
+    def write_data_array(fh, values, dtype=:float, precision=32)
+      pack_code = 
+        case dtype
+        when :float
+          precision == 64 ? 'E*' : 'e*'
+        when :int
+          # using signed!!!! (should these be unsigned??? NOT DOCUMENTED in
+          # imzml resources anywhere!)
+          case precision
+          when 8  ; 'c<*'
+          when 16 ; 's<*'
+          when 32 ; 'l<*'
+          when 64 ; 'q<*'
+          else
+            raise 'precision must be 8, 16, 32, or 64 for dtype==:int'
+          end
+        else
+          raise 'dtype must be :int or :float!'
+        end
+      fh.print values.pack(pack_code)
+    end
+
+    # returns an array of DataArrayInfo pairs
+    #
+    # These must be defined in the config hash with valid values (see
+    # write_data_array for what those are):
+    # 
+    #     :mz_data_type
+    #     :mz_precision
+    #     :intensity_data_type
+    #     :intensity_precision
+    #
+    # Also recognizes :data_structure (:processed or :continuous).  If
+    # :continuous, then the first spectrum is used to write the initial m/z
+    # and intensity pair, then in every spectrum after that the m/z data (if
+    # any) is ignored and only the intensities are written).  The return is
+    # the same, the data info for each intensity is coupled with the m/z
+    # info from the first m/z data.
+    def write_binary(filename, spectra_iterator, config)
+      config = DEFAULTS.merge(config)
+      raise ":data_structure must be :continuous or :processed" unless [:processed, :continuous].include?(config[:data_structure])
+      (mz_dtype, mz_prec, int_dtype, int_prec) = config.values_at(:mz_data_type, :mz_precision, :intensity_data_type, :intensity_precision)
+      mz_prec_in_bytes = mz_prec / 8
+      int_prec_in_bytes = int_prec / 8
+      File.open(filename, 'wb') do |out|
+        out.print [config[:uuid]].pack("H*")
+
+        if config[:data_structure] == :continuous
+          # write the first m/z and get its info
+          mzs = spectra_iterator.peek.mzs
+          mz_info = Mspire::Imzml::DataArrayInfo.new mzs.size, out.pos, mzs.size * mz_prec_in_bytes
+          write_data_array(out, mzs, mz_dtype, mz_prec)
+        end
+        spectra_iterator.map do |spec|
+
+          if config[:data_structure] == :processed
+            mzs = spec.mzs
+            mz_info = Mspire::Imzml::DataArrayInfo.new mzs.size, out.pos, mzs.size * mz_prec_in_bytes
+            write_data_array(out, spec.mzs, mz_dtype, mz_prec)
+          end
+
+          ints = spec.intensities
+          int_info = Mspire::Imzml::DataArrayInfo.new ints.size, out.pos, ints.size * int_prec_in_bytes
+          write_data_array(out, spec.intensities, int_dtype, int_prec)
+          [mz_info, int_info]
+        end
+      end
+    end
+
+    # converts image related hash values to an array of loose describe!
+    # params
+    def image_hash_to_cvs(hash)
+      cvs = []
+      cvs << case hash[:data_structure].to_sym
+      when :processed ; 'IMS:1000031'
+      when :continuous ; 'IMS:1000030'
+      else ; raise ":data_structure must be :processed or :continuous"
+      end 
+
+      cvs << case hash[:scan_pattern].to_sym
+      when :meandering ; 'IMS:1000410'
+      when :random ; 'IMS:1000412'
+      when :flyback ; 'IMS:1000413'
+      else ; raise ":scan_pattern must be :meandering, :random or :flyback"
+      end
+
+      cvs << case hash[:scan_type].to_sym
+      when :horizontal ; 'IMS:1000480'
+      when :vertical ; 'IMS:1000481'
+      else ; raise ":scan_type must be :horizontal or :vertical"
+      end
+
+      cvs << case hash[:linescan_direction].to_s
+      when 'left-right' ; 'IMS:1000402'
+      when 'right-left' ; 'IMS:1000403'
+      when 'bottom-up' ; 'IMS:1000400'
+      when 'top-down' ; 'IMS:1000401'
+      when 'none' ; 'IMS:1000404'
+      else ; raise ":linescan_direction unacceptable"
+      end
+
+      cvs << case hash[:linescan_sequence].to_s
+      when 'top-down' ; 'IMS:1000493'
+      when 'bottom-up' ; 'IMS:1000492'
+      when 'left-right' ; 'IMS:1000491'
+      when 'right-left' ; 'IMS:1000490'
+      else ; raise "linescan_sequence unacceptable"
+      end
+
+      max_pix_dims = hash[:max_dimensions_pixels].split(/x/i)
+      cvs.push *['IMS:1000042', 'IMS:1000043'].zip(max_pix_dims).to_a
+
+      real_dims = hash[:max_dimensions_microns].split(/x/i)
+      cvs.push *['IMS:1000044', 'IMS:1000045'].zip(real_dims).map {|pair| [*pair, "UO:0000017"] }
+
+      pix_dims = hash[:pixel_size].split(/x/i)
+      cvs.push *["IMS:1000046", "IMS:1000047"].zip(pix_dims).map {|pair| [*pair, "UO:0000017"] }
+      cvs
+    end
+
+    def experiment_hash_to_cvs(hash)
+      cvs = { 
+        :matrix_solution_concentration => "MS:1000835",
+        :matrix_solution => "MS:1000834",
+        :solvent => 'IMS:1001211',
+        :solvent_flowrate => 'IMS:1001213',
+        :spray_voltage => 'IMS:1001212',
+        :target_material => 'IMS:10000202'
+      }.map do |key,cv_acc|
+        [cv_acc, hash[key]] if hash[key]
+      end.compact
+
+      mats = hash[:matrix_application_types]
+      if mats
+        app_types = mats.map do |typ|
+          case typ.to_s
+          when 'sprayed' ; 'MS:1000838'
+          when 'precoated' ; 'MS:1000839'
+          when 'printed' ; 'MS:1000837'
