ms-msrun 0.0.1 → 0.1.0

data/lib/ms/msrun/search.rb

@@ -1,3 +1,4 @@
+require 'ms/mass'
 
 module Ms
   class Msrun
@@ -13,13 +14,12 @@ module Ms
 
     module Search
 
-      PROTON_MASS = 1.007276
-
-      # returns a string, or writes the string to file if given an out_filename
-      # if given a filename or IO object, returns the number of spectra
-      # written
-      def to_mgf(file_or_io=nil, opts={})
+      # returns a string unless :output given (may be a String (filename) or a
+      # writeable IO object in which case the data is written to file or io
+      # and the number of spectra written is returned
+      def to_mgf(opts={})
         opts = {
+          :output => nil,  # an output file or io object
           :bottom_mh => 0.0,
           :top_mh => nil,
           :ms_levels => (2..-1),  # range or intger, -1 at end will be substituted for last level
@@ -30,8 +30,10 @@ module Ms
           :prec_int_precision => 2,
           :frag_mz_precision => 5,
           :frag_int_precision => 1,
+          :charge_states_for_unknowns => [2,3],
+          :determine_plus_ones => false,
         }.merge(opts)
-       (_first_scan, _last_scan, _bottom_mh, _top_mh, _ms_levels, _min_peaks, _charge_states, _prec_mz_precision, _prec_int_precision, _frag_mz_precision, _frag_int_precision) = opts.values_at(:first_scan, :last_scan, :bottom_mh, :top_mh, :ms_levels, :min_peaks, :charge_states, :prec_mz_precision, :prec_int_precision, :frag_mz_precision, :frag_int_precision)
+       (_first_scan, _last_scan, _bottom_mh, _top_mh, _ms_levels, _min_peaks, _charge_states_for_unknowns, _prec_mz_precision, _prec_int_precision, _frag_mz_precision, _frag_int_precision, _determine_plus_ones) = opts.values_at(:first_scan, :last_scan, :bottom_mh, :top_mh, :ms_levels, :min_peaks, :charge_states_for_unknowns, :prec_mz_precision, :prec_int_precision, :frag_mz_precision, :frag_int_precision, :determine_plus_ones)
 
        sep = ' '
 
@@ -40,7 +42,7 @@ module Ms
        end
 
        if _last_scan.nil? or _last_scan == -1
-         _last_scan = scans.last.num
+         _last_scan = self.scan_nums.last
        end
 
        if !_ms_levels.is_a?(Integer) && _ms_levels.last == -1
@@ -50,26 +52,30 @@ module Ms
         prec_string = "PEPMASS=%0.#{_prec_mz_precision}f %0.#{_prec_int_precision}f\n"
         frag_string = "%0.#{_frag_mz_precision}f%s%0.#{_frag_int_precision}f\n"
 
-        any_input(file_or_io) do |out, out_type|
-          scans.each do |scan|
+        any_output(opts[:output]) do |out, out_type|
+          each_scan(:ms_level => _ms_levels) do |scan|
             sn = scan.num
 
-            next unless _ms_levels === scan.ms_level
             next unless sn >= _first_scan and sn <= _last_scan
             next unless scan.num_peaks >= _min_peaks
 
-            # tic under precursor > 95% and true = save the spectrum info
-            scan.spectrum.save!
-            if scan.plus1?(0.95)
-              _charge_states = [1]
+            
+            if _determine_plus_ones
+              # tic under precursor > 95% and true = save the spectrum info
+              if scan.plus1?(0.95)
+                _charge_states = [1]
+              end
             end
 
-            # (scanHeader.precursorMZ * iCharge) - (iCharge - 1)*dChargeMass;
+            chrg_sts = scan.precursor.charge_states
+            if chrg_sts.nil? || !chrg_sts.first.is_a?(Integer)
+              chrg_sts = _charge_states_for_unknowns
+            end
 
             pmz = scan.precursor && scan.precursor.mz
 
-            _charge_states.each do |z|
-              mh = (pmz * z) - (z - 1)*PROTON_MASS
+            chrg_sts.each do |z|
+              mh = (pmz * z) - (z - 1)*Ms::Mass::PROTON
               next unless (mh >= _bottom_mh) 
               next unless (mh <= _top_mh) if _top_mh
               out.puts "BEGIN IONS"
@@ -81,8 +87,6 @@ module Ms
               end
               out.puts "END IONS\n\n"
             end
-
-            scan.spectrum.flush!
           end
 
