def merge_alignment_stats(dirs, BASECALLER_RESULTS, ALIGNMENT_RESULTS, flows):
datasets_json = {}
try:
f = open(os.path.join(BASECALLER_RESULTS, "datasets_basecaller.json"),
'r')
datasets_json = json.load(f)
f.close()
except:
printtime("ERROR: problem parsing %s" %
os.path.join(BASECALLER_RESULTS, "datasets_basecaller.json"))
traceback.print_exc()
return
for dataset in datasets_json['datasets']:
# What needs merging:
# - alignment.summary
# - alignTable.txt
# Some time in the future:
# - alignStats_err.json
# Merge alignStats metrics
try:
input_prefix_list = [
os.path.join(dir, ALIGNMENT_RESULTS,
dataset['file_prefix'] + '.') for dir in dirs
]
input_prefix_list = [
prefix for prefix in input_prefix_list
if os.path.exists(prefix + 'alignment.summary')
]
composite_prefix = os.path.join(ALIGNMENT_RESULTS,
dataset['file_prefix'] + '.')
if input_prefix_list:
mergeAlignStatsResults(input_prefix_list, composite_prefix)
else:
printtime("Nothing to merge: " + dataset['file_prefix'])
except:
printtime("ERROR: merging %s stats unsuccessful" %
(dataset['file_prefix'] + '.bam'))
datasets_basecaller = {}
try:
f = open(os.path.join(BASECALLER_RESULTS, "datasets_basecaller.json"),
'r')
datasets_basecaller = json.load(f)
f.close()
except:
printtime("ERROR: problem parsing %s" %
os.path.join(BASECALLER_RESULTS, "datasets_basecaller.json"))
traceback.print_exc()
return
try:
graph_max_x = int(50 * math.ceil(0.014 * int(flows)))
except:
graph_max_x = 400
input_prefix_list = []
for dataset in datasets_basecaller["datasets"]:
printtime("Barcode processing, rename")
src = os.path.join(ALIGNMENT_RESULTS,
dataset['file_prefix'] + '.alignment.summary')
if os.path.exists(src):
input_prefix_list.append(
os.path.join(ALIGNMENT_RESULTS, dataset['file_prefix'] + '.'))
#terrible hack to make aggregate_alignment happy
X_name = 'nomatch'
read_group = dataset['read_groups'][0]
if 'barcode_name' in datasets_basecaller['read_groups'][
read_group]:
X_name = datasets_basecaller['read_groups'][read_group][
'barcode_name']
dst = os.path.join(ALIGNMENT_RESULTS,
'alignment_%s.summary' % X_name)
try:
os.symlink(os.path.relpath(src, os.path.dirname(dst)), dst)
except:
printtime("ERROR: Unable to symlink '%s' to '%s'" % (src, dst))
