cohort_parser.rb in pets-0.2.5

- old
+ new

@@ -1,10 +1,11 @@
 class Cohort_Parser
 	def self.load(options)
 		fields2extract = get_fields2extract(options)
 		field_numbers = fields2extract.values
 		records = read_records(options, fields2extract, field_numbers)
+		options[:extracted_fields] = fields2extract.keys
 		cohort, rejected_terms, rejected_recs = create_cohort(records, options)
 		return cohort, rejected_terms, rejected_recs
 	end 
 
 	def self.read_records(options, fields2extract, field_numbers)
@@ -44,11 +45,11 @@
 		return records
 	end
 
 	def self.get_fields2extract(options)
 		fields2extract = {}
-		[:id_col, :ont_col, :chromosome_col, :start_col, :end_col].each do |field|
+		[:id_col, :ont_col, :chromosome_col, :start_col, :end_col, :sex_col].each do |field|
 			col = options[field]
 			if !col.nil?
 				col = col.to_i if !options[:header]
 				fields2extract[field] = col
 			end
@@ -68,11 +69,11 @@
 		rejected_recs = []
 		cohort = Cohort.new()
 		records.each do |id, record|
 			rec = record.first
 			terms = rec.first
-			if options[:names]
+			if options[:names] # Translate hpo names 2 codes
 				init_term_number = terms.length
 				terms, rec_rejected_terms = ont.translate_names(terms)
 				if !rec_rejected_terms.empty?
 					STDERR.puts "WARNING: record #{id} has the unknown term NAMES '#{rec_rejected_terms.join(',')}'. Terms removed."
 					rejected_terms.concat(rec_rejected_terms)
@@ -85,10 +86,14 @@
 			if rec.length > 1 # there is genomic region attributes
 				variants = record.map{|v| v[1..3] }
 			else
 				variants = [] # Not exists genomic region attributes so we create a empty array
 			end
-			cohort.add_record([id, terms, check_variants(variants)])
+			other_attr = {}
+			if options[:extracted_fields].include?(:sex_col) # Check for additional attributes. -1 is applied to ignore :id in extracted fields
+				other_attr[:sex] = record.first[options[:extracted_fields].index(:sex_col) -1]
+			end
+			cohort.add_record([id, terms, check_variants(variants)], other_attr)
 		end
 		return cohort, rejected_terms.uniq, rejected_recs
 	end
 
 	def self.check_variants(vars)