def pass_fail_matrix():
""" Constructs the stagewise pass/fail matrix from the error_log. """
err_log = wd.err_log_path
matrix_path, junk = os.path.split(err_log)
matrix_path += "/yap_pass_fail_matrix.log"
sample_log_dict = {}
pass_fail = {}
for i in glob.glob(err_log + "/*"):
path, file = os.path.split(i)
file, ext = os.path.splitext(file)
sample_log_dict[file.rstrip("_err")] = yap_file_io.read_file(i) # len-1 chunks
for i in sample_log_dict.keys():
pass_fail_dict(sample_log_dict[i], pass_fail)
if not set(pass_fail.keys()).issubset(set(['CUFFDIFF', 'CUFFCOMPARE', 'CUFFMERGE', 'MACS2'])):
if wd.run_preprocess_analysis == "yes" and pass_fail.get("PREPROCESS") is None:
pass_fail["PREPROCESS"] = "FAIL"
elif wd.run_preprocess_analysis == "no":
pass_fail["PREPROCESS"] = "N/A"
if wd.run_reference_alignment == "yes" and pass_fail.get("ALIGNMENT") is None:
pass_fail["ALIGNMENT"] = "FAIL"
elif wd.run_reference_alignment == "no":
pass_fail["ALIGNMENT"] = "N/A"
if wd.run_postprocess_analysis == "yes" and pass_fail.get("POSTPROCESS") is None:
pass_fail["POSTPROCESS"] = "FAIL"
elif wd.run_postprocess_analysis == "no":
pass_fail["POSTPROCESS"] = "N/A"
if pass_fail["PREPROCESS"] == "FAIL":
pass_fail["ALIGNMENT"] = "FAIL"
pass_fail["POSTPROCESS"] = "FAIL"
sample_log_dict[i] = pass_fail
pass_fail = {}
print_matrix(sample_log_dict, matrix_path)
def read_barcodes(bar_file):
'''Reads barcode file and generates dictionary'''
bar_arr = yap_file_io.read_file(bar_file)
bar1 = []
barcode_dict = {}
for b in range(len(bar_arr)):
matchobj = re.search('\s*(\w*)(\s*).*', bar_arr[b].strip('\n'),
re.M | re.I)
if matchobj:
split_by = matchobj.group(2)
barcode = bar_arr[b].split(split_by)[1]
bar_id = bar_arr[b].split(split_by)[0]
barcode_dict[bar_id] = barcode
barcode_dict['unmatched'] = 'barcode_unmatched'
return barcode_dict
def read_barcodes(bar_file):
'''Reads barcode file and generates dictionary'''
bar_arr = yap_file_io.read_file(bar_file)
bar1 = []
barcode_dict = {}
for b in range(len(bar_arr)):
matchobj = re.search(
'\s*(\w*)(\s*).*', bar_arr[b].strip('\n'), re.M | re.I)
if matchobj:
split_by = matchobj.group(2)
barcode = bar_arr[b].split(split_by)[1]
bar_id = bar_arr[b].split(split_by)[0]
barcode_dict[bar_id] = barcode
barcode_dict['unmatched'] = 'barcode_unmatched'
return barcode_dict
def sample_filter():
""" Filters failed samples after the alignment step. """
ignore_list = []
inp_files_list=[]
list_of_samples=[]
list_of_samples_to_compare=[]
for i in wd.inp_files_list:
pass_fail = {}
err_log = wd.err_log_path + "/" + i[2] + "_err.log"
if os.path.exists(err_log):
pass_fail_dict(yap_file_io.read_file(err_log), pass_fail)
if 'FAIL' in pass_fail.itervalues():
ignore_list.append(i)
if len(ignore_list) != 0:
list_of_samples_to_compare = remove_corrupted_samples(wd.list_of_samples_to_compare,ignore_list)
list_of_samples = remove_corrupted_samples(wd.list_of_samples, ignore_list)
inp_files_list = [i for i in wd.inp_files_list if i not in ignore_list]
return inp_files_list,list_of_samples,list_of_samples_to_compare,ignore_list
else:
return wd.inp_files_list, wd.list_of_samples, wd.list_of_samples_to_compare,ignore_list
def sample_filter():
""" Filters failed samples after the alignment step. """
ignore_list = []
inp_files_list = []
list_of_samples = []
list_of_samples_to_compare = []
for i in wd.inp_files_list:
pass_fail = {}
err_log = wd.err_log_path + "/" + i[2] + "_err.log"
if os.path.exists(err_log):