# Merge alignStats_err.json right here!
merged_align_stats = {}
align_stats_num_bases = 400
for dir in dirs:
current_align_stats = {}
try:
f = open(
os.path.join(dir, ALIGNMENT_RESULTS, 'alignStats_err.json'),
'r')
current_align_stats = json.load(f)
f.close()
except:
printtime(
"Merge alignStats_err.json: skipping %s" %
os.path.join(dir, ALIGNMENT_RESULTS, 'alignStats_err.json'))
continue
if not merged_align_stats:
merged_align_stats = current_align_stats
align_stats_num_bases = len(
merged_align_stats.get("read_length", []))
continue
for idx in range(align_stats_num_bases):
merged_align_stats['nread'][idx] += current_align_stats['nread'][
idx]
merged_align_stats['unaligned'][idx] += current_align_stats[
'unaligned'][idx]
merged_align_stats['filtered'][idx] += current_align_stats[
'filtered'][idx]
merged_align_stats['clipped'][idx] += current_align_stats[
'clipped'][idx]
merged_align_stats['aligned'][idx] += current_align_stats[
'aligned'][idx]
merged_align_stats['n_err_at_position'][
idx] += current_align_stats['n_err_at_position'][idx]
merged_align_stats['cum_aligned'][idx] += current_align_stats[
'cum_aligned'][idx]
merged_align_stats['cum_err_at_position'][
idx] += current_align_stats['cum_err_at_position'][idx]
merged_align_stats['accuracy_total_bases'] += current_align_stats[
'accuracy_total_bases']
merged_align_stats['accuracy_total_errors'] += current_align_stats[
'accuracy_total_errors']
merged_align_stats['total_mapped_target_bases'] += current_align_stats[
'total_mapped_target_bases']
merged_align_stats['total_mapped_reads'] += current_align_stats[
'total_mapped_reads']
try:
f = open(os.path.join(ALIGNMENT_RESULTS, 'alignStats_err.json'), "w")
json.dump(merged_align_stats, f, indent=4)
f.close()
except:
printtime("ERROR; Failed to write merged alignStats_err.json")
traceback.print_exc()
mergeAlignStatsResults(input_prefix_list, ALIGNMENT_RESULTS + "/")
try:
base_error_plot.generate_base_error_plot(
os.path.join(ALIGNMENT_RESULTS, 'alignStats_err.json'),
os.path.join(ALIGNMENT_RESULTS, 'base_error_plot.png'),
int(graph_max_x))
ionstats_plots.alignment_rate_plot(
os.path.join(ALIGNMENT_RESULTS, 'alignStats_err.json'),
os.path.join(BASECALLER_RESULTS, 'ionstats_basecaller.json'),
os.path.join(ALIGNMENT_RESULTS, 'alignment_rate_plot.png'),
int(graph_max_x))
printtime("Base error plot has been created successfully")
except:
printtime("ERROR: Failed to generate base error plot")
traceback.print_exc()
# Generate alignment_barcode_summary.csv
barcodelist_path = 'barcodeList.txt'
if not os.path.exists(barcodelist_path):
barcodelist_path = '../barcodeList.txt'
if not os.path.exists(barcodelist_path):
barcodelist_path = '../../barcodeList.txt'
if not os.path.exists(barcodelist_path):
barcodelist_path = '../../../barcodeList.txt'
if not os.path.exists(barcodelist_path):
barcodelist_path = '../../../../barcodeList.txt'
if os.path.exists(barcodelist_path):
printtime("Barcode processing, aggregate")
aggregate_alignment("./", barcodelist_path)
def alignment_post_processing(BASECALLER_RESULTS, ALIGNMENT_RESULTS, flows,
mark_duplicates, force_alignstats):
datasets_basecaller = {}
try:
f = open(os.path.join(BASECALLER_RESULTS, "datasets_basecaller.json"),
'r')
datasets_basecaller = json.load(f)
f.close()
except:
printtime("ERROR: problem parsing %s" %
os.path.join(BASECALLER_RESULTS, "datasets_basecaller.json"))
traceback.print_exc()
return
try:
graph_max_x = int(50 * math.ceil(0.014 * int(flows)))
except:
graph_max_x = 400
input_prefix_list = []
for dataset in datasets_basecaller["datasets"]:
if not os.path.exists(
os.path.join(BASECALLER_RESULTS, dataset['basecaller_bam'])):
continue
printtime("Barcode processing, rename")
src = os.path.join(ALIGNMENT_RESULTS,
dataset['file_prefix'] + '.alignment.summary')
if os.path.exists(src):
input_prefix_list.append(
os.path.join(ALIGNMENT_RESULTS, dataset['file_prefix'] + '.'))
