ms-msrun 0.0.1 → 0.1.0

data/spec/ms/msrun/search_spec.rb

@@ -4,8 +4,8 @@ require 'ms/msrun'
 
 class SearchSpec < MiniTest::Spec
 
-  xit 'works' do
-    @file = '/home/jtprince/ms-msrun/spec/files/opd1/000.v1.mzXML'
+  it 'creates mgf formatted files' do
+    @file = TESTFILES + '/opd1/000.v1.mzXML'
     params = {
       :bottom_mh => 300.0,
       :top_mh => 4500.0,
@@ -27,7 +27,7 @@ class SearchSpec < MiniTest::Spec
     ]
     Ms::Msrun.open(@file) do |ms|
       no_scans.each do |k,v|
-        ms.to_mgf(nil, k => v).must_equal ""
+        ms.to_mgf( k => v).must_equal ""
       end
     end
 
@@ -43,14 +43,13 @@ class SearchSpec < MiniTest::Spec
       [:bottom_mh, 500],
     ]
     Ms::Msrun.open(@file) do |ms|
-      no_scans.each do |k,v|
-        reply = ms.to_mgf(nil, k => v)
-        puts reply
+      some_scans.each do |k,v|
+        reply = ms.to_mgf(k => v)
         reply.must_match(/BEGIN.IONS/)
         reply.must_match(/END.IONS/)
       end
     end
-
+    # TODO: should write some more specs here
   end
 
 end

data/spec/ms/msrun/sha1_spec.rb

@@ -0,0 +1,23 @@
+require File.expand_path(File.dirname(__FILE__) + '/../../spec_helper')
+
+require 'ms/msrun/sha1'
+
+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!
+  xit 'can determine a sha1 for an mzxml file' do
+    @files.each do |file|
+      (actual, recorded) = Ms::Msrun::Sha1.digest_mzxml_file file
+      [actual, recorded].each {|v| assert !v.nil? }
+      actual.must_equal recorded
+    end
+  end
+
+
+end

data/spec/ms/msrun_spec.rb

@@ -8,27 +8,111 @@ module MsrunSpec
   before do
     @file = nil  # you need to define this!
     @key = nil   # <- do nothing with this.
+    @nums = (1..20).to_a  # define scan numbers
   end
 
   def key
     @key || @key = YAML.load_file(@file + '.key.yml')
   end
 
+  def hash_match(hash, obj)
+    #$DEBUG = 1
+    puts "SCAN: #{obj.num}" if $DEBUG && obj.respond_to?(:num)
+    hash.each do |k,v|
+      if v.is_a?(Hash)
+        hash_match(v, obj.send(k.to_sym))
+      else
+        puts "#{k}: #{v} but was #{obj.send(k.to_sym)}" if obj.send(k.to_sym) != v
+        obj.send(k.to_sym).must_equal v
+      end
+    end
+  end
+
   it 'reads header information' do
     Ms::Msrun.open(@file) do |ms|
       key['header'].each do |k,v|
+        #puts "K: #{k} Vexp: #{v} Vact: #{ms.send(k.to_sym)}"
         ms.send(k.to_sym).must_equal v
       end
     end
   end
 
+  it 'can access random scans' do
+    Ms::Msrun.open(@file) do |ms|
+      scan = ms.scan(20)
+      hash_match(key['scans'][20], scan)
+    end
+  end
+
+  it 'can read all scans' do
+    num_required_scans = key['scans'].size
+    Ms::Msrun.open(@file) do |ms|
+      ms.each do |scan|
+        if hash = key['scans'][scan.num]
+          hash_match(hash, scan)
+          num_required_scans -= 1
+        end
+      end
+    end
+    num_required_scans.must_equal 0
+  end
+
+  it 'can read scans of a certain ms_level' do
+    nums = [1,5,9,13,17]
+    Ms::Msrun.open(@file) do |ms|
+      ms.each(:ms_level => 1) do |scan|
+        scan.num.must_equal nums.shift 
+      end
+    end
+    nums = [2,3,4,6,7,8,10,11,12,14,15,16,18,19,20]
+    Ms::Msrun.foreach(@file, :ms_level => 2) do |scan|
+      scan.num.must_equal nums.shift 
+    end
+  end
+
+  it 'can avoid reading spectra' do 
+    nums = @nums.map
+    Ms::Msrun.foreach(@file, :spectrum => false) do |scan|
+      assert scan.spectrum.nil?
+      scan.num.must_equal nums.shift
+    end
+  end
+
+  it 'can avoid reading precursor information' do 
+    Ms::Msrun.foreach(@file, :precursor => false) do |scan|
+      assert scan.precursor.nil?
+    end
+  end
+
+  it 'gives scan counts for different ms levels' do
+    Ms::Msrun.open(@file) do |ms|
+      key['scan_count'].each do |index, count|
+        ms.scan_count(index).must_equal count
+      end
+    end
+  end
+
+  it 'gives start and end mz even if the information is not given' do
+    Ms::Msrun.open(@file) do |ms|
+      ms.start_and_end_mz_brute_force.must_equal(key['start_and_end_mz'][1])
+    end
+  end
+
 end
 