pass_fail_dict(yap_file_io.read_file(err_log), pass_fail)
if 'FAIL' in pass_fail.itervalues():
ignore_list.append(i)
if len(ignore_list) != 0:
list_of_samples_to_compare = remove_corrupted_samples(
wd.list_of_samples_to_compare, ignore_list)
list_of_samples = remove_corrupted_samples(wd.list_of_samples,
ignore_list)
inp_files_list = [i for i in wd.inp_files_list if i not in ignore_list]
return inp_files_list, list_of_samples, list_of_samples_to_compare, ignore_list
else:
return wd.inp_files_list, wd.list_of_samples, wd.list_of_samples_to_compare, ignore_list
def pass_fail_matrix():
""" Constructs the stagewise pass/fail matrix from the error_log. """
err_log = wd.err_log_path
matrix_path, junk = os.path.split(err_log)
matrix_path += "/yap_pass_fail_matrix.log"
sample_log_dict = {}
pass_fail = {}
for i in glob.glob(err_log + "/*"):
path, file = os.path.split(i)
file, ext = os.path.splitext(file)
sample_log_dict[file.rstrip("_err")] = yap_file_io.read_file(
i) # len-1 chunks
for i in sample_log_dict.keys():
pass_fail_dict(sample_log_dict[i], pass_fail)
if not set(pass_fail.keys()).issubset(
set(['CUFFDIFF', 'CUFFCOMPARE', 'CUFFMERGE', 'MACS2'])):
if wd.run_preprocess_analysis == "yes" and pass_fail.get(
"PREPROCESS") is None:
pass_fail["PREPROCESS"] = "FAIL"
elif wd.run_preprocess_analysis == "no":
pass_fail["PREPROCESS"] = "N/A"
if wd.run_reference_alignment == "yes" and pass_fail.get(
"ALIGNMENT") is None:
pass_fail["ALIGNMENT"] = "FAIL"
elif wd.run_reference_alignment == "no":
pass_fail["ALIGNMENT"] = "N/A"
if wd.run_postprocess_analysis == "yes" and pass_fail.get(
"POSTPROCESS") is None:
pass_fail["POSTPROCESS"] = "FAIL"
elif wd.run_postprocess_analysis == "no":
pass_fail["POSTPROCESS"] = "N/A"
if pass_fail["PREPROCESS"] == "FAIL":
pass_fail["ALIGNMENT"] = "FAIL"
pass_fail["POSTPROCESS"] = "FAIL"
sample_log_dict[i] = pass_fail
pass_fail = {}
print_matrix(sample_log_dict, matrix_path)
workflow_config_file = sys.argv[2]
run_mode = "--check"
else:
print_usage = 'True'
else:
print_usage = 'True'
if print_usage == 'True':
print "Options : "
print " To print help : ", " yap --help or yap -h"
print " To do configuration file format check : ", " yap --check [workflow_configuration_filename], ", "\"eg: yap --check workflow_configuration.cfg \""
print " To run YAP: ", " yap -n [number of processors] [workflow_configuration_filename], ", "\"eg: yap -n 2 workflow_configuration.cfg \"", "(yap run with 2 processors)"
print "exiting the program"
exit_status = 'True'
if exit_status != 'True':
#reads the main workflow configuration file
workflow_file_data = yap_file_io.read_file(workflow_config_file)
if len(workflow_file_data) > 0:
#pass the file data to workflow parser
workflow_struct, workflow_errorlist = yap_tools.workflow_parser(workflow_file_data, workflow_config_file,nprocs)
if len(workflow_struct) < 2 or len(workflow_errorlist) > 0:
print"Format Error : while parsing the workflow configuration file= ", workflow_config_file, "\n"
print"YAP analysis general metadata mising or syntax error.", "\n"
print"Note:Use symbol(:begin) and (:end) to define command sections,enclose variable and corresponding values in double quotes(eg \"..\")"
print"Use (:=) in variable value assignment (eg \"variable\" := \"value\")"
print"To add comments,start the line with symbol(#)"
for i in workflow_errorlist:
print i
exit_status = 'True'
if exit_status != 'True':
if len(workflow_struct) != 0:
#check if the workflow configuration passes the sanity checks