def calc_lvafs(list_CLV_target_range_base_counts):
dict_CLV_bases = {'A':0,'C':1,'G':2,'T':3,'N':4}
list_CLV_target_range_LVAF = [(int_CLV_temp_range_position,division_zero_tolerant(max([list_CLV_temp_base_counts[1][int_CLV_i] for int_CLV_i in range(4) if int_CLV_i != dict_CLV_bases[list_CLV_temp_base_counts[0]]]), \
sum(list_CLV_temp_base_counts[1][:4]))) for int_CLV_temp_range_position, list_CLV_temp_base_counts in list_CLV_target_range_base_counts]
return list_CLV_target_range_LVAF
def calc_indel_score(list_CIS_target_position_base_counts,list_CIS_target_range_base_counts,int_CIS_target_position,list_CIS_target_base_change,float_CIS_LOD_stdev_test,float_CIS_LOD_stdev_cutoff_BE,int_CIS_NOP_target_range):
if len(list_CIS_target_base_change[0]) == len(list_CIS_target_base_change[1]):
raise ValueError('Indel not recognized')
#get indel sequence from base change
if len(list_CIS_target_base_change[0]) < len(list_CIS_target_base_change[1]):
int_CIS_indel_type = 0
str_CIS_indel_seq = list_CIS_target_base_change[1][1:]
else:
int_CIS_indel_type = 1
str_CIS_indel_seq = list_CIS_target_base_change[0][1:]
#get Nsupp from pileup
if str_CIS_indel_seq in list_CIS_target_position_base_counts[2][int_CIS_indel_type]:
int_CIS_target_Nsupp = list_CIS_target_position_base_counts[2][int_CIS_indel_type][str_CIS_indel_seq]
else:
int_CIS_target_Nsupp = 0
int_CIS_target_Rdepth = sum(list_CIS_target_position_base_counts[1][:4])
float_CIS_target_VAF = division_zero_tolerant(int_CIS_target_Nsupp,int_CIS_target_Rdepth)
list_CIS_target_range_LVAF = calc_lvafs(list_CIS_target_range_base_counts)
dict_CIS_target_formal_test = formal_testing(float_CIS_target_VAF,list_CIS_target_range_LVAF,int_CIS_target_position,float_CIS_LOD_stdev_test,float_CIS_LOD_stdev_cutoff_BE,int_CIS_NOP_target_range)
return [int_CIS_target_Nsupp,int_CIS_target_Rdepth,float_CIS_target_VAF,[dict_CIS_target_formal_test]]
def get_quality(dict_GQ_read, int_GQ_phred_cutoff):
list_GQ_low_phreds = [
1 if Chr2Phred(str_GQ_temp_chr) < int_GQ_phred_cutoff else 0
for str_GQ_temp_chr in dict_GQ_read['quals']
]
int_GQ_len_read = len(dict_GQ_read['read'])
float_GQ_low_q_fraction = division_zero_tolerant(sum(list_GQ_low_phreds),
int_GQ_len_read)
int_GQ_low_q_fraction_ceil = ceil(float_GQ_low_q_fraction * 100)
return int_GQ_low_q_fraction_ceil
def calc_tvafs(list_CTV_target_range_base_counts,*list_CTV_target_base_changes):
dict_CTV_bases = {'A':0,'C':1,'G':2,'T':3,'N':4}
dict_CTV_base_changes = {}
for list_CTV_temp_base_change in list_CTV_target_base_changes:
if not list_CTV_temp_base_change[0] in dict_CTV_base_changes:
dict_CTV_base_changes[list_CTV_temp_base_change[0]] = []
dict_CTV_base_changes[list_CTV_temp_base_change[0]].append(dict_CTV_bases[list_CTV_temp_base_change[1]])
list_CTV_target_range_TVAF = [(int_CTV_temp_range_position,division_zero_tolerant(max([list_CTV_temp_base_counts[1][int_CTV_i] for int_CTV_i in dict_CTV_base_changes[list_CTV_temp_base_counts[0]]]), \