+          when 'drieddroplet' ; 'MS:1000836'
+          else ; raise "invalid matrix_application_type"
+          end
+        end
+        cvs.push(*app_types)
+      end
+      cvs
+    end
+
+    # returns an array with x, y pairs in the correct order
+    # accounts for scan_pattern, scan_type, linescan_direction,
+    # linescan_sequence and shots_per_position
+    def x_y_positions(config)
+      (scan_pattern, scan_type, scan_direction, scan_sequence, shots_per_pos) = config.values_at(:scan_pattern, :scan_type, :linescan_direction, :linescan_sequence, :shots_per_position).map(&:to_s)
+      shots_per_pos = shots_per_pos.to_i
+      #puts "EXAMIN ARGS: "
+      #p [ scan_pattern, scan_type, scan_direction, scan_sequence, shots_per_pos, shots_per_pos]
+
+      # the true dimensions we need to work off come from the pixel dimensions.
+      (slen, plen) = config[:max_dimensions_pixels].split(/x/i).map(&:to_i)
+
+      flip = (scan_type == 'vertical')
+      plen, slen = slen, plen if flip
+
+      pindices = (1..plen).to_a  # mindices if linescan_direction is 'horizontal'
+      sindices = (1..slen).to_a  # nindices if linescan_direction is 'horizontal'
+
+      if flip
+        sindices.reverse! if scan_direction == 'bottom-top' || scan_direction == 'right-left'
+        pindices.reverse! if scan_sequence == 'right-left' || scan_sequence == 'bottom-top'
+      end
+
+      indices = pindices.map do |a|
+        row = sindices.map do |b|
+          flip ? [a,b] : [b,a]
+        end
+        sindices.reverse! if scan_pattern == 'meandering'
+        row
+      end.flatten(1)
+
+      indices.map {|pair| shots_per_pos.times.map { pair } }.flatten(1)
+    end
+
+    def create_file_description(source_files, config)
+      Mspire::Mzml::FileDescription.new  do |fd|
+
+        fd.file_content = Mspire::Mzml::FileContent.new :params => [
+          'MS:1000579', # MS1 Spectrum
+          config[:profile] ? 'MS:1000128' : 'MS:1000127',
+          ['IMS:1000080', "{"+config[:uuid_hyphenated]+"}"],
+          ['IMS:1000091', config[:ibd_sha1]],
+          (config[:data_structure] == :processed) ? 'IMS:1000031' : 'IMS:1000030',
+        ]
+
+        fd.source_files.replace(source_files)
+        if [:name, :organization, :address, :email].any? {|key| config[key] }
+          contact = Mspire::Mzml::Contact.new
+          [ [:name, 'MS:1000586'], 
+            [:organization, 'MS:1000590'],
+            [:address, 'MS:1000587'],
+            [:email, 'MS:1000589']
+          ].each do |key, accession|
+            contact.describe!(accession, config[key]) if config[key]
+          end
+          fd.contacts << contact
+        end
+      end
+    end
+
+    def create_referenceable_params_by_id_hash
+      # m/z array and no compression (because we'll ref peaks in different
+      # file)
+      rparms = {
+        :mz_array => [
+          ["MS:1000514", "MS:1000040"], # m/z array, units = m/z
+          "MS:1000576",                 # no compression
+          ["IMS:1000101", true],        # external data
+          'MS:1000523'                  # 64-bit float
+      ],
+        :intensity_array =>  [
+          ["MS:1000515", "MS:1000131"], # intensity array, units = number of counts
+          "MS:1000576",                 # no compression
+          ["IMS:1000101", true],        # external data
+          'MS:1000521'                  # 32-bit float
+      ],
+        :scan1 => [
+          # this should probably be ascertained from the mzml file:
+          "MS:1000093", # increasing m/z scan
+          # "MS:1000095" # linear  # <- this should probably be gathered
+          # from mzml (leave outfor now)
+          # could include the filter string here in future
+      ],
+      :spectrum1 => [
+        "MS:1000579", # MS1 spectrum
+        ["MS:1000511", 1], # ms level - default implementation uses 0 but I disagree...
+        "MS:1000127",  # centroid spectrum
+        "MS:1000130" # <- positive scan
+      ]}.map do |id, list|
+        Mspire::Mzml::ReferenceableParamGroup.new id, params: list
+      end
+      Hash[ rparms.map {|parm| [parm.id, parm]} ]
+    end
+
+    # mzml_filenames can each be a partial or relative path
+    # by default the file will write to the same directory and basename
+    #
+    #     contact:
+    #       :name => name of the person or organization
+    #       :address => address of the person or organization
+    #       :url => url of person or organization
+    #       :email => email of person or organization
+    #       :organization => home institution of contact
+    #     experiment:
+    #       :solvent => the solvent used
+    #       :solvent_flowrate => flowrate of the solvent (ml/min)
+    #       :spray_voltage => spray voltage in kV
+    #       :target_material => the material the target is made of
+    #     general:
+    #       :omit_zeros => remove zero values
+    #       :combine => use this outfile base name to combine files
+    #                   also works on single files to rename them.
+    #                   MUST be present if more than one mzml given.
+    #     imaging:
+    #       :data_structure => :processed or :continuous  
+    #       :scan_pattern => meandering random flyback
+    #       :scan_type => horizontal or vertical
+    #       :linescan_direction => scan_direction.join('|')