 class Mzxml_v1 < MiniTest::Spec
   include MsrunSpec
   before do
     super
-    @file = '/home/jtprince/ms-msrun/spec/files/opd1/000.v1.mzXML'
+    @file = TESTFILES + '/opd1/000.v1.mzXML'
+  end
+
+  it 'can give start and end mz' do
+    # scan has attributes startMz endMz
+    Ms::Msrun.open(@file) do |ms|
+      #ms.start_and_end_mz.must_equal([300.0, 1500.0])
+      ms.start_and_end_mz.must_equal(key['start_and_end_mz'][1])
+    end
   end
 end
 
@@ -36,7 +120,7 @@ class Mzxml_v2_0 < MiniTest::Spec
   include MsrunSpec
   before do
     super
-    @file = '/home/jtprince/ms-msrun/spec/files/opd1/020.v2.0.readw.mzXML'
+    @file = TESTFILES + '/opd1/020.v2.0.readw.mzXML'
   end
 end
 
@@ -44,7 +128,31 @@ class Mzxml_v2_1 < MiniTest::Spec
   include MsrunSpec
   before do
     super
-    @file = '/home/jtprince/ms-msrun/spec/files/opd1/000.v2.1.mzXML'
+    @file = TESTFILES + '/opd1/000.v2.1.mzXML'
+  end
+
+  it 'gives nil if scans do not have start and end mz info' do
+    # scans do not have startMz endMz or filterLine
+    Ms::Msrun.open(@file) do |ms|
+      ms.start_and_end_mz.must_equal([nil, nil])
+    end
   end
+
+  it 'gives start and end mz if filterLine present' do
+    newname = @file + ".TMP.mzXML"
+    File.open(newname, 'w') do |out|
+      IO.foreach(@file) do |line|
+        if line =~ /msLevel="1"/
+          out.puts %Q{        filterLine="FTMS + p NSI Full ms [300.00-1500.00]"}
+        end
+        out.print line
+      end
+    end
+    Ms::Msrun.open(newname) do |ms|
+      ms.start_and_end_mz.must_equal([300.0, 1500.0])
+    end
+    File.unlink(newname) if File.exist?(newname)
+  end
+
 end
 

data/spec/ms/scan_spec.rb

@@ -7,11 +7,11 @@ class ScanUnitSpec < MiniTest::Spec
   before do
     @scan = Ms::Scan.new
     @scan.precursor = Ms::Precursor.new
-    @scan.spectrum = Ms::Spectrum.new([1,2,3,4], [2,4,4,2])
+    @scan.spectrum = Ms::Spectrum.new([[1,2,3,4], [2,4,4,2]])
   end
 
 
-  it 'determines if its +1 or not' do
+ it 'determines if its +1 or not' do
     # these have not been checked for accuracy, just sanity
     reply = [0.1,2.5, 3.5, 5].map do |prec_mz|
       @scan.precursor.mz = prec_mz

data/spec/ms/spectrum/compare_spec.rb

@@ -13,11 +13,11 @@ class CompareSpec < MiniTest::Spec
   include Ms
 
   before do
-    @a = Spectrum.new([0,2,3,4], [5,6,7,8])
-    @b = Spectrum.new([0, 1.5, 3.5, 5.5], [9,10,11,12])
+    @a = Spectrum.new([[0,2,3,4], [5,6,7,8]])
+    @b = Spectrum.new([[0, 1.5, 3.5, 5.5], [9,10,11,12]])
 
-    @c = Spectrum.new([0, 1], [8,9])
-    @d = Spectrum.new([0.6, 0.75], [10,11])
+    @c = Spectrum.new([[0, 1], [8,9]])
+    @d = Spectrum.new([[0.6, 0.75], [10,11]])
   end
 
   it 'compares spectra' do
@@ -35,10 +35,17 @@ class CompareSpec < MiniTest::Spec
   end
 
   it 'computes similarity score' do
-    p @a.sim_score(@b)
-    p @a.sim_score(@a)
+    @a.sim_score(@a, :radius => 0.1).must_equal 1.0
+    # this is just frozen, not verified:
+    @a.sim_score(@b).must_be_close_to 0.702945603476432, 0.000001
   end
 