#terrible hack to make aggregate_alignment happy
X_name = 'nomatch'
read_group = dataset['read_groups'][0]
if 'barcode_name' in datasets_basecaller['read_groups'][
read_group]:
X_name = datasets_basecaller['read_groups'][read_group][
'barcode_name']
dst = os.path.join(ALIGNMENT_RESULTS,
'alignment_%s.summary' % X_name)
try:
os.symlink(os.path.relpath(src, os.path.dirname(dst)), dst)
except:
printtime("ERROR: Unable to symlink '%s' to '%s'" % (src, dst))
# Special legacy post-processing.
# Generate merged rawlib.bam on barcoded runs
composite_bam_filename = os.path.join(ALIGNMENT_RESULTS, 'rawlib.bam')
if not os.path.exists(composite_bam_filename):
bam_file_list = []
for dataset in datasets_basecaller["datasets"]:
bam_name = os.path.join(
ALIGNMENT_RESULTS,
os.path.basename(dataset['file_prefix']) + '.bam')
if os.path.exists(bam_name):
bam_file_list.append(bam_name)
blockprocessing.merge_bam_files(bam_file_list, composite_bam_filename,
composite_bam_filename + '.bai',
mark_duplicates)
force_alignstats = True
if force_alignstats:
## Generate data for error plot for barcoded run from composite bam
printtime("Call alignStats to generate raw accuracy")
try:
cmd = "alignStats"
cmd += " -n 12"
cmd += " --alignSummaryFile alignStats_err.txt"
cmd += " --alignSummaryJsonFile alignStats_err.json"
cmd += " --alignSummaryMinLen 1"
#cmd += " --alignSummaryMaxLen %s" % str(int(graph_max_x))
cmd += " --alignSummaryMaxLen %s" % str(int(400))
cmd += " --alignSummaryLenStep 1"
cmd += " --alignSummaryMaxErr 10"
cmd += " --infile %s" % composite_bam_filename
cmd = cmd + " --outputDir %s" % ALIGNMENT_RESULTS
printtime("DEBUG: Calling '%s'" % cmd)
os.system(cmd)
except:
printtime("alignStats failed")
mergeAlignStatsResults(input_prefix_list, ALIGNMENT_RESULTS + "/")
try:
base_error_plot.generate_base_error_plot(
os.path.join(ALIGNMENT_RESULTS, 'alignStats_err.json'),
os.path.join(ALIGNMENT_RESULTS, 'base_error_plot.png'),
int(graph_max_x))
ionstats_plots.alignment_rate_plot(
os.path.join(ALIGNMENT_RESULTS, 'alignStats_err.json'),
os.path.join(BASECALLER_RESULTS, 'ionstats_basecaller.json'),
os.path.join(ALIGNMENT_RESULTS, 'alignment_rate_plot.png'),
int(graph_max_x))
# Create aligned histogram plot
# Create AQ20 plot
printtime("Base error plot has been created successfully")
except:
printtime("ERROR: Failed to generate base error plot")
traceback.print_exc()
# Generate alignment_barcode_summary.csv
barcodelist_path = 'barcodeList.txt'
if not os.path.exists(barcodelist_path):
barcodelist_path = '../barcodeList.txt'
if not os.path.exists(barcodelist_path):
barcodelist_path = '../../barcodeList.txt'
if not os.path.exists(barcodelist_path):
barcodelist_path = '../../../barcodeList.txt'
if not os.path.exists(barcodelist_path):
barcodelist_path = '../../../../barcodeList.txt'
if os.path.exists(barcodelist_path):
printtime("Barcode processing, aggregate")
aggregate_alignment("./", barcodelist_path)
# These graphs are likely obsolete
makeAlignGraphs()
if os.path.exists(ionstats_path):
file_list.append(ionstats_path)
elif os.path.exists(ionstats_path_CA):
file_list.append(ionstats_path_CA)
else:
raise Exception('')
ionstats.reduce_stats(file_list, ionstats_file)
# Make alignment_rate_plot.png
stats = json.load(open(ionstats_file))
l = stats['full']['max_read_length']
graph_max_x = int(round(l + 49, -2))
ionstats_plots.alignment_rate_plot(
'alignStats_err.json',
'ionstats_basecaller.json',
