mspire 0.8.4 → 0.8.5

data/script/mascot_dat_to_peptide_hit_qvalues.rb

@@ -0,0 +1,118 @@
+#!/usr/bin/env ruby
+
+require 'trollop'
+require 'set'
+require 'mspire/ident/peptide_hit/qvalue'
+require 'mspire/error_rate/qvalue'
+
+begin
+  require 'mascot/dat'
+rescue LoadError
+  puts "You need the mascot-dat gem for this to work!"
+  puts "AND IT MUST BE THE PRINCELAB GITHUB FORK until changes get incorporated upstream!"
+  puts ">     gem install mascot-dat"
+  raise LoadError
+end
+raise "need princelab mascot-dat gem!" unless Mascot::DAT::VERSION == "0.3.1.1"
+
+# target-decoy bundle
+SearchBundle = Struct.new(:target, :decoy) do
+  # combines all bundles under self so that all targets are grouped and all
+  # decoys are grouped.  returns self
+  def combine(bundles)
+    (targets, decoys) = bundles.map {|bundle| [bundle.target, bundle.decoy] }
+    .transpose.map {|ars| ars.reduce(:+) }
+    self
+  end
+end
+
+PSM = Struct.new(:search_id, :id, :aaseq, :charge, :score)
+
+# turns 1+ into 1
+def charge_string_to_charge(st)
+  md = st.match(/(\d)([\+\-])/)
+  i = md[1].to_i 
+  i *= -1 if (md[2] == '-')
+  i
+end
+
+def read_mascot_dat_hits(dat_file)
+  filename =nil
+  IO.foreach(dat_file) do |line| 
+    if line =~ /^FILE=(.*?).mgf/i
+      filename = $1.dup
+      break
+    end
+  end
+  dat = Mascot::DAT.open(dat_file)
+
+  data = [:peptides, :decoy_peptides].map do |mthd|
+    psms = []
+    dat.send(mthd).each do |psm|
+      next unless psm.query
+      query = dat.query(psm.query)
+      charge = charge_string_to_charge(query.charge)
+      psms << PSM.new(filename, query.title, psm.pep, charge, psm.score) if psm.score
+    end
+    psms
+  end
+  dat.close
+  SearchBundle.new(*data)
+end
+
+
+def putsv(*args)
+  puts(*args) if $VERBOSE
+  $stdout.flush
+end
+
+EXT = Mspire::Ident::PeptideHit::Qvalue::FILE_EXTENSION
+combine_base  = "combined"
+
+opts = Trollop::Parser.new do
+  #banner %Q{usage: #{File.basename(__FILE__)} <target>.xml <decoy>.xml ...
+  banner %Q{usage: #{File.basename(__FILE__)} <mascot>.dat ...
+outputs: <mascot>.phq.tsv
+assumes a decoy search was run *with* the initial search
+phq.tsv?: see schema/peptide_hit_qvalues.phq.tsv
+}
+  opt :combine, "groups target and decoy hits together from all files, writing to #{combine_base}#{EXT}", :default => false
+  opt :z_together, "do not group by charge state", :default => false
+  opt :verbose, "be verbose", :default => false
+end
+
+opt = opts.parse(ARGV)
+if ARGV.size == 0
+  opts.educate
+  exit 
+end
+
+$VERBOSE = opt.delete(:verbose)
+
+files = ARGV.to_a
+
+bundles = files.map do |file|
+  # assumes the file has both target and decoy hits
+  read_mascot_dat_hits(file)
+end
+
+to_run = {}
+if opt[:combine]
+  putsv "combining all target hits together and all decoy hits together"
+  bundle = SearchBundle.new.combine(bundles)
+  to_run[combine_base + EXT] = bundle
+else
+  files.zip(bundles) do |file, bundle|
+    to_run[file.chomp(File.extname(file)) + EXT] = bundle
+  end
+end
+
+to_run.each do |file, bundle|
+  putsv "calculating qvalues for #{file}"
+  hit_qvalue_pairs = Mspire::ErrorRate::Qvalue.target_decoy_qvalues(bundle.target, bundle.decoy, :z_together => opt[:z_together])
+  # {|hit| hit.search_scores[:ionscore] }
+  outfile = Mspire::Ident::PeptideHit::Qvalue.to_file(file, *hit_qvalue_pairs.transpose)
+  putsv "created: #{outfile}"
+end
+
+