sum(list_CTV_temp_base_counts[1][:4]))) for int_CTV_temp_range_position, list_CTV_temp_base_counts in list_CTV_target_range_base_counts \
if list_CTV_temp_base_counts[0] in dict_CTV_base_changes]
return list_CTV_target_range_TVAF
def calc_snp_score(list_CSS_target_position_base_counts,list_CSS_target_range_base_counts,int_CSS_target_position,list_CSS_target_base_change,float_CSS_LOD_stdev_test,float_CSS_LOD_stdev_cutoff_BE,int_CSS_NOP_target_range):
if not len(list_CSS_target_base_change[0]) == len(list_CSS_target_base_change[1]) == 1:
raise ValueError('SNP base change not recognized')
dict_CSS_bases = {'A':0,'C':1,'G':2,'T':3,'N':4}
int_CSS_target_Nsupp = list_CSS_target_position_base_counts[1][dict_CSS_bases[list_CSS_target_base_change[1]]]
int_CSS_target_Rdepth = sum(list_CSS_target_position_base_counts[1][:4])
float_CSS_target_VAF = division_zero_tolerant(int_CSS_target_Nsupp,int_CSS_target_Rdepth)
list_CSS_target_range_VAF = [calc_lvafs(list_CSS_target_range_base_counts),calc_tvafs(list_CSS_target_range_base_counts,list_CSS_target_base_change)]
list_CSS_target_formal_test = [formal_testing(float_CSS_target_VAF,list_CSS_temp_target_range_VAF,int_CSS_target_position,float_CSS_LOD_stdev_test,float_CSS_LOD_stdev_cutoff_BE,int_CSS_NOP_target_range) for list_CSS_temp_target_range_VAF in list_CSS_target_range_VAF]
return [int_CSS_target_Nsupp,int_CSS_target_Rdepth,float_CSS_target_VAF,list_CSS_target_formal_test]
def adjust_quality_distr(list_AQD_path_input_files, int_AQD_phred_cutoff,
int_AQD_input_max_low_q_percent,
int_AQD_max_read_filter_percent, int_AQD_read_chunks,
int_AQD_number_processes):
list_AQD_quality_distr = [0 for _ in range(101)]
with contextlib.ExitStack() as stack:
list_AQD_input_files = [stack.enter_context(open(str_AQD_temp_path_input_file, 'r')) for str_AQD_temp_path_input_file \
in list_AQD_path_input_files]
it_AQD_input_fastq = fastq_iterator(fastq_zip_equal(*[FastqGeneralIterator(obj_AQD_temp_input_file) \
for obj_AQD_temp_input_file in list_AQD_input_files]))
for list_AQD_reads_for_processing in iter_double_chunked(
it_AQD_input_fastq, int_AQD_read_chunks,
int_AQD_number_processes):
with concurrent.futures.ProcessPoolExecutor() as executor:
list_AQD_read_chunk_distr = executor.map(
quality_read_chunk, list_AQD_reads_for_processing,
itertools.repeat(int_AQD_phred_cutoff))
for list_AQD_temp_read_chunk_distr in list_AQD_read_chunk_distr:
list_AQD_quality_distr = [
int_AQD_temp_this_score + int_AQD_temp_total
for int_AQD_temp_this_score, int_AQD_temp_total in zip(
list_AQD_temp_read_chunk_distr,
list_AQD_quality_distr)
]
int_AQD_total_reads = sum(list_AQD_quality_distr)
int_AQD_reads_pot_filter = 0
for int_AQD_i in range(101):
int_AQD_reads_pot_filter += list_AQD_quality_distr[int_AQD_i]
int_AQD_min_low_q = int_AQD_i
if division_zero_tolerant(
int_AQD_reads_pot_filter, int_AQD_total_reads
) * 100 >= 100 - int_AQD_max_read_filter_percent:
break