'alignment_rate_plot.png', int(graph_max_x))
print("Ionstats plot created successfully")
except:
print("ERROR: Failed to generate alignment rate plot")
try:
# Make base_error_plot.png
base_error_plot.generate_base_error_plot(
'alignStats_err.json',
'base_error_plot.png',int(graph_max_x))
except:
print("ERROR: Failed to generate base error plot")
traceback.print_exc()
if os.path.exists(ionstats_path):
file_list.append(ionstats_path)
elif os.path.exists(ionstats_path_CA):
file_list.append(ionstats_path_CA)
else:
raise Exception('')
ionstats.reduce_stats(file_list, ionstats_file)
# Make alignment_rate_plot.png
stats = json.load(open(ionstats_file))
l = stats['full']['max_read_length']
graph_max_x = int(round(l + 49, -2))
ionstats_plots.alignment_rate_plot('alignStats_err.json',
'ionstats_basecaller.json',
'alignment_rate_plot.png',
int(graph_max_x))
print("Ionstats plot created successfully")
except:
print("ERROR: Failed to generate alignment rate plot")
try:
# Make base_error_plot.png
base_error_plot.generate_base_error_plot('alignStats_err.json',
'base_error_plot.png',
int(graph_max_x))
except:
print("ERROR: Failed to generate base error plot")
traceback.print_exc()
if args.zip and len(args.files) > 1:
def merge_alignment_stats(dirs, BASECALLER_RESULTS, ALIGNMENT_RESULTS, flows):
datasets_json = {}
try:
f = open(os.path.join(BASECALLER_RESULTS,"datasets_basecaller.json"),'r')
datasets_json = json.load(f);
f.close()
except:
printtime("ERROR: problem parsing %s" % os.path.join(BASECALLER_RESULTS,"datasets_basecaller.json"))
traceback.print_exc()
return
for dataset in datasets_json['datasets']:
# What needs merging:
# - alignment.summary
# - alignTable.txt
# Some time in the future:
# - alignStats_err.json
# Merge alignStats metrics
try:
input_prefix_list = [os.path.join(dir,ALIGNMENT_RESULTS, dataset['file_prefix']+'.') for dir in dirs]
input_prefix_list = [prefix for prefix in input_prefix_list if os.path.exists(prefix+'alignment.summary')]
composite_prefix = os.path.join(ALIGNMENT_RESULTS, dataset['file_prefix']+'.')
if input_prefix_list:
mergeAlignStatsResults(input_prefix_list,composite_prefix)
else:
printtime("Nothing to merge: "+dataset['file_prefix'])
except:
printtime("ERROR: merging %s stats unsuccessful" % (dataset['file_prefix']+'.bam'))
datasets_basecaller = {}
try:
f = open(os.path.join(BASECALLER_RESULTS,"datasets_basecaller.json"),'r')
datasets_basecaller = json.load(f);
f.close()
except:
printtime("ERROR: problem parsing %s" % os.path.join(BASECALLER_RESULTS,"datasets_basecaller.json"))
traceback.print_exc()
return
try:
graph_max_x = int(50 * math.ceil(0.014 * int(flows)))
except:
graph_max_x = 400
input_prefix_list = []
for dataset in datasets_basecaller["datasets"]:
printtime("Barcode processing, rename")
src = os.path.join(ALIGNMENT_RESULTS,dataset['file_prefix']+'.alignment.summary')
if os.path.exists(src):
input_prefix_list.append(os.path.join(ALIGNMENT_RESULTS,dataset['file_prefix']+'.'))
#terrible hack to make aggregate_alignment happy
X_name = 'nomatch'
read_group = dataset['read_groups'][0]
if 'barcode_name' in datasets_basecaller['read_groups'][read_group]:
X_name = datasets_basecaller['read_groups'][read_group]['barcode_name']
dst = os.path.join(ALIGNMENT_RESULTS, 'alignment_%s.summary' % X_name)
try:
os.symlink(os.path.relpath(src,os.path.dirname(dst)),dst)
except:
printtime("ERROR: Unable to symlink '%s' to '%s'" % (src, dst))