           if out_type == :string_io
@@ -96,7 +100,7 @@ module Ms
 
 
       # yields an IO object and the type input (:io, :filename, :string_io)
-      def any_input(arg, &block)
+      def any_output(arg, &block)
         # this is pretty ugly, can we clean up?
           if arg.is_a? IO  # an IO object passed in
             block.call(arg, :io)

data/lib/ms/msrun/sha1.rb

@@ -0,0 +1,36 @@
+require 'digest/sha1'
+
+module Ms
+  class Msrun
+    # the mzXML digest is from the start of the document to the end of the
+    # first sha1 tag: '...<sha1>'
+    module Sha1
+      module_function
+
+      # returns [calculated digest, recorded digest] for an mzXML file 
+      def digest_mzxml_file(file)
+        recorded_digest = nil
+
+        incr_digest = ""
+        #incr_digest = Digest::SHA1.new
+        endpos = nil
+        File.open(file, 'rb') do |io|
+          while line = io.gets
+            if line.include?("<sha1>")  
+              incr_digest << line[0, line.index("<sha1>") + 6]
+              if line =~ %r{<sha1>(.*)</sha1>}
+                recorded_digest = $1.dup
+                break
+              else
+                incr_digest << line
+              end
+            end
+          end
+        end
+
+        [Digest::SHA1.hexdigest(incr_digest), recorded_digest]
+        #[incr_digest.hexdigest, recorded_digest]
+      end
+    end
+  end
+end

data/lib/ms/mzxml.rb

@@ -0,0 +1,12 @@
+
+module Ms
+  module Mzxml
+    module_function
+    def parent_basename_and_dir(xml_value)
+      fn = xml_value.gsub(/\\/, '/')
+      dn = File.dirname(fn)
+      dn = nil if dn == '.' && !fn.include?('/')
+      [File.basename(fn), dn]
+    end
+  end
+end

data/lib/ms/precursor.rb

@@ -1,8 +1,9 @@
 
 module Ms
   # charge_states are the possible charge states of the precursor
-  # parent references a scan
-  PrecursorAtts = [:mz, :intensity, :parent, :charge_states]
+  # parent references a scan and scan references the scan the precursor
+  # belongs to (i.e., the scan holding the precursor information).
+  PrecursorAtts = [:mz, :intensity, :parent, :charge_states, :scan]
 end
 
 Ms::Precursor = Struct.new(*Ms::PrecursorAtts)

data/lib/ms/precursor/lazy_parent.rb

@@ -0,0 +1,28 @@
+require 'ms/precursor'
+
+module Ms
+  class Precursor
+    class LazyParent
+      undef :parent
+
+      def parent
+        #########################
+        #########################
+        #########################
+        #########################
+        #########################
+        #########################
+        
+        WorkingHERE!!
+
+        #########################
+        #########################
+        #########################
+        #########################
+        #########################
+        scan.num
+      end
+    end
+  end
+end
+

data/lib/ms/scan.rb

@@ -2,16 +2,14 @@ require 'ms/precursor'
 
 module Ms ; end
 
-#              0      1          2      3         4        5           6     7
-#        8 
+#              0      1          2      3         4        5           6     7           8 
 MsScanAtts =  [:num, :ms_level, :time, :start_mz, :end_mz, :num_peaks, :tic, :precursor, :spectrum]
 
-Ms::Scan = Struct.new(*MsScanAtts)
+Ms::Scan = Struct.new(*MsScanAtts) do
 
 # time in seconds
 # everything else in float/int
 
-class Ms::Scan
  
   def to_s
     "<Scan num=#{num} ms_level=#{ms_level} time=#{time}>"
@@ -63,31 +61,6 @@ class Ms::Scan
     arr.join(" ")
   end
 
-  # adds the attribute parent to each scan with a parent
-  # (level 1 = no parent; level 2 = prev level 1, etc.
-  def self.add_parent_scan(scans)
-    prev_scan = nil
-    parent_stack = [nil]
-    ## we want to set the level to be the first mslevel we come to
-    prev_level = 1
-    scans.each do |scan|
-      if scan then prev_level = scan.ms_level; break; end
-    end
-    scans.each do |scan|
-      next unless scan  ## the first one is nil, (others?)
-      level = scan.ms_level
-      if prev_level < level
-        parent_stack.unshift prev_scan
-      end
-      if prev_level > level
-        (prev_level - level).times do parent_stack.shift end
-      end
-      scan.parent = parent_stack.first
-      prev_level = level
-      prev_scan = scan
-    end
-  end
-
 end
 