+    #       :linescan_sequence => scan_sequence.join('|')
+    #       :max_dimensions_pixels => maximum X by Y pixels (e.g. 300x100)
+    #       :shots_per_position => number of spectra per position
+    #       :pixel_size => X by Y of a single pixel in microns (μm)
+    #       :max_dimensions_microns => maximum X by Y in microns (e.g. 25x20)
+    def write(mzml_filenames, config={})
+
+      base = config[:combine] || mzml_filenames.first.chomp(File.extname(mzml_filenames.first))
+      config[:imzml_filename] = base + ".imzML"
+      config[:ibd_filename] = base + ".ibd"
+
+      uuid_with_hyphens = UUID.new.generate
+      config[:uuid_hyphenated] = uuid_with_hyphens
+      config[:uuid] = uuid_with_hyphens.gsub('-','')
+
+      sourcefile_id_parallel_to_spectra = []
+      sourcefile_ids = []
+      all_spectra_iter = Enumerator.new do |yielder|
+        mzml_filenames.each_with_index do |mzml_filename,i|
+          sourcefile_id = "source_file_#{i}"
+          sourcefile_ids << sourcefile_id
+          Mspire::Mzml.open(mzml_filename) do |mzml|
+            mzml.each do |spec| 
+              sourcefile_id_parallel_to_spectra << sourcefile_id 
+              yielder << spec
+            end
+          end
+        end
+      end
+
+      data_info_pairs = write_binary(config[:ibd_filename], all_spectra_iter, config)
+      config[:ibd_sha1] = Digest::SHA1.hexdigest(IO.read(config[:ibd_filename]))
+
+      source_files = mzml_filenames.zip(sourcefile_ids).map do |mzml_filename, source_file_id|
+        sfile = Mspire::Mzml::SourceFile[ mzml_filename ]
+        sfile.id = source_file_id
+        sfile.describe! 'MS:1000584'
+        sfile.describe! 'MS:1000569', Digest::SHA1.hexdigest(IO.read(mzml_filename))
+        sfile
+      end
+      sourcefile_id_to_sourcefile = Hash[ source_files.group_by(&:id).map {|k,v| [k,v.first] } ]
+
+      imzml_obj = Mspire::Mzml.new do |imzml|
+        imzml.id = UUID.new.generate.gsub('-','')
+        imzml.cvs = Mspire::Mzml::CV::DEFAULT_CVS
+
+        imzml.file_description = create_file_description(source_files, config)
+
+        rparms_by_id = create_referenceable_params_by_id_hash
+
+        warn "using positive scan in every case but need to get this from original mzml!"
+
+        imzml.referenceable_param_groups = rparms_by_id.values
+        # skip sample list for now
+        mspire_software = Mspire::Mzml::Software.new( "mspire", Mspire::VERSION, params: ["MS:1000799"] )
+        imzml.software_list << mspire_software
+
+        scan_setting_params = image_hash_to_cvs( config )
+        scan_setting_params.push *experiment_hash_to_cvs( config )
+        imzml.scan_settings_list = [Mspire::Mzml::ScanSettings.new("scansettings1", params: scan_setting_params)]
+
+        warn 'todo: need to borrow instrumentConfiguration from original mzml'
+
+        default_instrument_config = Mspire::Mzml::InstrumentConfiguration.new("borrow_from_mzml")
+        warn 'todo: need to include default softare from mzml in default_instrument_config'
+        #default_instrument_config.software = Software.new( from the mzml file! )
+        imzml.instrument_configurations << default_instrument_config
+
+        # this is a generic 'file format conversion' but its the closest we
+        # have for mzml to imzml (which is really mzml to mzml)
+        data_processing_obj = Mspire::Mzml::DataProcessing.new('mzml_to_imzml')
+        data_processing_obj.processing_methods << Mspire::Mzml::ProcessingMethod.new(1, mspire_software, params: ['MS:1000530'] )
+        imzml.data_processing_list << data_processing_obj
+
+        warn "not implemented 'omit_zeros' yet"
+        # low intensity data point removal: "MS:1000594"
+        imzml.run = Mspire::Mzml::Run.new("run1", default_instrument_config) do |run|
+          spec_list = Mspire::Mzml::SpectrumList.new(data_processing_obj)
+          data_info_pairs.zip(x_y_positions(config), sourcefile_id_parallel_to_spectra).each_with_index do |(pair, xy, sourcefile_id),i|
+            # TODO: we should probably copy the id from the orig mzml (e.g.
+            # scan=1)
+            spectrum = Mspire::Mzml::Spectrum.new("spectrum#{i}", params: [rparms_by_id[:spectrum1]])
+            spectrum.source_file = sourcefile_id_to_sourcefile[sourcefile_id]
+            scan_list = Mspire::Mzml::ScanList.new(params: ['MS:1000795'])  # no combination
+            scan = Mspire::Mzml::Scan.new( params: [rparms_by_id[:scan1], ["IMS:1000050", xy[0]], ["IMS:1000051", xy[1]]] )
+            scan.instrument_configuration = default_instrument_config
+            spectrum.scan_list = (scan_list << scan)
+            
+            data_arrays = %w(mz intensity).zip(pair).map do |type, data_array_info|
+              rparmgroup = rparms_by_id[(type + "_array").to_sym]
+              data_array = Mspire::Mzml::DataArray.new
+              data_array.type = type
+              data_array.external = true
+              data_array.describe_many! [rparmgroup, *%w(IMS:1000103 IMS:1000102 IMS:1000104).zip(data_array_info).map.to_a]
+              data_array
+            end
+            spectrum.data_arrays = data_arrays
+            spec_list << spectrum
+          end
+          run.spectrum_list = spec_list
+        end # run
+      end # imzml
+      imzml_obj.to_xml(config[:imzml_filename])
+    end
+  end
+end