data/spec/mspire/ident/peptide/db/creator_spec.rb

@@ -0,0 +1,65 @@
+require 'spec_helper'
+
+require 'yaml'
+path = 'mspire/ident/peptide/db'
+require path + "/creator" 
+
+describe 'creating a peptide centric database' do
+  subject { Mspire::Ident::Peptide::Db::Creator.new }
+
+  describe 'amino acid expansion' do
+
+    it 'can expand out wildcard amino acid combinations' do
+      array = subject.expand_peptides('ALXX', 'X' =>  %w(* % &), 'L' => %w(P Q) )
+      array.sort.should == %w(AP** AP*% AP*& AP%* AP%% AP%& AP&* AP&% AP&& AQ** AQ*% AQ*& AQ%* AQ%% AQ%& AQ&* AQ&% AQ&&).sort
+    end
+
+    it 'will not expand explosive combinations (>MAX_NUM_AA_EXPANSION)' do
+      # this is from real data
+      worst_case = 'LTLLRPEKHEAATGVDTICTHRVDPIGPGLXXEXLYWELSXLTXXIXELGPYTLDR'
+      subject.expand_peptides(worst_case, 'X' =>  %w(* % &)).nil?.should == true
+    end
+
+    it 'returns the peptide in the array if no expansion' do
+      array = subject.expand_peptides('ZZZZZ', 'X' =>  %w(* % &), 'L' => %w(P Q) )
+      array.should == ['ZZZZZ']
+    end
+  end
+
+  describe 'the commandline utility' do
+
+    before do
+      @fasta_file = [TESTFILES, path, 'uni_11_sp_tr.fasta'].join('/')
+      @output_file = [TESTFILES, path, "uni_11_sp_tr.msd_clvg2.min_aaseq4.yml"].join('/')
+    end
+
+    it 'converts a fasta file into peptide centric db' do
+      output_files = Mspire::Ident::Peptide::Db::Creator.cmdline([@fasta_file])
+      output_files.first.should == File.expand_path(@output_file)
+      File.exist?(@output_file).should == true
+      hash = {}
+      YAML.load_file(@output_file).each do |k,v|
+        hash[k] = v.split("\t")
+      end
+      sorted = hash.sort
+      # these are merely frozen, not perfectly defined
+      sorted.first.should == ["AAFDDAIAELDTLSEESYK", ["P62258"]]
+      sorted.last.should == ["YWCRLGPPRWICQTIVSTNQYTHHR", ["D2KTA8"]]
+      sorted.size.should == 728
+      #File.unlink(@output_file)
+    end
+
+    it 'lists approved enzymes and exits' do
+      output = capture_stdout do
+        begin
+          Mspire::Ident::Peptide::Db::Creator.cmdline(['--list-enzymes'])
+        rescue SystemExit
+          1.should == 1 # we exited
+        end
+      end
+      lines = output.split("\n")
+      lines.include?("trypsin").should == true
+      lines.include?("chymotrypsin").should == true
+    end
+  end
+end

data/spec/mspire/ident/peptide/db/io_spec.rb

@@ -0,0 +1,21 @@
+require 'spec_helper'
+
+require 'mspire/ident/peptide/db/io'
+
+describe 'reading a peptide centric DB' do
+  before do
+    @pepcentric = TESTFILES + "/mspire/ident/peptide/db/uni_11_sp_tr.PEPTIDE_CENTRIC.yml"
+  end
+
+  it 'reads the file on disk with random access or is enumerable' do
+    Mspire::Ident::Peptide::Db::IO.open(@pepcentric) do |io|
+      io["AVTEQGHELSNEER"].should == %w(sp|P31946|1433B_HUMAN	sp|P31946-2|1433B_HUMAN)
+      io["VRAAR"].should == ["tr|D3DX18|D3DX18_HUMAN"]
+      io["SILLY WILLY"].should be_nil
+      io.each_with_index do |key_prots, i|
+        key_prots.first.should be_an_instance_of String
+        key_prots.last.should be_a_kind_of Array
+      end
+    end
+  end
+end

data/spec/mspire/ident/peptide/db_spec.rb

@@ -4,105 +4,15 @@ require 'yaml'
 path = 'mspire/ident/peptide/db'
 require path 
 