+  it 'computes a pic score' do
+    @a.pic_score(@a, :radius => 0.01).must_equal 100.0
+    @a.pic_score(@d, :radius => 0.01).must_equal 0.0
+    # frozen:
+    @a.pic_score(@b).must_be_close_to 68.4981684981685, 0.000001 
+  end
 end
 
 

data/spec/ms/spectrum/filter_spec.rb

@@ -14,14 +14,14 @@ class FilterSpec < MiniTest::Spec
   include Ms
 
   before do
-    @a = Spectrum.new([0,5,10, 15,16,17,18, 20.1], [0,1,2, 3,8,10,4, 0])
-    @null = Spectrum.new
+    @a = Spectrum.new([[0,5,10, 15,16,17,18, 20.1], [0,1,2, 3,8,10,4, 0]])
+    @null = Spectrum.new([[],[]])
   end
 
   it 'filters spectra' do
     spec = @a.filter(:bins, :bin_width => 10, :num_peaks => 2)
     spec.mzs.must_equal [5,10,16,17,20.1]
-    spec.ints.must_equal [1,2,8,10,0]
+    spec.intensities.must_equal [1,2,8,10,0]
 
     @a.filter(:bins, :bin_width => 100, :num_peaks => 8).must_equal @a
     @a.filter(:bins, :bin_width => 1, :num_peaks => 1).must_equal @a

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: ms-msrun
 version: !ruby/object:Gem::Version 
-  version: 0.0.1
+  version: 0.1.0
 platform: ruby
 authors: 
 - John Prince
@@ -9,18 +9,28 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2009-03-17 00:00:00 -06:00
+date: 2009-11-11 00:00:00 -07:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
-  name: axml
+  name: ms-core
   type: :runtime
   version_requirement: 
   version_requirements: !ruby/object:Gem::Requirement 
     requirements: 
-    - - ~>
+    - - ">="
       - !ruby/object:Gem::Version 
-        version: 0.0.5
+        version: "0"
+    version: 
+- !ruby/object:Gem::Dependency 
+  name: nokogiri
+  type: :runtime
+  version_requirement: 
+  version_requirements: !ruby/object:Gem::Requirement 
+    requirements: 
+    - - ">="
+      - !ruby/object:Gem::Version 
+        version: "0"
     version: 
 - !ruby/object:Gem::Dependency 
   name: runarray
@@ -35,30 +45,38 @@ dependencies:
 description: A library for working with LC/MS runs. Part of mspire.  Has parsers for mzXML v1, 2, and 3, mzData and mzML.  Can convert to commonly desired search output (such as mgf)
 email: jtprince@gmail.com
 executables: 
-- ms_to_search.rb
 - base64_to_array.rb
 - ms_to_obiwarp.rb
+- ms_to_search.rb
 extensions: []
 
 extra_rdoc_files: 
 - README
 - LICENSE
 files: 
-- lib/ms/spectrum.rb
-- lib/ms/msrun/axml/mzxml.rb
-- lib/ms/msrun/search.rb
-- lib/ms/msrun.rb
-- lib/ms/scan.rb
-- lib/ms/spectrum/compare.rb
+- lib/lmat.rb
 - lib/ms/spectrum/filter.rb
+- lib/ms/spectrum/compare.rb
+- lib/ms/scan.rb
+- lib/ms/msrun/sha1.rb
+- lib/ms/msrun/regexp/mzxml.rb
+- lib/ms/msrun/search.rb
+- lib/ms/msrun/index.rb
+- lib/ms/msrun/nokogiri/mzxml.rb
+- lib/ms/msrun/axml/mzxml.rb
+- lib/ms/msrun/mzxml.rb
+- lib/ms/msrun/nokogiri.rb
 - lib/ms/precursor.rb
-- lib/bsearch.rb
-- lib/lmat.rb
+- lib/ms/mzxml.rb
+- lib/ms/msrun.rb
+- lib/ms/precursor/lazy_parent.rb
 - README
 - LICENSE
 - Rakefile
 has_rdoc: true
 homepage: http://mspire.rubyforge.org/projects/ms-msrun
+licenses: []
+
 post_install_message: 
 rdoc_options: 
 - --main
@@ -81,17 +99,20 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version 
       version: "0"
   version: 
-requirements: 
-- xmlparser (preferrably) or libxml
+requirements: []
+
 rubyforge_project: mspire
-rubygems_version: 1.3.1
+rubygems_version: 1.3.2
 signing_key: 
-specification_version: 2
+specification_version: 3
 summary: A library for working with LC/MS runs
 test_files: 
-- spec/ms/msrun_spec.rb
-- spec/ms/msrun/search_spec.rb
-- spec/ms/spectrum/compare_spec.rb
 - spec/ms/spectrum/filter_spec.rb
+- spec/ms/spectrum/compare_spec.rb
+- spec/ms/msrun/search_spec.rb
+- spec/ms/msrun/sha1_spec.rb
+- spec/ms/msrun/index_spec.rb
+- spec/ms/msrun/scan_spec.rb
+- spec/ms/msrun_spec.rb
 - spec/ms/scan_spec.rb
 - spec/lmat_spec.rb