data/lib/mspire/imzml/writer/commandline.rb

@@ -0,0 +1,109 @@
+# encoding: UTF-8
+
+require 'mspire/imzml/writer'
+require 'trollop'
+require 'yaml'
+
+module Mspire
+  module Imzml
+    class Writer
+    end
+  end
+end
+
+module Mspire::Imzml::Writer::Commandline
+
+  def self.run(argv, progname=$0)
+
+    default_config = "config.yaml"
+    scan_patterns = %w(flyback meandering random)
+    scan_types = %w(horizontal vertical)
+    scan_direction = %w(left-right right-left bottom-up top-down none)
+    scan_sequence = %w(top-down bottom-up left-right right-left)
+    default_dims = '800x600'
+    default_pixel_size = '1x1'
+    matrix_application_types = %w(sprayed precoated printed drieddroplet)
+
+    parser = Trollop::Parser.new do
+      banner <<-EOS
+usage: #{File.basename(progname)} [OPTIONS] <file>.mzML ..."
+output: <file>.imzML and <file>.ibd
+
+* imzML docs: 
+    http://www.maldi-msi.org/index.php?option=com_content&view=article&id=187&Itemid=67
+* explanation of vocabulary (followed here): 
+    http://www.maldi-msi.org/index.php?option=com_content&view=article&id=193&Itemid=66
+* github repository:
+    https://github.com/princelab/mzml_to_imzml
+      EOS
+      text "\ngeneral:"
+      opt :config, "read a config file for default values. Command line options overide those from the config file ", :type => :string
+      opt :print_config, "print current options to #{default_config} and exit"
+      opt :omit_zeros, "remove zero values"
+      opt :combine, "combine all files and set the base name of resulting imzML and ibd files", :type => String
+      opt :outfile, "use a specific basename for the resulting file. Acts like --combine for multiple files", :type => String
+
+      text "\nimaging:"
+      opt :continuous, "assumes m/z values are the same for every scan. The 'processed' storage format is used unless this flag is given."
+      opt :scan_pattern, scan_patterns.join('|'), :default => scan_patterns.first
+      opt :scan_type, scan_types.join('|'), :default => scan_types.first
+      opt :linescan_direction, scan_direction.join('|'), :default => scan_direction.first
+      opt :linescan_sequence, scan_sequence.join('|'), :default => scan_sequence.first
+      opt :max_dimensions_pixels, "maximum X by Y pixels (e.g. 300x100)", :default => default_dims
+      opt :shots_per_position, "number of spectra per position", :default => 1
+      opt :pixel_size, "X by Y of a single pixel in microns (μm)", :default => default_pixel_size 
+      opt :max_dimensions_microns, "maximum X by Y in microns (e.g. 25x20)", :default => default_dims
+
+      text "\ncontact: "
+      opt :name, "name of the person or organization", :type => :string
+      opt :address, "address of the person or organization", :type => :string
+      opt :url, "url of person or organization", :type => :string
+      opt :email, "email of person or organization", :type => :string
+      opt :organization, "home institution of contact", :type => :string
+
+      text "\nDESI: "
+      opt :solvent, "the solvent used", :type => :string
+      opt :solvent_flowrate, "flowrate of the solvent (ml/min)", :type => :float
+      opt :spray_voltage, "spray voltage in kV", :type => :float
+      opt :target_material, "the material the target is made of", :type => :string
+
+      text "\nMALDI: "
+      opt :matrix_application_types, "#{matrix_application_types.join('|')} (comma separated)", :type => :string
+      opt :matrix_solution_concentration, "in grams per liter", :type => :float
+      opt :matrix_solution, "the chemical solution used as matrix (e.g., DHB)", :type => :string
+
+      # things to add: data types for m/z and intensity
+      # filters (cutoff / max # peaks, etc.)
+      # ms_level, etc.
+    end
+
+    opts = parser.parse(argv)
+    opts = Hash[YAML.load_file(opts[:config]).map {|k,v| [k.to_sym, v]}].merge(opts) if opts[:config]
+
+    opts[:combine] ||= opts.delete(:outfile)
+
+    if opts.delete(:print_config)
+      puts "writing defaults to: #{default_config}"
+      string_opts = Hash[ opts.map {|k,v| [k.to_s, v] } ]
+      %w(help).each {|key| string_opts.delete key }
+      string_opts.delete_if {|k,v| k =~ /_given$/ }
+      File.write(default_config, string_opts.to_yaml) 
+      exit
+    end
+
+    if argv.size == 0
+      return parser.educate
+    end
+
+    opts[:data_structure] = (opts.delete(:continuous) ? :continuous : :processed)
+    opts[:matrix_application_types] = opts[:matrix_application_types].split(',') if opts[:matrix_application_types]
+
+    # prep args a little
+    writer = Mspire::Imzml::Writer.new
+    writer.write(argv, opts)
+
+  end
+end
+
+
+

data/lib/mspire/mzml/data_array.rb

@@ -7,6 +7,8 @@ module Mspire
     class DataArray < Array
       alias_method :array_initialize, :initialize
       include Mspire::CV::Paramable
+      alias_method :params_initialize, :initialize
+      alias_method :params_to_xml, :to_xml
 