if int_AQD_min_low_q > int_AQD_input_max_low_q_percent:
print('Adjusted filter criterion to ' + str(int_AQD_min_low_q) + '%')
return int_AQD_min_low_q
else:
return int_AQD_input_max_low_q_percent
def quality_filter(dict_QF_read, int_QF_phred_cutoff,
float_QF_max_low_q_fraction, int_QF_start_end_trim,
int_QF_end_end_trim, float_QF_end_trim_max_low_q_fraction):
list_QF_low_phreds = [
1 if Chr2Phred(str_QF_temp_chr) < int_QF_phred_cutoff else 0
for str_QF_temp_chr in dict_QF_read['quals']
]
int_QF_len_read = len(dict_QF_read['read'])
bool_QF_read_end_cut = False
#check if read meets basic quality criteria
float_QF_low_q_fraction = division_zero_tolerant(sum(list_QF_low_phreds),
int_QF_len_read)
if float_QF_low_q_fraction <= float_QF_max_low_q_fraction:
bool_QF_read_passed = True
int_QF_end_cut_point = int_QF_len_read
#check if quality at read end is low
int_QF_end_sum_low_q_bases = sum(
list_QF_low_phreds[int_QF_start_end_trim:])
for int_QF_i in range(int_QF_start_end_trim,
int_QF_len_read - int_QF_end_end_trim):
if int_QF_end_sum_low_q_bases / (
int_QF_len_read -
int_QF_i) > float_QF_end_trim_max_low_q_fraction:
int_QF_end_cut_point = int_QF_i
bool_QF_read_end_cut = True
break
int_QF_end_sum_low_q_bases -= list_QF_low_phreds[int_QF_i]
dict_QF_output_read = {
str_QF_temp_read_item:
dict_QF_read[str_QF_temp_read_item][:int_QF_end_cut_point]
for str_QF_temp_read_item in ['read', 'quals']
}
dict_QF_output_read['name'] = dict_QF_read['name']
else:
bool_QF_read_passed = False
dict_QF_output_read = dict_QF_read
return [bool_QF_read_passed, bool_QF_read_end_cut, dict_QF_output_read]
def get_mrd_scores(list_GMS_samples,float_GMS_LOD_stdev_test,float_GMS_LOD_stdev_cutoff_BE,int_GMS_NOP_target_range,int_GMS_R1_Nsupp_cutoff):
list_GMS_R1_tests = ['_R1.tsv','_R2.tsv']
list_GMS_EC_tests = ['_R1R2.tsv','_RF.tsv']
list_GMS_MRD_scores = []
for dict_GMS_temp_sample in list_GMS_samples:
list_GMS_temp_final_line = []
#general sample info
list_GMS_sample_ident = [dict_GMS_temp_sample['number'],dict_GMS_temp_sample['sample'],dict_GMS_temp_sample['primer'],f'{dict_GMS_temp_sample["target"]["chr"]}:' \
+ f'{dict_GMS_temp_sample["target"]["position"]}:{">".join(dict_GMS_temp_sample["target"]["base_change"])}']
#check if primer on target region
if dict_GMS_temp_sample['target_region']['bounds'][0] <= dict_GMS_temp_sample['target']['position'] <= dict_GMS_temp_sample['target_region']['bounds'][1] \
and dict_GMS_temp_sample['target']['chr'] == dict_GMS_temp_sample['target_region']['chr']:
#tests to run
#get MRD scores
list_GMS_EC_results = [CalcMRDScore.calc_mrd_score(dict_GMS_temp_sample['base_filename']+str_GMS_temp_file_end,dict_GMS_temp_sample['target']['chr'], \
dict_GMS_temp_sample['target_region']['bounds'][0],dict_GMS_temp_sample['target_region']['bounds'][1]-dict_GMS_temp_sample['target_region']['bounds'][0]+1,dict_GMS_temp_sample['target']['position'], \