data/lib/bsearch.rb

@@ -1,120 +0,0 @@
-#
-# Ruby/Bsearch - a binary search library for Ruby.
-#
-# Copyright (C) 2001 Satoru Takabayashi <satoru@namazu.org>
-#     All rights reserved.
-#     This is free software with ABSOLUTELY NO WARRANTY.
-#
-# You can redistribute it and/or modify it under the terms of 
-# the Ruby's licence.
-#
-# Example:
-#
-#  % irb -r ./bsearch.rb
-#  >> %w(a b c c c d e f).bsearch_first {|x| x <=> "c"}
-#  => 2
-#  >> %w(a b c c c d e f).bsearch_last {|x| x <=> "c"}
-#  => 4
-#  >> %w(a b c e f).bsearch_first {|x| x <=> "c"}
-#  => 2
-#  >> %w(a b e f).bsearch_first {|x| x <=> "c"}
-#  => nil
-#  >> %w(a b e f).bsearch_last {|x| x <=> "c"}
-#  => nil
-#  >> %w(a b e f).bsearch_lower_boundary {|x| x <=> "c"}
-#  => 2
-#  >> %w(a b e f).bsearch_upper_boundary {|x| x <=> "c"}
-#  => 2
-#  >> %w(a b c c c d e f).bsearch_range {|x| x <=> "c"}
-#  => 2...5
-#  >> %w(a b c d e f).bsearch_range {|x| x <=> "c"}
-#  => 2...3
-#  >> %w(a b d e f).bsearch_range {|x| x <=> "c"}
-#  => 2...2
-
-module Bsearch
-  VERSION = '1.5'
-end
-
-class Array
-  # 
-  # The binary search algorithm is extracted from Jon Bentley's
-  # Programming Pearls 2nd ed. p.93
-  #
-
-  #
-  # Return the lower boundary. (inside)
-  #
-  def bsearch_lower_boundary (range = 0 ... self.length, &block)
-    lower  = range.first() -1
-    upper = if range.exclude_end? then range.last else range.last + 1 end
-    while lower + 1 != upper
-      mid = ((lower + upper) / 2).to_i # for working with mathn.rb (Rational)
-      if yield(self[mid]) < 0
-	lower = mid
-      else 
-	upper = mid
-      end
-    end
-    return upper
-  end
-
-  #
-  # This method searches the FIRST occurrence which satisfies a
-  # condition given by a block in binary fashion and return the 
-  # index of the first occurrence. Return nil if not found.
-  #
-  def bsearch_first (range = 0 ... self.length, &block)
-    boundary = bsearch_lower_boundary(range, &block)
-    if boundary >= self.length || yield(self[boundary]) != 0
-      return nil
-    else 
-      return boundary
-    end
-  end
-
-  alias bsearch bsearch_first
-
-  #
-  # Return the upper boundary. (outside)
-  #
-  def bsearch_upper_boundary (range = 0 ... self.length, &block)
-    lower  = range.first() -1
-    upper = if range.exclude_end? then range.last else range.last + 1 end
-    while lower + 1 != upper
-      mid = ((lower + upper) / 2).to_i # for working with mathn.rb (Rational)
-      if yield(self[mid]) <= 0
-	lower = mid
-      else 
-	upper = mid
-      end
-    end
-    return lower + 1 # outside of the matching range.
-  end
-
-  #
-  # This method searches the LAST occurrence which satisfies a
-  # condition given by a block in binary fashion and return the 
-  # index of the last occurrence. Return nil if not found.
-  #
-  def bsearch_last (range = 0 ... self.length, &block)
-    # `- 1' for canceling `lower + 1' in bsearch_upper_boundary.
-    boundary = bsearch_upper_boundary(range, &block) - 1
-
-    if (boundary <= -1 || yield(self[boundary]) != 0)
-      return nil
-    else
-      return boundary
-    end
-  end
-
-  #
-  # Return the search result as a Range object.
-  #
-  def bsearch_range (range = 0 ... self.length, &block)
-    lower = bsearch_lower_boundary(range, &block)
-    upper = bsearch_upper_boundary(range, &block)
-    return lower ... upper
-  end
-end
-