       DEFAULT_DTYPE = :float64
       DEFAULT_COMPRESSION = true
@@ -18,11 +20,35 @@ module Mspire
         int32: 'MS:1000519', # signed
       }
 
-      # the type of data array (:mz, :intensity, :mz_external, or :intensity_external)
-      attr_accessor :type
+      def type=(arg)
+        all_accs = %w(MS:1000514 MS:1000515)
+        params.delete_if {|param| all_accs.include?(param.accession) } if params
+        case arg
+        when :mz
+          describe! all_accs[0] # , nil, "MS:1000040"
+        when :intensity
+          describe! all_accs[1] # , nil, "MS:1000131"
+        end
+        arg
+      end
+
+      def type
+        if params
+          if params.any? {|param| param.accession == 'MS:1000514' }
+            :mz
+          elsif params.any? {|param| param.accession == 'MS:1000515' }
+            :intensity
+          end
+        end
+      end
+
+      # set this if the data is written to an external file (such as the ibd
+      # file for imzML files)
+      attr_accessor :external
 
       def initialize(*args)
         array_initialize(*args)
+        params_initialize
       end
 
       # returns a new Mspire::Mzml::DataArray object (an array)
@@ -72,7 +98,7 @@ module Mspire
       #
       #     args:
       #       array-like  set-like               # where set-like responds to include?
-      #       array-like  type=:float64, compression=true
+      #       array-like  dtype=:float64, compression=true
       def self.to_binary(array_ish, *args)
         if args.first.respond_to?(:include?)
           accessions = args.first
@@ -117,15 +143,11 @@ module Mspire
           end
 
         builder.binaryDataArray(encodedLength: encoded_length) do |bda_n|
-          @params.each {|param| param.to_xml(bda_n) } if @params
-          unless @external
+          params_to_xml(bda_n)
+          unless self.external
             Mspire::CV::Param[ DTYPE_TO_ACC[dtype] ].to_xml(bda_n)
             Mspire::CV::Param[ compression ? 'MS:1000574' : 'MS:1000576' ].to_xml(bda_n)
-            if @type
-              accession = ( (@type == :mz) ? 'MS:1000514' : 'MS:1000515' )
-              Mspire::CV::Param[accession].to_xml(bda_n)
-              bda_n.binary(base64)
-            end
+            bda_n.binary(base64)
           end
         end
       end
@@ -138,9 +160,7 @@ module Mspire
               if data_ar.is_a?(Mspire::Mzml::DataArray)
                 data_ar
               else
-                real_data_array = Mspire::Mzml::DataArray.new(data_ar)
-                real_data_array.type = typ
-                real_data_array
+                Mspire::Mzml::DataArray.new(data_ar)
               end
             ar.type = typ unless ar.type
             ar.to_xml(bdal_n)

data/lib/mspire/mzml/data_array_container_like.rb

@@ -26,11 +26,6 @@ module Mspire
       # (optional) an array of Mspire::Mzml::DataArray
       attr_accessor :data_arrays
 
-      def initialize(id, opts={params: []})
-        @id = id
-        describe_many!(opts[:params])
-      end
-
       def default_array_length
         if @data_arrays
           if @data_arrays.first

data/lib/mspire/mzml/instrument_configuration.rb

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

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_many!(opts[:params])
+        super(opts)
         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_many!(opts[:params])
+        super(opts)
       end
 
       def to_xml(builder)
@@ -24,7 +24,7 @@ module Mspire
 
       def to_xml_definition(builder)
         builder.referenceableParamGroup(id: @id) do |fc_n|
-          @params.each {|obj| obj.to_xml(fc_n) }
+          params.each {|obj| obj.to_xml(fc_n) }
         end
         builder
       end

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_many!(opts[:params])
+        super(opts)
         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_many!(opts[:params])
+        super(opts)
         block.call(self) if block
       end
 

data/lib/mspire/mzml/scan.rb

@@ -21,7 +21,7 @@ module Mspire
       attr_accessor :scan_windows
 
       def initialize(opts={params: []}, &block)
-        describe_many!(opts[:params])
+        super(opts)
         block.call(self) if block
       end
 

data/lib/mspire/mzml/scan_list.rb

@@ -13,13 +13,13 @@ module Mspire
       include Mspire::CV::Paramable
 
       def initialize(opts={params: []}, &block)
-        describe_many!(opts[:params])
+        super(opts)
         block.call(self) if block
       end
 
       def to_xml(builder)
         builder.scanList(count: self.size) do |sl_n|
-          @params.each {|param| param.to_xml(sl_n) } if @params
+          super(sl_n)
           self.each do |scan|
             scan.to_xml(sl_n)
           end

data/lib/mspire/mzml/scan_settings.rb

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

data/lib/mspire/mzml/spectrum.rb

@@ -43,6 +43,7 @@ module Mspire
     class Spectrum
       include Mspire::SpectrumLike
       include Mspire::Mzml::DataArrayContainerLike
+      alias_method :params_initialize, :initialize
 