 

data/lib/ms/spectrum/compare.rb

@@ -17,14 +17,54 @@ end
 module Ms
   class Spectrum
     module Compare
+  
+      # percent ion current score: the percent of total ion current of the
+      # calling object that can be explained by other. Ranges from 0-100
+      def pic_score(other, opts={})
+        opts= {:normalize => true}.merge(opts)
+        (a_spec, b_spec) = 
+          if opts[:normalize] == true
+            [self.normalize, other.normalize]
+          else
+            [self, other]
+          end
+        overlapping_current = 0.0
+        a_spec.compare(b_spec, opts.merge( {:yield_diff => false} )) do |sint, oint|
+          overlapping_current += [oint, sint].min
+        end
+        100.0 * (overlapping_current / a_spec.intensities.sum)
+      end
+
 
+      # Zhang Analytical Chemistry. 2004 76(14)
+      # "the ratio between the sum of geometric mean and sum of arithmetic mean
+      # of all ions, after the two spectra are normalized to the same total
+      # ion intensity"
+      # The score will be 1 for identical spectra
       # (Σ (Ii*Ij)^½) / (ΣIi * ΣIj)^½
+      #
+      # options:
+      # 
+      #   :normalize => true | false (d: true) normalizes intensities first
+      #
+      # relevant options given for 'compare' will be passed through:
+      #
+      #     :radius
+      #     :type
       def sim_score(other, opts={})
+        opts = {:normalize => true}.merge(opts)
+        (a_spec, b_spec) = 
+          if opts[:normalize] == true
+            [self.normalize, other.normalize]
+          else
+            [self, other]
+          end
         numer = 0.0
-        compare(other, {:yield_diff => false}) do |sint, oint|
+        
+        a_spec.compare(b_spec, opts.merge( {:yield_diff => false} )) do |sint, oint|
           numer += Math.sqrt(sint * oint)
         end
-        numer / Math.sqrt( self.ints.sum * other.ints.sum )
+        numer / Math.sqrt( a_spec.intensities.sum * b_spec.intensities.sum )
       end
 
       # opts[:type] == :mutual_best

data/lib/ms/spectrum/filter.rb

@@ -37,7 +37,7 @@ module Ms
         ret_ints = []
         ret_mzs = include.map {|int, mz| [mz, int] }.sort.map {|mz,int| ret_ints << int ; mz }
 
-        return Spectrum.new(ret_mzs, ret_ints)
+        return Spectrum.new([ret_mzs, ret_ints])
       end
 
     end

data/spec/ms/msrun/index_spec.rb

@@ -0,0 +1,60 @@
+require File.expand_path(File.dirname(__FILE__) + '/../../spec_helper')
+
+require 'rexml/document'
+require 'ms/msrun/index'
+
+class MsMsrunIndexSpec < MiniTest::Spec
+
+  before do
+    @indices = @files.map do |file|
+      indices = Ms::Msrun::Index.new(file)
+    end
+  end
+
+  def initialize(*args)
+    @files = %w(000.v1.mzXML 000.v2.1.mzXML 020.v2.0.readw.mzXML).map {|v| TESTFILES + '/opd1/' + v }
+    super *args
+  end
+
+  it 'is indexed by scan num and gives doublets of byte and length' do
+    @files.zip(@indices) do |file, index|
+      index.each_with_index do |pair,i|
+        string = IO.read(file, pair.last, pair.first).strip
+        string[0,5].must_equal '<scan'
+        string[-7..-1].must_match %r{</scan>|/peaks>|/msRun>}
+        string.must_match %r{num="#{i+1}"}
+      end
+    end
+  end
+
+  it 'gives scan_nums' do
+    @indices.each do |index|
+      index.scan_nums.must_equal((1..20).to_a)
+    end
+  end
+
+  it 'is enumerable' do
+    @indices.each do |index|
+      scan_nums = index.scan_nums
+      index.each_with_index do |doublet,i|
+        index[scan_nums[i]].must_equal doublet
+      end
+    end
+  end
+
+  it 'gives header length' do
+    header_lengths = [824, 1138, 1147]
+    @indices.zip(@files, header_lengths) do |index, file, header_length|
+      index.header_length.must_equal header_length
+    end
+  end
+
+  it 'gives a scan for #first and #last' do
+    # TODO: fill in with actual data too
+    @indices.each do |index|
+      index.first.wont_equal nil
+      index.last.wont_equal nil
+    end
+  end
+
+end

