paco_translator.rb in pets-0.2.3

- old
+ new

@@ -5,37 +5,56 @@
 HPO_FILE = File.join(EXTERNAL_DATA, 'hp.obo')
 $: << File.expand_path(File.join(ROOT_PATH, '..', 'lib', 'pets'))
 
 require 'generalMethods.rb'
 require 'optparse'
+require 'semtools'
 
 ###############
 #METHODS
 ###############
 
-def translate_codes_to_terms(patient_data, hpo_storage)
-	patients_with_hpo_names = {}
-	hpo_names = []
-	patient_data.each do |patientID, hpos_and_cnvs|
-		hpos = hpos_and_cnvs.shift.split('|')
-		hpos.each do |hpo|
-			hpo_names << hpo_storage[hpo][1]
-		end
-		hpos_and_cnvs << hpo_names.join('|')
-		patients_with_hpo_names[patientID] = hpos_and_cnvs
-		hpo_names = []
-	end
-	return patients_with_hpo_names
+def translate_hpo(patient_data, hpo, translate)
+  reject_pats = []
+  patient_data.each do |patientID, patient_record|
+    hpos, chr, start, stop = patient_record
+    if translate == 'names'
+      # hpos, rejected = hpo.translate_codes2names(hpos)
+      hpos, rejected = hpo.translate_ids(hpos)
+    elsif translate =='codes'
+      # hpos, rejected = hpo.translate_names2codes(hpos)
+      hpos, rejected = hpo.translate_names(hpos)
+      STDERR.puts(" The ontology names '#{rejected.join(',')}' were not found") if !rejected.empty?
+    end
+    if hpos.empty?
+      reject_pats << patientID 
+    else
+      patient_record[0] = hpos
+    end
+  end
+  reject_pats.each do | rj_pat|
+    patient_data.delete(rj_pat)
+  end
 end
 
-def save_translated_file(patients_with_hpo_names, output_file)
-	handler = File.open(output_file, 'w')
-	patients_with_hpo_names.each do |id, data|
-		patientID = id.gsub(/_i[0-9]/,'')
-		handler.puts "#{patientID}\t#{data.join("\t")}"
-	end
-	handler.close
+def save_translated_file(patients_with_hpo_names, output_file, mode)
+  File.open(output_file, 'w') do |f|
+    if mode == 'paco'
+      f.puts "patient_id\tchr\tstart\tstop\tphenotypes"
+    end
+    patients_with_hpo_names.each do |id, patient_record|
+      hpos, chr, start, stop = patient_record
+      id = id.gsub(/_i[0-9]+$/,'')
+      if mode == 'default'
+        f.puts "#{id}\t#{hpos.join('|')}\t#{[chr, start, stop].join("\t")}"
+      elsif mode == 'paco'
+        f.puts "#{id}\t#{[chr, start, stop].join("\t")}\t#{hpos.join('|')}"
+      else
+        abort('Wrong save_mode] option, please try default or paco')
+      end
+    end
+  end
 end
 
 ###############
 #OPTIONS
 ###############
@@ -58,11 +77,11 @@
   opts.on("-e", "--end_col INTEGER/STRING", "Column name if header is true, otherwise 0-based position of the column with the end mutation coordinate") do |data|
     options[:end_col] = data
   end
   
   options[:header] = true
-  opts.on("-H", "--header", "Set if the file has a line header. Default true") do 
+  opts.on("-H", "--header", "File has a line header. Default true") do 
     options[:header] = false
   end
 
   options[:output_file] = 'paco_file_with_hpo_names.txt'
   opts.on("-o", "--output_file PATH", "Output paco file with HPO names") do |data|
@@ -74,29 +93,43 @@
     options[:input_file] = value
   end
 
   options[:hpo_col] = nil
   opts.on("-p", "--hpo_term_col INTEGER/STRING", "Column name if header true or 0-based position of the column with the HPO terms") do |data|
-  	options[:hpo_col] = data
+    options[:hpo_col] = data
   end
 
   options[:start_col] = nil
   opts.on("-s", "--start_col INTEGER/STRING", "Column name if header is true, otherwise 0-based position of the column with the start mutation coordinate") do |data|
-  	options[:start_col] = data
+    options[:start_col] = data
   end
 
+  options[:hpo_separator] = '|'
+  opts.on("-S", "--hpo_separator STRING", "Set which character must be used to split the HPO profile. Default '|'") do |data|
+    options[:hpo_separator] = data
+  end
 
+  options[:save_mode] = 'default'
+  opts.on("-m", "--save_mode STRING", "Set output data mode") do |data|
+    options[:save_mode] = data
+  end
+
+  options[:translate] = nil
+  opts.on("-t", "--translate STRING", "Set 'names' to translate from hpo codes to names or set 'codes' to translate from hpo names to codes. By default, ther is not translation") do |data|
+    options[:translate] = data
+  end
 end.parse!
 
-
 ###############
 #MAIN
 ###############
+hpo_file = ENV['hpo_file']
+hpo_file = HPO_FILE if hpo_file.nil?
 
-hpo_storage = load_hpo_file(HPO_FILE)
-patient_data, $patient_number = load_patient_cohort(options)
-patients_with_hpo_names = translate_codes_to_terms(patient_data, hpo_storage)
-
-save_translated_file(patients_with_hpo_names, options[:output_file])
-
-
-Process.exit	
+patient_data = load_patient_cohort(options)
+if !options[:translate].nil?
+  # hpo = Ontology.new
+  # hpo.load_data(hpo_file)
+  hpo = Ontology.new(file: hpo_file, load_file: true)
+  translate_hpo(patient_data, hpo, options[:translate])
+end
+save_translated_file(patient_data, options[:output_file], options[:save_mode])