       # (optional) an Mspire::Mzml::SourceFile object
       attr_accessor :source_file
@@ -144,7 +145,7 @@ module Mspire
       #
       def initialize(id, opts={params: []}, &block)
         @id = id
-        describe_many! opts[:params]
+        params_initialize(opts)
         block.call(self) if block
       end
 

data/mspire.gemspec

@@ -5,13 +5,14 @@
 
 Gem::Specification.new do |s|
   s.name = "mspire"
-  s.version = "0.7.6"
+  s.version = "0.7.7"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["John T. Prince", "Simon Chiang"]
   s.date = "2012-03-27"
   s.description = "mass spectrometry proteomics, lipidomics, and tools, a rewrite of mspire, merging of ms-* gems"
   s.email = "jtprince@gmail.com"
+  s.executables = ["mzml_to_imzml"]
   s.extra_rdoc_files = [
     "LICENSE",
     "README.rdoc"
@@ -21,6 +22,7 @@ Gem::Specification.new do |s|
     "README.rdoc",
     "Rakefile",
     "VERSION",
+    "bin/mzml_to_imzml",
     "lib/core_ext/array/in_groups.rb",
     "lib/cv.rb",
     "lib/cv/param.rb",
@@ -56,6 +58,8 @@ Gem::Specification.new do |s|
     "lib/mspire/ident/protein.rb",
     "lib/mspire/ident/protein_group.rb",
     "lib/mspire/ident/search.rb",
+    "lib/mspire/imzml/writer.rb",
+    "lib/mspire/imzml/writer/commandline.rb",
     "lib/mspire/isotope.rb",
     "lib/mspire/isotope/aa.rb",
     "lib/mspire/isotope/distribution.rb",
@@ -128,6 +132,7 @@ Gem::Specification.new do |s|
     "spec/mspire/ident/pepxml/search_hit/modification_info_spec.rb",
     "spec/mspire/ident/pepxml_spec.rb",
     "spec/mspire/ident/protein_group_spec.rb",
+    "spec/mspire/imzml/writer_spec.rb",
     "spec/mspire/isotope/aa_spec.rb",
     "spec/mspire/isotope/distribution_spec.rb",
     "spec/mspire/isotope_spec.rb",
@@ -151,8 +156,11 @@ Gem::Specification.new do |s|
     "spec/mspire_spec.rb",
     "spec/obo_spec.rb",
     "spec/spec_helper.rb",
+    "spec/testfiles/continuous_binary.tmp.ibd",
     "spec/testfiles/mspire/ident/peptide/db/uni_11_sp_tr.fasta",
     "spec/testfiles/mspire/ident/peptide/db/uni_11_sp_tr.msd_clvg2.min_aaseq4.yml",
+    "spec/testfiles/mspire/imzml/continuous_binary_check.ibd",
+    "spec/testfiles/mspire/imzml/processed_binary_check.ibd",
     "spec/testfiles/mspire/mzml/j24z.idx_comp.3.mzML",
     "spec/testfiles/mspire/mzml/mspire_simulated.MSn.check.mzML",
     "spec/testfiles/mspire/mzml/openms.noidx_nocomp.12.mzML",
@@ -175,7 +183,8 @@ Gem::Specification.new do |s|
     "spec/testfiles/mspire/quant/pdcd5_lfq_tabdel.txt_qspecgp",
     "spec/testfiles/mspire/quant/remove_rest_of_proteins.rb",
     "spec/testfiles/mspire/quant/unlog_transform.rb",
-    "spec/testfiles/plms1/output.key"
+    "spec/testfiles/plms1/output.key",
+    "spec/testfiles/processed_binary.tmp.ibd"
   ]
   s.homepage = "http://github.com/princelab/mspire"
   s.licenses = ["MIT"]