# Merge alignStats_err.json right here!
merged_align_stats = {}
align_stats_num_bases = 400
for dir in dirs:
current_align_stats = {}
try:
f = open(os.path.join(dir,ALIGNMENT_RESULTS,'alignStats_err.json'),'r')
current_align_stats = json.load(f);
f.close()
except:
printtime("Merge alignStats_err.json: skipping %s" % os.path.join(dir,ALIGNMENT_RESULTS,'alignStats_err.json'))
continue
if not merged_align_stats:
merged_align_stats = current_align_stats
align_stats_num_bases = len(merged_align_stats.get("read_length",[]))
continue
for idx in range(align_stats_num_bases):
merged_align_stats['nread'][idx] += current_align_stats['nread'][idx]
merged_align_stats['unaligned'][idx] += current_align_stats['unaligned'][idx]
merged_align_stats['filtered'][idx] += current_align_stats['filtered'][idx]
merged_align_stats['clipped'][idx] += current_align_stats['clipped'][idx]
merged_align_stats['aligned'][idx] += current_align_stats['aligned'][idx]
merged_align_stats['n_err_at_position'][idx] += current_align_stats['n_err_at_position'][idx]
merged_align_stats['cum_aligned'][idx] += current_align_stats['cum_aligned'][idx]
merged_align_stats['cum_err_at_position'][idx] += current_align_stats['cum_err_at_position'][idx]
merged_align_stats['accuracy_total_bases'] += current_align_stats['accuracy_total_bases']
merged_align_stats['accuracy_total_errors'] += current_align_stats['accuracy_total_errors']
merged_align_stats['total_mapped_target_bases'] += current_align_stats['total_mapped_target_bases']
merged_align_stats['total_mapped_reads'] += current_align_stats['total_mapped_reads']
try:
f = open(os.path.join(ALIGNMENT_RESULTS,'alignStats_err.json'),"w")
json.dump(merged_align_stats, f, indent=4)
f.close()
except:
printtime("ERROR; Failed to write merged alignStats_err.json")
traceback.print_exc()
mergeAlignStatsResults(input_prefix_list,ALIGNMENT_RESULTS+"/")
try:
base_error_plot.generate_base_error_plot(
os.path.join(ALIGNMENT_RESULTS,'alignStats_err.json'),
os.path.join(ALIGNMENT_RESULTS,'base_error_plot.png'),int(graph_max_x))
ionstats_plots.alignment_rate_plot(
os.path.join(ALIGNMENT_RESULTS,'alignStats_err.json'),
os.path.join(BASECALLER_RESULTS,'ionstats_basecaller.json'),
os.path.join(ALIGNMENT_RESULTS,'alignment_rate_plot.png'),int(graph_max_x))
printtime("Base error plot has been created successfully")
except:
printtime("ERROR: Failed to generate base error plot")
traceback.print_exc()
# Generate alignment_barcode_summary.csv
barcodelist_path = 'barcodeList.txt'
if not os.path.exists(barcodelist_path):
barcodelist_path = '../barcodeList.txt'
if not os.path.exists(barcodelist_path):
barcodelist_path = '../../barcodeList.txt'
if not os.path.exists(barcodelist_path):
barcodelist_path = '../../../barcodeList.txt'
if not os.path.exists(barcodelist_path):
barcodelist_path = '../../../../barcodeList.txt'
if os.path.exists(barcodelist_path):
printtime("Barcode processing, aggregate")
aggregate_alignment ("./",barcodelist_path)
def alignment_post_processing(
BASECALLER_RESULTS,
ALIGNMENT_RESULTS,
flows,
mark_duplicates,
force_alignstats):
datasets_basecaller = {}
try:
f = open(os.path.join(BASECALLER_RESULTS,"datasets_basecaller.json"),'r')
datasets_basecaller = json.load(f);
f.close()
except:
printtime("ERROR: problem parsing %s" % os.path.join(BASECALLER_RESULTS,"datasets_basecaller.json"))
traceback.print_exc()
return
try:
graph_max_x = int(50 * math.ceil(0.014 * int(flows)))
except:
graph_max_x = 400
input_prefix_list = []
for dataset in datasets_basecaller["datasets"]:
if not os.path.exists(os.path.join(BASECALLER_RESULTS, dataset['basecaller_bam'])):
continue
printtime("Barcode processing, rename")
src = os.path.join(ALIGNMENT_RESULTS,dataset['file_prefix']+'.alignment.summary')
if os.path.exists(src):
input_prefix_list.append(os.path.join(ALIGNMENT_RESULTS,dataset['file_prefix']+'.'))