-module Kernel
- 
-  def capture_stdout
-    out = StringIO.new
-    $stdout = out
-    yield
-    out.rewind
-    return out.read
-  ensure
-    $stdout = STDOUT
-  end
- 
-end
-
-
-describe 'a uniprot fasta file' do
-
+describe 'reading a peptide centric db' do
   before do
-    @fasta_file = [TESTFILES, path, 'uni_11_sp_tr.fasta'].join('/')
-  end
-
-  describe 'amino acid expansion' do
-
-    it 'can expand out wildcard amino acid combinations' do
-      array = Mspire::Ident::Peptide::Db.expand_peptides('ALXX', 'X' =>  %w(* % &), 'L' => %w(P Q) )
-      array.sort.should == %w(AP** AP*% AP*& AP%* AP%% AP%& AP&* AP&% AP&& AQ** AQ*% AQ*& AQ%* AQ%% AQ%& AQ&* AQ&% AQ&&).sort
-    end
-
-    it 'will not expand explosive combinations (>MAX_NUM_AA_EXPANSION)' do
-      # this is from real data
-      worst_case = 'LTLLRPEKHEAATGVDTICTHRVDPIGPGLXXEXLYWELSXLTXXIXELGPYTLDR'
-      Mspire::Ident::Peptide::Db.expand_peptides(worst_case, 'X' =>  %w(* % &)).nil?.should == true
-    end
-
-    it 'returns the peptide in the array if no expansion' do
-      array = Mspire::Ident::Peptide::Db.expand_peptides('ZZZZZ', 'X' =>  %w(* % &), 'L' => %w(P Q) )
-      array.should == ['ZZZZZ']
-    end
-
+    @pepcentric = TESTFILES + "/mspire/ident/peptide/db/uni_11_sp_tr.PEPTIDE_CENTRIC.yml"
   end
 
-  describe 'creating a peptide centric database' do
-    before do
-
-      #@output_file = [TESTFILES, path, 'uni_11_sp_tr.'].join('/')
-      @output_file = [TESTFILES, path, "uni_11_sp_tr.msd_clvg2.min_aaseq4.yml"].join('/')
-    end
-
-    it 'converts a fasta file into peptide centric db' do
-      output_files = Mspire::Ident::Peptide::Db.cmdline([@fasta_file])
-      output_files.first.should == File.expand_path(@output_file)
-      File.exist?(@output_file).should == true
-      hash = {}
-      YAML.load_file(@output_file).each do |k,v|
-        hash[k] = v.split("\t")
-      end
-      sorted = hash.sort
-      # these are merely frozen, not perfectly defined
-      sorted.first.should == ["AAFDDAIAELDTLSEESYK", ["sp|P62258|1433E_HUMAN"]]
-      sorted.last.should == ["YWCRLGPPRWICQTIVSTNQYTHHR", ["tr|D2KTA8|D2KTA8_HUMAN"]]
-      sorted.size.should == 728
-      File.unlink(@output_file)
-    end
-
-    it 'lists approved enzymes and exits' do
-      output = capture_stdout do
-        begin
-          Mspire::Ident::Peptide::Db.cmdline(['--list-enzymes'])
-        rescue SystemExit
-          1.should == 1 # we exited
-        end
-      end
-      lines = output.split("\n")
-      lines.include?("trypsin").should == true
-      lines.include?("chymotrypsin").should == true
-    end
-  end
-
-  describe 'reading a peptide centric database' do
-    before do
-      outfiles = Mspire::Ident::Peptide::Db.cmdline([@fasta_file])
-      @outfile = outfiles.first
-    end
-
-    it 'creates a hash that can retrieve peptides as an array' do
-      hash = Mspire::Ident::Peptide::Db.new(@outfile)
-      hash["AVTEQGHELSNEER"].should == %w(sp|P31946|1433B_HUMAN	sp|P31946-2|1433B_HUMAN)
-      hash["VRAAR"].should == ["tr|D3DX18|D3DX18_HUMAN"]
-    end
-
-    it 'reads the file on disk with random access or is enumerable' do
-      Mspire::Ident::Peptide::Db::IO.open(@outfile) do |io|
-        io["AVTEQGHELSNEER"].should == %w(sp|P31946|1433B_HUMAN	sp|P31946-2|1433B_HUMAN)
-        io["VRAAR"].should == ["tr|D3DX18|D3DX18_HUMAN"]
-        io.each_with_index do |key_prots, i|
-          key_prots.first.should be_an_instance_of String
-          key_prots.last.should be_a_kind_of Array
-        end
-      end
-    end
+  it 'creates a hash that can retrieve peptides as an array' do
+    hash = Mspire::Ident::Peptide::Db.new(@pepcentric)
+    hash["AVTEQGHELSNEER"].should == ["sp|P31946|1433B_HUMAN", "sp|P31946-2|1433B_HUMAN"]
+    hash["VRAAR"].should == ["tr|D3DX18|D3DX18_HUMAN"]
+    hash["BANNANA"].should == nil
   end
 end