data/spec/ms/msrun/scan_spec.rb

@@ -0,0 +1,78 @@
+require File.expand_path(File.dirname(__FILE__) + '/../../spec_helper')
+
+require 'ms/msrun'
+
+Scan1 = '<scan num="4"
+        msLevel="2"
+        peaksCount="74"
+        polarity="+"
+        scanType="Full"
+        retentionTime="PT3.64S"
+        collisionEnergy="35"
+        lowMz="325"
+        highMz="2000"
+	basePeakMz="1090.05"
+	basePeakIntensity="789939"
+	totIonCurrent="2.27206e+006">
+    <precursorMz precursorIntensity="1.52022e+006">1222.033203</precursorMz>
+    <peaks precision="32"
+           byteOrder="network"
+           pairOrder="m/z-int">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</peaks>
+  </scan>
+  </scan>
+'
+
+Scan2 = '<scan num="4"
+        msLevel="2"
+        peaksCount="74"
+        polarity="+"
+        scanType="Full"
+        retentionTime="PT3.64S"
+        collisionEnergy="35"
+        lowMz="325"
+        highMz="2000"
+	basePeakMz="1090.05"
+	basePeakIntensity="789939"
+	totIonCurrent="2.27206e+006">
+    <precursorMz precursorIntensity="1.52022e+006">1222.033203</precursorMz>
+    <peaks precision="32"
+           byteOrder="network"
+           pairOrder="m/z-int">Q7Rw2EUK8ABDuWjORQSgAEO5+NhFHLAAQ8gNTEU10ABD2QKcRLCAAEPcdCJFxBAAQ92bFkZzPABEAIucRSbQAEQJuBJFTyAARAqM7EX9+ABECsPMRT7QAEQPyWRF1LgARBoLhETLoABEHjlsRZQwAEQefjRE3yAARCERUkYvuABEIYiKRMMAAEQh01xFifAARCNEKEYpuABEI3kcRWrgAEQjvPhFHRAARCZm7EZtNABEJpXwRnyMAEQmy+pF4QAARCb/RkWfIABEKY28RZQIAEQ6gGBFyHAARDrIYEVEsABEPLuARXJgAEQ9RdREQ8AARD2BIEaV7ABEP7tcRdzoAERChMxHVPcAREK2gEZQAABEQwDkRa/gAEREyLxE3UAARFDwhEXwKABEUT7uRTPQAERTbZZFhfAARFQQvEaK9gBEVOfcRaDQAERWgBJGY5AARFh4ukcWmgBEWMAqRppEAERZgzhF+SgARFsBakXOKABEW4IiR8GBAERbw5pGUvwARFwJcEUooABEXat2RmUoAERegQRHibgARF7FmkU2AABEY6z2RgBIAERvwkxF/pgARHJ9YEYXEABEcrySRgFkAER0gDJINYsARHTFKEeJloBEdPgURoz8AER3hVhIlXggRHfIAEd2YgBEd/acRuh2AER4IiI/gAAARIW+tEVbkABEhdH8RnqcAESIQXhJQNswRIhimEecEgBEiH0cRvKMAESInkBGguYARIp9kEWm0ABEkUz6RdCQAESR//BFH5AARJQhYkYGbABEmxG2RjnYAA==</peaks>
+'
+
+Scan3 = '<scan num="4"
+        msLevel="2"
+        peaksCount="74"
+        polarity="+"
+        scanType="Full"
+        retentionTime="PT3.64S"
+        collisionEnergy="35"
+        lowMz="325"
+        highMz="2000"
+	basePeakMz="1090.05"
+	basePeakIntensity="789939"
+	totIonCurrent="2.27206e+006">
+    <precursorMz precursorIntensity="1.52022e+006">1222.033203</precursorMz>
+    <peaks precision="32"
+           byteOrder="network"
+           pairOrder="m/z-int">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</peaks>
+           </scan>
+           </scan>
+           </msRun>
+'
+
+#class Sha1Spec < MiniTest::Spec
+  #def initialize(*args)
+    #@files = %w(000.v1.mzXML 020.v2.0.readw.mzXML 000.v2.1.mzXML).map do |file|
+      #TESTFILES + "/opd1/#{file}"
+    #end
+    #super(*args)
+  #end
+
+  ### NOTE: this does NOT match up to real files yet!
+  #it 'can read xml scans with extra or missing tags' do
+    #Scan.new(from_xml)    
+  #end
+
+
+#end