pets 0.2.3 → 0.2.4

data/bin/paco_translator.rb

@@ -1,59 +1,10 @@
 #! /usr/bin/env ruby
 
 ROOT_PATH = File.dirname(__FILE__)
-EXTERNAL_DATA = File.expand_path(File.join(ROOT_PATH, '..', 'external_data'))
-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_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, 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
+require 'pets'
 
 ###############
 #OPTIONS
@@ -68,9 +19,9 @@ OptionParser.new do |opts|
     options[:chromosome_col] = data
   end
 
-  options[:pat_id_col] = nil
+  options[:id_col] = nil
   opts.on("-d", "--pat_id_col INTEGER/STRING", "Column name if header is true, otherwise 0-based position of the column with the patient id") do |data|
-    options[:pat_id_col] = data
+    options[:id_col] = data
   end
 
   options[:end_col] = nil
@@ -93,9 +44,9 @@ OptionParser.new do |opts|
     options[:input_file] = value
   end
 
-  options[:hpo_col] = nil
+  options[:ont_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[:ont_col] = data
   end
 
   options[:start_col] = nil
@@ -103,33 +54,39 @@ OptionParser.new do |opts|
     options[:start_col] = data
   end
 
-  options[:hpo_separator] = '|'
+  options[: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
+    options[:separator] = data
+  end
+
+  options[:n_phens] = nil
+  opts.on("--n_phens INTEGER", "Remove records with N or less phenotypes") do |data|
+    options[:n_phens] = data.to_i
   end
 
-  options[:save_mode] = 'default'
+  options[:save_mode] = :default
   opts.on("-m", "--save_mode STRING", "Set output data mode") do |data|
-    options[:save_mode] = data
+    options[:save_mode] = data.to_sym
   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
+  options[:names] = false
+  opts.on("-n", "--hpo_names", "Define if the input HPO are human readable names. Default false") do
+    options[:names] = true
+  end
+
+  options[:translate] = false
+  opts.on("-t", "--translate", "Set to translate from hpo codes to names. By default, ther is not translation") do 
+    options[:translate] = true
   end
 end.parse!
 
 ###############
 #MAIN
 ###############
-hpo_file = ENV['hpo_file']
-hpo_file = HPO_FILE if hpo_file.nil?
-
-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])
+hpo_file = !ENV['hpo_file'].nil? ? ENV['hpo_file'] : HPO_FILE
+Cohort.load_ontology(:hpo, hpo_file)
+Cohort.act_ont = :hpo
+
+patient_data, rejected_hpos, rejected_patients = Cohort_Parser.load(options)
+rejected_patients_by_phen = patient_data.filter_by_term_number(options[:n_phens]) if !options[:n_phens].nil?
+patient_data.save(options[:output_file], options[:save_mode], options[:translate])

data/bin/phen2reg.rb

@@ -2,16 +2,13 @@
 # Rojano E. & Seoane P., September 2016
 # Program to predict the position from given HPO codes, sorted by their association values.
 
-REPORT_FOLDER=File.expand_path(File.join(File.dirname(__FILE__), '..', 'templates'))
 ROOT_PATH = File.dirname(__FILE__)
 $: << File.expand_path(File.join(ROOT_PATH, '..', 'lib', 'pets'))
 
-require 'generalMethods.rb'
-require 'phen2reg_methods.rb'
 require 'optparse'
 require 'report_html'
 require 'semtools'
-
+require 'pets'
 
 ##########################
 #OPT-PARSER

data/bin/profiles2phenopacket.rb

@@ -0,0 +1,110 @@
+#! /usr/bin/env ruby
+
+ROOT_PATH = File.dirname(__FILE__)
+$: << File.expand_path(File.join(ROOT_PATH, '..', 'lib', 'pets'))
+
+require 'optparse'
+require 'pets'
+
+#############################
+## METHODS
+#############################
+def load_index(path_index)
+  vcf_index = {}
+  File.open(path_index).each do |line|
+    id, path = line.chomp.split("\t")
+    vcf_index[id] = path
+  end
+  return vcf_index
+end
+
+
+##########################
+#OPT-PARSER
+##########################
+
+options = {}
+OptionParser.new do |opts|
+  opts.banner = "Usage: #{__FILE__} [options]"
+
+  options[:chromosome_col] = nil
+  opts.on("-c", "--chromosome_col INTEGER/STRING", "Column name if header is true, otherwise 0-based position of the column with the chromosome") do |data|
+    options[:chromosome_col] = data
+  end
+
+  options[:id_col] = nil
+  opts.on("-d", "--pat_id_col INTEGER/STRING", "Column name if header is true, otherwise 0-based position of the column with the patient id") do |data|
+    options[:id_col] = data
+  end
+
+  options[:end_col] = nil
+  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[:genome_assembly] = 'hg38'
+  opts.on("-G", "--genome_assembly STRING", "Genome assembly version. Please choose between hg18, hg19 and hg38. Default hg38") do |data|
+    options[:genome_assembly] = data
+  end
+
+  options[:header] = true
+  #chr\tstart\tstop
+  opts.on("-H", "--header", "Set if the file has a line header. Default true") do 
+    options[:header] = false
+  end
+
+  options[:input_file] = nil
+  opts.on("-i", "--input_file PATH", "Input file with patient data") do |data|
+    options[:input_file] = data
+  end
+
+  options[:vcf_index] = nil
+  opts.on("-I", "--vcf_index PATH", "VCF file with patient id pointing to vcf path") do |data|
+    options[:vcf_index] = data
+  end
+
+  options[:names] = false
+  opts.on("-n", "--hpo_names", "Define if the input HPO are human readable names. Default false") do
+    options[:names] = true
+  end
+
+  options[:output_folder] = nil
+  opts.on("-o", "--output_file PATH", "Output folder") do |data|
+    options[:output_folder] = data
+  end
+
+  options[:ont_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[:ont_col] = data
+  end
+
+  options[:separator] = '|'
+  opts.on("-S", "--hpo_separator STRING", "Set which character must be used to split the HPO profile. Default '|'") do |data|
+  	options[:separator] = 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
+  end
+
+  opts.on_tail("-h", "--help", "Show this message") do
+    puts opts
+    exit
+  end
+
+end.parse!
+
+#############################################################
+## MAIN
+#############################################################
+hpo_file = !ENV['hpo_file'].nil? ? ENV['hpo_file'] : HPO_FILE
+Cohort.load_ontology(:hpo, hpo_file, options[:excluded_hpo])
+Cohort.act_ont = :hpo
+
+patient_data, rejected_hpos_L, rejected_patients_L = Cohort_Parser.load(options)
+rejected_hpos_C, rejected_patients_C = patient_data.check(hard=true)
+patient_data.link2ont(Cohort.act_ont)
+
+vcf_index = load_index(options[:vcf_index]) if !options[:vcf_index].nil?
+patient_data.export_phenopackets(options[:output_folder], options[:genome_assembly], vcf_index: vcf_index)

data/bin/reg2phen.rb

@@ -3,14 +3,12 @@
 #data2predict = file to predict
 #training_file.txt = file with training data (association values and hpo codes).
 
-REPORT_FOLDER=File.expand_path(File.join(File.dirname(__FILE__), '..', 'templates'))
 ROOT_PATH = File.dirname(__FILE__)
 $: << File.expand_path(File.join(ROOT_PATH, '..', 'lib', 'pets'))
 
 require 'optparse'
-require 'generalMethods.rb'
 require 'semtools'
-require 'reg2phen_methods'
+require 'pets'
 
 ##########################
 #OPT-PARSER