data/spec/mspire/imzml/writer_spec.rb

@@ -0,0 +1,101 @@
+require 'spec_helper'
+
+require 'mspire/imzml/writer'
+require 'mspire/spectrum'
+
+describe Mspire::Imzml::Writer do
+
+  before(:each) do
+    @spectra = [
+      Mspire::Spectrum.new([[1,2,3],[4,5,6]]),
+      Mspire::Spectrum.new([[1,2,3],[4,5,6]]),
+      Mspire::Spectrum.new([[1,2,3],[4,5,6]]),
+    ]
+    @config = {
+      uuid: "d097f8103a8e012f2b130024e8b4cdae",
+      mz_data_type: :float,
+      mz_precision: 32,
+      # just for fun, use floats for m/z and ints for intensities
+      intensity_data_type: :int,
+      intensity_precision: 32,
+    }
+  end
+
+  it 'has an iterator that creates the proper x and y positions' do
+    #config = { :scan_pattern => 'meandering', :scan_type => 'horizontal', :linescan_direction => 'left-right', :linescan_sequence => 'top-down', :max_dimensions_pixels => '3x2', :pixel_size => '2x2', :max_dimensions_microns => '6x4', :shots_per_position => 2 }
+    writer = Mspire::Imzml::Writer.new
+    testing = {
+
+      %w(flyback horizontal left-right top-bottom) => [[1, 1], [1, 1], [2, 1], [2, 1], [3, 1], [3, 1], [1, 2], [1, 2], [2, 2], [2, 2], [3, 2], [3, 2]],
+      %w(flyback horizontal left-right bottom-top) => [[1, 1], [1, 1], [2, 1], [2, 1], [3, 1], [3, 1], [1, 2], [1, 2], [2, 2], [2, 2], [3, 2], [3, 2]],
+      %w(flyback horizontal right-left bottom-top) => [[1, 1], [1, 1], [2, 1], [2, 1], [3, 1], [3, 1], [1, 2], [1, 2], [2, 2], [2, 2], [3, 2], [3, 2]],
+      %w(flyback horizontal right-left top-bottom) => [[1, 1], [1, 1], [2, 1], [2, 1], [3, 1], [3, 1], [1, 2], [1, 2], [2, 2], [2, 2], [3, 2], [3, 2]],
+      %w(meandering horizontal left-right top-bottom) => [[1, 1], [1, 1], [2, 1], [2, 1], [3, 1], [3, 1], [3, 2], [3, 2], [2, 2], [2, 2], [1, 2], [1, 2]],
+      %w(meandering horizontal left-right bottom-top) => [[1, 1], [1, 1], [2, 1], [2, 1], [3, 1], [3, 1], [3, 2], [3, 2], [2, 2], [2, 2], [1, 2], [1, 2]],
+      %w(meandering horizontal right-left bottom-top) => [[1, 1], [1, 1], [2, 1], [2, 1], [3, 1], [3, 1], [3, 2], [3, 2], [2, 2], [2, 2], [1, 2], [1, 2]],
+      %w(meandering horizontal right-left top-bottom) => [[1, 1], [1, 1], [2, 1], [2, 1], [3, 1], [3, 1], [3, 2], [3, 2], [2, 2], [2, 2], [1, 2], [1, 2]],
+      %w(flyback vertical top-bottom left-right) => [[1, 1], [1, 1], [1, 2], [1, 2], [2, 1], [2, 1], [2, 2], [2, 2], [3, 1], [3, 1], [3, 2], [3, 2]],
+      %w(flyback vertical bottom-top right-left) => [[3, 2], [3, 2], [3, 1], [3, 1], [2, 2], [2, 2], [2, 1], [2, 1], [1, 2], [1, 2], [1, 1], [1, 1]],
+      %w(flyback vertical bottom-top left-right) => [[1, 2], [1, 2], [1, 1], [1, 1], [2, 2], [2, 2], [2, 1], [2, 1], [3, 2], [3, 2], [3, 1], [3, 1]],
+      %w(flyback vertical top-bottom right-left) => [[3, 1], [3, 1], [3, 2], [3, 2], [2, 1], [2, 1], [2, 2], [2, 2], [1, 1], [1, 1], [1, 2], [1, 2]],
+      %w(meandering vertical top-bottom left-right) => [[1, 1], [1, 1], [1, 2], [1, 2], [2, 2], [2, 2], [2, 1], [2, 1], [3, 1], [3, 1], [3, 2], [3, 2]],
+      %w(meandering vertical bottom-top right-left) => [[3, 2], [3, 2], [3, 1], [3, 1], [2, 1], [2, 1], [2, 2], [2, 2], [1, 2], [1, 2], [1, 1], [1, 1]],
+      %w(meandering vertical bottom-top left-right) => [[1, 2], [1, 2], [1, 1], [1, 1], [2, 1], [2, 1], [2, 2], [2, 2], [3, 2], [3, 2], [3, 1], [3, 1]],
+      %w(meandering vertical top-bottom right-left) => [[3, 1], [3, 1], [3, 2], [3, 2], [2, 2], [2, 2], [2, 1], [2, 1], [1, 1], [1, 1], [1, 2], [1, 2]]
+    }
+    cats = %w(scan_pattern scan_type linescan_direction linescan_sequence).map(&:to_sym)
+    testing. each do |(configs, expected)|
+      config = Hash[ cats.zip(configs).map.to_a ]
+    config[:max_dimensions_pixels] = '3x2'
+    config[:shots_per_position] = 2
+    writer.x_y_positions(config).should == expected
+    end
+  end
+
+  it 'writes :processed binary file with spectra and returns DataArrayInfo objects' do
+    expected_file = TESTFILES + "/mspire/imzml/processed_binary_check.ibd"
+    write_to = TESTFILES + "/mspire/imzml/processed_binary.tmp.ibd"
+    array = subject.write_binary(write_to, @spectra.each, @config.merge( {:data_structure => :processed} ))
+    IO.read(write_to).should == IO.read(expected_file)
+    array.should be_an(Array)
+    length = 3
+    offsets = (16..76).step(12)
+    encoded_length = 12
+
+    array.each do |info_pair|
+      info_pair.each do |obj| 
+        obj.should be_a(Mspire::Imzml::DataArrayInfo)
+        obj.length.should == length
+        obj.encoded_length.should == encoded_length
+        obj.offset.should == offsets.next
+      end
+    end
+    File.unlink(write_to) if File.exist?(write_to)
+  end
+
+  it 'writes :continuous binary file with spectra and returns DataArrayInfo objects' do
+    expected_file = TESTFILES + "/mspire/imzml/continuous_binary_check.ibd"
+    write_to = TESTFILES + "/mspire/imzml/continuous_binary.tmp.ibd"
+    array = subject.write_binary(write_to, @spectra.each, @config.merge( {:data_structure => :continuous} ))
+    IO.read(write_to).should == IO.read(expected_file)
+    array.should be_an(Array)
+
+    length = 3
+    offsets = (28..52).step(12)
+    encoded_length = 12
+    first_offset = 16
+
+    array.each do |info_pair|
+      info_pair.each do |obj| 
+        obj.should be_a(Mspire::Imzml::DataArrayInfo)
+        obj.length.should == length
+        obj.encoded_length.should == encoded_length
+      end
+      info_pair.first.offset.should == first_offset
+      info_pair.last.offset.should == offsets.next
+    end
+    File.unlink(write_to) if File.exist?(write_to)
+  end
+
+
+end

