ms-error_rate 0.0.10 → 0.0.11

data/VERSION

@@ -1 +1 @@
-0.0.10
+0.0.11

data/bin/mascot_pepxml_to_peptide_hit_qvalues.rb

@@ -4,15 +4,20 @@ require 'trollop'
 require 'nokogiri'
 require 'set'
 
+require 'ms/ident/pepxml'
 require 'ms/ident/peptide_hit/qvalue'
-require 'ms/error_rate/qvalue/pepxml'
+
+EXT = Ms::Ident::PeptideHit::Qvalue::FILE_EXTENSION
+combine_base  = "combined"
 
 opts = Trollop::Parser.new do
-  banner %Q{usage: #{File.basename(__FILE__)} <fwd>.xml <decoy>.xml ...
-outputs: <fwd>.phq.csv
+  banner %Q{usage: #{File.basename(__FILE__)} <target>.xml <decoy>.xml ...
+outputs: <fwd>.phq.tsv
 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 :decoy_first, "decoy files are before target files", :default => false
   opt :verbose, "be verbose", :default => false
 end
 
@@ -22,16 +27,37 @@ if ARGV.size == 0 || (ARGV.size%2 != 0)
   opts.educate
   exit 
 end
+
 $VERBOSE = opt.delete(:verbose)
 
 files = ARGV.to_a
-files.each_slice(2).map do |target, decoy|
-  hit_qvalue_pairs = Ms::ErrorRate::Qvalue::Pepxml.target_decoy_qvalues(target, decoy, opt, &:ionscore)
+
+files.each_slice(2).map!(&:reverse) if opt[:decoy_first]
+
+groups_of_search_hits = files.map do |file|
+  Ms::Ident::Pepxml.search_hits(file)
+end
+
+to_run = {}
+if opt[:combine]
+  all_target = [] ; all_decoy = []
+  groups_of_search_hits.each_slice(2) do |target_hits,decoy_hits| 
+    all_target.push(*target_hits) ; all_decoy.push(*decoy_hits)
+  end
+  to_run[combine_base] = [all_target, all_decoy]
+else
+  groups_of_search_hits.each_slice(2).zip(files) do |target_hits_and_decoy_hits, file|
+    to_run[file.chomp(File.extname(file)) + EXT] = target_hits_and_decoy_hits
+  end
+end
+
+to_run.each do |file, target_decoy_pair|
+  hit_qvalue_pairs = Ms::ErrorRate::Qvalue.target_decoy_qvalues(target_decoy_pair.first, target_decoy_pair.last, :z_together => opt[:z_together])
   hits = [] ; qvals = []
   hit_qvalue_pairs.each do |hit, qval|
     hits << hit ; qvals << qval
   end
-  outfile = Ms::Ident::PeptideHit::Qvalue.to_phq(target.chomp(File.extname(target)), hits, qvals)
+  outfile = Ms::Ident::PeptideHit::Qvalue.to_file(file, hits, qvals)
   puts "created: #{outfile}" if $VERBOSE
 end
 

metadata

@@ -5,8 +5,8 @@ version: !ruby/object:Gem::Version
   segments: 
   - 0
   - 0
-  - 10
-  version: 0.0.10
+  - 11
+  version: 0.0.11
 platform: ruby
 authors: 
 - John Prince
@@ -127,7 +127,6 @@ files:
 - lib/ms/error_rate/qvalue.rb
 - lib/ms/error_rate/qvalue/mascot.rb
 - lib/ms/error_rate/qvalue/mascot/percolator.rb
-- lib/ms/error_rate/qvalue/pepxml.rb
 - lib/ms/error_rate/sbv.rb
 - lib/ms/error_rate/sbv/peptide_based.rb
 - lib/ms/error_rate/sbv/protein_based.rb

data/lib/ms/error_rate/qvalue/pepxml.rb

@@ -1,52 +0,0 @@
-require 'ms/error_rate/qvalue'
-
-module Ms ; end
-module Ms::ErrorRate ; end
-module Ms::ErrorRate::Qvalue ; end
-
-module Ms::ErrorRate::Qvalue::Pepxml
-  module_function
-
-  # returns an array of hit and qvalue pairs
-  # retrieves the aaseq, charge, and all search_score keys and values for use
-  # in the search_hit.  caller must provide a sort_by block, where the best
-  # hits are last.  charge is an integer, and all other search scores are cast
-  # as floats.  returns the output filename.
-  def target_decoy_qvalues(target_pepxml, decoy_pepxml, opt={}, &sort_by)
-
-    # this is a list of high quality peptide hits associated with each group
-    fields = [:aaseq, :charge]
-    ss_names = []
-    have_ss_names = false
-    (target_hits, decoy_hits) = [target_pepxml, decoy_pepxml].map do |file|
-      # begin with aaseq, charge
-      File.open(file) do |io|
-        doc = Nokogiri::XML.parse(io, nil, nil, Nokogiri::XML::ParseOptions::DEFAULT_XML | Nokogiri::XML::ParseOptions::NOBLANKS)
-        # we can work with namespaces, or just remove them ...
-        doc.remove_namespaces!
-        root = doc.root
-        search_hits = root.xpath('//search_hit')
-        search_hits.map do |search_hit| 
-          aaseq = search_hit['peptide']
-          charge = search_hit.parent.parent['assumed_charge'].to_i
-          search_score_nodes = search_hit.children.select {|node| node.name == 'search_score' }
-          ss_values = []
-          search_score_nodes.each do |node|
-            ss_names << node['name'].to_sym unless have_ss_names
-            ss_values << node['value'].to_f
-          end
-          have_ss_names = true
-          [aaseq, charge, *ss_values]
-        end
-      end
-    end
-
-    fields.push(*ss_names)
-
-    peptide_hit_class = Struct.new(*fields)
-    (t_hits, d_hits) = [target_hits, decoy_hits].map {|hits| hits.map {|hit_values| peptide_hit_class.new(*hit_values) } }
-
-    # hit and qvalue pairs
-    Ms::ErrorRate::Qvalue.target_decoy_qvalues(t_hits, d_hits, :z_together => opt[:z_together], &sort_by)
-  end
-end