make sample.txt file given a list of TCRalpha and TCRBeta barcodes

make_BC_list_Oct27_2017_final.py

from __future__ import print_function
import re
import subprocess
import sys

file_in = sys.argv[1]
#file_out = sys.argv[2]

### define functions
def num_ret(str=str):
	return(''.join([s for s in str if s.isdigit()]))

################ this will take a file in the format DBID;TCR_A;TCR_A_BC;TCR_B;TCR_B_BC and process it by creating dictionary for each HS run
def HS_spool(file_in):
	HS_calls ={}
	with open(file_in) as f_in:
		for line in f_in:
			line_parse = line.strip('\r\n').split(';')
			TCR_A = re.sub('TCR-', '',line_parse[1])
			TCR_A_BC = num_ret(line_parse[2])
			TCR_B = re.sub('TCR-', '', line_parse[3])
			TCR_B_BC = num_ret(line_parse[4])
			DBID = line_parse[0]
			if TCR_A and TCR_B:
					make_value_B = str(DBID) + ';'+TCR_B+';'+TCR_B_BC
					make_value_A = str(DBID) + ';'+TCR_A+';'+TCR_A_BC
					if TCR_B in HS_calls:
						get_exist = HS_calls.get(TCR_B)
						HS_calls[TCR_B] = get_exist + '$' +make_value_B
					else:
						HS_calls[TCR_B] = make_value_B
					if TCR_A in HS_calls:
						get_exist = HS_calls.get(TCR_A)
						HS_calls[TCR_A] = get_exist + '$' +make_value_A
					else:
						HS_calls[TCR_A] = make_value_A
			elif TCR_A:
				make_value = str(DBID) + ';'+TCR_A+';'+TCR_A_BC
				if TCR_A in HS_calls:
					get_exist = HS_calls.get(TCR_A)
					HS_calls[TCR_A] = get_exist + '$' +make_value
				else:
					HS_calls[TCR_A] = make_value
			else:
				make_value = str(DBID) + ';'+TCR_B+';'+TCR_B_BC
				if TCR_B in HS_calls:
					get_exist = HS_calls.get(TCR_B)
					HS_calls[TCR_B] = get_exist + '$' +make_value
				else:
					HS_calls[TCR_B] = make_value
	print ('IDENTIFIED   ', len(HS_calls), '   RUNS' )
	return(HS_calls)


################# this wil take the HS_dic created by HS_spool and write out barcodes from same DBID and same HS run in one line separated by commas
def out_f(Dic, file_out):
	f_out = open(file_out+'_SAMPLES.txt', 'w')
	for value in Dic.itervalues():
		if value:
			f_out.write(value+'\n')
	f_out.close()

## main call from python
def main(argv):
	HS_calls=HS_spool(file_in)
	for key, value in HS_calls.iteritems():
		value_parse = value.split('$')
		DBID_DIC ={}
		for item in value_parse:
			item_parse = item.split(';')
			if item_parse[2]:
				if item_parse[0] in DBID_DIC:
					get_exist = DBID_DIC.get(item_parse[0])
					DBID_DIC[item_parse[0]] = get_exist+','+item_parse[2]
				else:
					DBID_DIC[item_parse[0]] = item_parse[2]
		print ('IDENTIFIED  ', len(DBID_DIC), '  Individuals from ', key)
		out_f(DBID_DIC, key)


if __name__ == '__main__':main(sys.argv)