data/spec/testfiles/mspire/mzml/mspire_simulated.MSn.check.mzML

@@ -40,15 +40,15 @@
         </scanList>
         <binaryDataArrayList count="2">
           <binaryDataArray encodedLength="28">
+            <cvParam cvRef="MS" accession="MS:1000514" name="m/z array"/>
             <cvParam cvRef="MS" accession="MS:1000523" name="64-bit float"/>
             <cvParam cvRef="MS" accession="MS:1000574" name="zlib compression"/>
-            <cvParam cvRef="MS" accession="MS:1000514" name="m/z array"/>
             <binary>eJxjYACBD/YMEOAAoTgcABe3Abg=</binary>
           </binaryDataArray>
           <binaryDataArray encodedLength="28">
+            <cvParam cvRef="MS" accession="MS:1000515" name="intensity array"/>
             <cvParam cvRef="MS" accession="MS:1000523" name="64-bit float"/>
             <cvParam cvRef="MS" accession="MS:1000574" name="zlib compression"/>
-            <cvParam cvRef="MS" accession="MS:1000515" name="intensity array"/>
             <binary>eJxjYAABAQcwxSACpSUcAAjwAP0=</binary>
           </binaryDataArray>
         </binaryDataArrayList>
@@ -73,15 +73,15 @@
         </precursorList>
         <binaryDataArrayList count="2">
           <binaryDataArray encodedLength="28">
+            <cvParam cvRef="MS" accession="MS:1000514" name="m/z array"/>
             <cvParam cvRef="MS" accession="MS:1000523" name="64-bit float"/>
             <cvParam cvRef="MS" accession="MS:1000574" name="zlib compression"/>
-            <cvParam cvRef="MS" accession="MS:1000514" name="m/z array"/>
             <binary>eJxjYACBD/YMEOAAoXgcABe/Abw=</binary>
           </binaryDataArray>
           <binaryDataArray encodedLength="28">
+            <cvParam cvRef="MS" accession="MS:1000515" name="intensity array"/>
             <cvParam cvRef="MS" accession="MS:1000523" name="64-bit float"/>
             <cvParam cvRef="MS" accession="MS:1000574" name="zlib compression"/>
-            <cvParam cvRef="MS" accession="MS:1000515" name="intensity array"/>
             <binary>eJxjYAABEQcwxSABpUUcAAlYAQE=</binary>
           </binaryDataArray>
         </binaryDataArrayList>

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: mspire
 version: !ruby/object:Gem::Version
-  version: 0.7.6
+  version: 0.7.7
   prerelease: 
 platform: ruby
 authors:
@@ -175,7 +175,8 @@ dependencies:
 description: mass spectrometry proteomics, lipidomics, and tools, a rewrite of mspire,
   merging of ms-* gems
 email: jtprince@gmail.com
-executables: []
+executables:
+- mzml_to_imzml
 extensions: []
 extra_rdoc_files:
 - LICENSE
@@ -185,6 +186,7 @@ files:
 - README.rdoc
 - Rakefile
 - VERSION
+- bin/mzml_to_imzml
 - lib/core_ext/array/in_groups.rb
 - lib/cv.rb
 - lib/cv/param.rb
@@ -220,6 +222,8 @@ files:
 - lib/mspire/ident/protein.rb
 - lib/mspire/ident/protein_group.rb
 - lib/mspire/ident/search.rb
+- lib/mspire/imzml/writer.rb
+- lib/mspire/imzml/writer/commandline.rb
 - lib/mspire/isotope.rb
 - lib/mspire/isotope/aa.rb
 - lib/mspire/isotope/distribution.rb
@@ -292,6 +296,7 @@ files:
 - spec/mspire/ident/pepxml/search_hit/modification_info_spec.rb
 - spec/mspire/ident/pepxml_spec.rb
 - spec/mspire/ident/protein_group_spec.rb
+- spec/mspire/imzml/writer_spec.rb
 - spec/mspire/isotope/aa_spec.rb
 - spec/mspire/isotope/distribution_spec.rb
 - spec/mspire/isotope_spec.rb
@@ -315,8 +320,11 @@ files:
 - spec/mspire_spec.rb
 - spec/obo_spec.rb
 - spec/spec_helper.rb
+- spec/testfiles/continuous_binary.tmp.ibd
 - spec/testfiles/mspire/ident/peptide/db/uni_11_sp_tr.fasta
 - spec/testfiles/mspire/ident/peptide/db/uni_11_sp_tr.msd_clvg2.min_aaseq4.yml
+- spec/testfiles/mspire/imzml/continuous_binary_check.ibd
+- spec/testfiles/mspire/imzml/processed_binary_check.ibd
 - spec/testfiles/mspire/mzml/j24z.idx_comp.3.mzML
 - spec/testfiles/mspire/mzml/mspire_simulated.MSn.check.mzML
 - spec/testfiles/mspire/mzml/openms.noidx_nocomp.12.mzML
@@ -340,6 +348,7 @@ files:
 - spec/testfiles/mspire/quant/remove_rest_of_proteins.rb
 - spec/testfiles/mspire/quant/unlog_transform.rb
 - spec/testfiles/plms1/output.key
+- spec/testfiles/processed_binary.tmp.ibd
 homepage: http://github.com/princelab/mspire
 licenses:
 - MIT