dict_GMS_temp_sample['target']['base_change'],float_GMS_LOD_stdev_test,float_GMS_LOD_stdev_cutoff_BE,int_GMS_NOP_target_range) for str_GMS_temp_file_end in list_GMS_EC_tests]
list_GMS_R1_temp_results = [CalcMRDScore.calc_mrd_score(dict_GMS_temp_sample['base_filename']+str_GMS_temp_file_end,dict_GMS_temp_sample['target']['chr'], \
dict_GMS_temp_sample['target_region']['bounds'][0],dict_GMS_temp_sample['target_region']['bounds'][1]-dict_GMS_temp_sample['target_region']['bounds'][0]+1,dict_GMS_temp_sample['target']['position'], \
dict_GMS_temp_sample['target']['base_change'],float_GMS_LOD_stdev_test,float_GMS_LOD_stdev_cutoff_BE,int_GMS_NOP_target_range) for str_GMS_temp_file_end in list_GMS_R1_tests]
#check if R1 or R2 has more Nsupp
if list_GMS_R1_temp_results[0][0] >= list_GMS_R1_temp_results[1][0]:
list_GMS_R1_result = list_GMS_R1_temp_results[0]
str_GMS_large_indel_test = 'R1'
else:
list_GMS_R1_result = list_GMS_R1_temp_results[1]
str_GMS_large_indel_test = 'R2'
list_GMS_test_results = list_GMS_EC_results + [list_GMS_R1_result]
list_GMS_test_results_formatted = [format_mrd_output(list_GMS_temp_test_result) for list_GMS_temp_test_result in list_GMS_test_results]
#add R1/R2 and RF results
for list_GMS_temp_result_full_test in list_GMS_test_results_formatted[:2]:
list_GMS_temp_final_line += list_GMS_sample_ident + list_GMS_temp_result_full_test + ['#']
#add R1 results
list_GMS_temp_final_line += list_GMS_sample_ident + list_GMS_test_results_formatted[2][:3] + ['#',None]
##add final test columns
#indel-length
int_GMS_target_indel_length = max(map(len,dict_GMS_temp_sample['target']['base_change']))-1
list_GSM_final_decision = final_decision(list_GMS_test_results,int_GMS_target_indel_length,str_GMS_large_indel_test,int_GMS_R1_Nsupp_cutoff)
list_GMS_temp_final_line += list_GSM_final_decision + ['#',None]
#add additional info, RF distributions
#get read number
int_GMS_temp_fastq_read_number = fastq_get_read_number(dict_GMS_temp_sample['fastq_file'])
list_GMS_temp_RF_distr = get_RF_distr(dict_GMS_temp_sample['bam_file_RF'],3,0,32,2)
str_GMS_temp_RF_distr_join = " ".join([":".join(str(int_GMS_item) for int_GMS_item in tuple_GMS_item) for tuple_GMS_item in list_GMS_temp_RF_distr[3]])
list_GMS_temp_additional = [f'{dict_GMS_temp_sample["target_region"]["chr"]}:{dict_GMS_temp_sample["target_region"]["bounds"][0]}-{dict_GMS_temp_sample["target_region"]["bounds"][1]}', int_GMS_temp_fastq_read_number, \
division_zero_tolerant(list_GMS_test_results[0][1],int_GMS_temp_fastq_read_number),list_GMS_temp_RF_distr[0],list_GMS_temp_RF_distr[1],division_zero_tolerant(list_GMS_temp_RF_distr[0], \
list_GMS_temp_RF_distr[1]),list_GMS_temp_RF_distr[2], division_zero_tolerant(list_GMS_temp_RF_distr[1],list_GMS_temp_RF_distr[2]),str_GMS_temp_RF_distr_join]
list_GMS_temp_final_line += list_GMS_temp_additional
#target not on primer region
else:
list_GMS_temp_final_line = list_GMS_sample_ident + ['Target not on primer region']
list_GMS_MRD_scores.append(list_GMS_temp_final_line)
return list_GMS_MRD_scores