#terrible hack to make aggregate_alignment happy
X_name = 'nomatch'
read_group = dataset['read_groups'][0]
if 'barcode_name' in datasets_basecaller['read_groups'][read_group]:
X_name = datasets_basecaller['read_groups'][read_group]['barcode_name']
dst = os.path.join(ALIGNMENT_RESULTS, 'alignment_%s.summary' % X_name)
try:
os.symlink(os.path.relpath(src,os.path.dirname(dst)),dst)
except:
printtime("ERROR: Unable to symlink '%s' to '%s'" % (src, dst))
# Special legacy post-processing.
# Generate merged rawlib.bam on barcoded runs
composite_bam_filename = os.path.join(ALIGNMENT_RESULTS,'rawlib.bam')
if not os.path.exists(composite_bam_filename):
bam_file_list = []
for dataset in datasets_basecaller["datasets"]:
bam_name = os.path.join(ALIGNMENT_RESULTS,os.path.basename(dataset['file_prefix'])+'.bam')
if os.path.exists(bam_name):
bam_file_list.append(bam_name)
blockprocessing.merge_bam_files(bam_file_list,composite_bam_filename,composite_bam_filename+'.bai',mark_duplicates)
force_alignstats = True
if force_alignstats:
## Generate data for error plot for barcoded run from composite bam
printtime("Call alignStats to generate raw accuracy")
try:
cmd = "alignStats"
cmd += " -n 12"
cmd += " --alignSummaryFile alignStats_err.txt"
cmd += " --alignSummaryJsonFile alignStats_err.json"
cmd += " --alignSummaryMinLen 1"
#cmd += " --alignSummaryMaxLen %s" % str(int(graph_max_x))
cmd += " --alignSummaryMaxLen %s" % str(int(400))
cmd += " --alignSummaryLenStep 1"
cmd += " --alignSummaryMaxErr 10"
cmd += " --infile %s" % composite_bam_filename
cmd = cmd + " --outputDir %s" % ALIGNMENT_RESULTS
printtime("DEBUG: Calling '%s'" % cmd)
os.system(cmd)
except:
printtime("alignStats failed")
mergeAlignStatsResults(input_prefix_list,ALIGNMENT_RESULTS+"/")
try:
base_error_plot.generate_base_error_plot(
os.path.join(ALIGNMENT_RESULTS,'alignStats_err.json'),
os.path.join(ALIGNMENT_RESULTS,'base_error_plot.png'),int(graph_max_x))
ionstats_plots.alignment_rate_plot(
os.path.join(ALIGNMENT_RESULTS,'alignStats_err.json'),
os.path.join(BASECALLER_RESULTS,'ionstats_basecaller.json'),
os.path.join(ALIGNMENT_RESULTS,'alignment_rate_plot.png'),int(graph_max_x))
# Create aligned histogram plot
# Create AQ20 plot
printtime("Base error plot has been created successfully")
except:
printtime("ERROR: Failed to generate base error plot")
traceback.print_exc()
# Generate alignment_barcode_summary.csv
barcodelist_path = 'barcodeList.txt'
if not os.path.exists(barcodelist_path):
barcodelist_path = '../barcodeList.txt'
if not os.path.exists(barcodelist_path):
barcodelist_path = '../../barcodeList.txt'
if not os.path.exists(barcodelist_path):
barcodelist_path = '../../../barcodeList.txt'
if not os.path.exists(barcodelist_path):
barcodelist_path = '../../../../barcodeList.txt'
if os.path.exists(barcodelist_path):
printtime("Barcode processing, aggregate")
aggregate_alignment ("./",barcodelist_path)
# These graphs are likely obsolete
makeAlignGraphs()