def main():
parser = ArgumentParser(description='Add readgroups based on Illumina ' +
'IDs to given BAM')
parser.add_argument('-c', '--cn', required=True,
help='Sequencing Center Name')
parser.add_argument('-l', '--lb', required=True, help='Library')
parser.add_argument('-s', '--sm', required=True, help='Sample')
parser.add_argument('bam', help='Input BAM')
args = parser.parse_args()
ReadGroup = namedtuple('ReadGroup', 'ID CN LB PL SM',
defaults=[args.cn, args.lb, 'ILLUMINA', args.sm])
in_bam = args.bam
in_bam_stem = Path(in_bam).stem
# Intermediate BAM with readgroup tagged reads, but no readgroups in header
temp_bam = in_bam_stem + '.temp.bam'
# Header file with readgroups
header_sam = in_bam_stem + '.header.sam'
# Final BAM
final_bam = in_bam_stem + '.readgroupsadded.bam'
readgroups = dict()
header_dict = dict()
# Add readgroup tag to each read and create readgroups for header
with pysam.AlignmentFile(in_bam, 'rb') as in_bam_fh, \
pysam.AlignmentFile(temp_bam, 'wb', template=in_bam_fh) as temp_bam_fh:
header_dict = in_bam_fh.header.to_dict()
for read in in_bam_fh.fetch(until_eof=True):
readgroup_id = '.'.join(read.query_name.split(':')[2:4])
read.set_tag('RG', readgroup_id, 'Z')
temp_bam_fh.write(read)
if readgroup_id not in readgroups:
readgroups[readgroup_id] = ReadGroup(readgroup_id)
# Add readgroups to header
header_dict['RG'] = [readgroup._asdict() for readgroup in readgroups.values()]
# Write updated header to file
header_sam_fh = pysam.AlignmentFile(header_sam, 'wh', header=header_dict)
header_sam_fh.close()
# Create new bam with readgroups in header and read tags
Path(final_bam).touch()
pysam.reheader(header_sam, temp_bam, save_stdout=final_bam)
# Delete intermediate BAM and header file
Path(temp_bam).unlink()
Path(header_sam).unlink()
def reheader_and_rename_chromosomes(in_bam_file, out_bam_file, replacements):
with pysam.Samfile(in_bam_file) as f:
h = str(f.header)
org_header = h
for a, b in replacements.items():
h = h.replace(f"SN:{a}", f"SN:{b}")
if h == org_header:
raise ValueError("No replacement happened")
tf = tempfile.NamedTemporaryFile()
tf.write(h.encode("utf-8"))
tf.flush()
out_bam_file.write_text("") # must be there for save_stdout to work..
pysam.reheader(
tf.name,
str(in_bam_file.absolute()),
save_stdout=str(out_bam_file.absolute()).encode("utf-8"),
)
pysam.index(str(out_bam_file))
def main(args):
# Detect the input format.
try:
with open(args.bam, 'r') as f:
for line in f:
is_sam = True
break
except UnicodeDecodeError:
is_sam = False
# Rename the sample(s).
old_lines = pysam.view('-H', args.bam, '--no-PG').strip().split('\n')
new_lines = []
for old_line in old_lines:
old_fields = old_line.split('\t')
if old_fields[0] != '@RG':
new_lines.append(old_line)
continue
new_fields = []
for old_field in old_fields:
if 'SM:' not in old_field:
new_fields.append(old_field)
continue
new_fields.append(f'SM:{args.name}')
new_lines.append('\t'.join(new_fields))
# Write the output file.
if is_sam:
for new_line in new_lines:
sys.stdout.write(new_line + '\n')
alignments = pysam.view(args.bam, '--no-PG')
sys.stdout.write(alignments)
else:
with tempfile.TemporaryDirectory() as t:
with open(f'{t}/header.sam', 'w') as f:
f.write('\n'.join(new_lines))
alignments = pysam.reheader(f'{t}/header.sam',
args.bam,
'--no-PG')
sys.stdout.buffer.write(alignments)
def premap_stampy(data_folder,
adaID,
VERBOSE=0,
threads=1,
summary=True,
maxreads=-1,
subsrate=0.05,
gapopen=40,
gapextend=3):
'''Call stampy for actual mapping'''
if VERBOSE:
print 'Premapping: adaID ', adaID
if summary:
summary_filename = get_premap_summary_filename(data_folder, adaID)
# Stampy can handle both gzipped and uncompressed fastq inputs
input_filenames = get_read_filenames(data_folder, adaID, gzip=True)
if not os.path.isfile(input_filenames[0]):
input_filenames = get_read_filenames(data_folder, adaID, gzip=False)
if not all(map(os.path.isfile, input_filenames)):
raise OSError('Input files for mapping not found: ' +
input_filenames[0])
# parallelize if requested
if threads == 1:
call_list = [
stampy_bin,
'--overwrite',
'-g',
get_reference_premap_index_filename(data_folder, adaID, ext=False),
'-h',
get_reference_premap_hash_filename(data_folder, adaID, ext=False),
'-o',
get_premapped_filename(data_folder, adaID, type='sam'),
'--insertsize=450',
'--insertsd=100',
'--substitutionrate=' + str(subsrate),
'--gapopen=' + str(gapopen),
'--gapextend=' + str(gapextend),
]
if maxreads > 0:
call_list.append('--numrecords=' + str(maxreads))
call_list.extend(['-M'] + input_filenames)
call_list = map(str, call_list)
if VERBOSE >= 2:
print ' '.join(call_list)
sp.call(call_list)
if summary:
with open(get_premap_summary_filename(data_folder, adaID),
'a') as f:
f.write('\nStampy premapped (single thread).\n')
# Convert to compressed BAM
convert_sam_to_bam(
get_premapped_filename(data_folder, adaID, type='bam'))
if summary:
with open(summary_filename, 'a') as f:
f.write('\nSAM file converted to compressed BAM: '+\
get_premapped_filename(data_folder, adaID, type='bam')+'\n')
else:
# Multithreading works as follows: call qsub + stampy, monitor the process
# IDs with qstat at regular intervals, and finally merge results with pysam
output_file_parts = [
get_premapped_filename(data_folder,
adaID,
type='bam',
part=(j + 1)) for j in xrange(threads)
]
# Submit map script
jobs_done = np.zeros(threads, bool)
job_IDs = np.zeros(threads, 'S30')
# Submit map call
import hivwholeseq
JOBDIR = hivwholeseq.__path__[0].rstrip('/') + '/'
JOBLOGOUT = JOBDIR + 'logout'
JOBLOGERR = JOBDIR + 'logerr'
cluster_time = ['23:59:59', '1:59:59']
vmem = '8G'
for j in xrange(threads):
call_list = [
'qsub', '-cwd', '-b', 'y', '-S', '/bin/bash', '-o', JOBLOGOUT,
'-e', JOBLOGERR, '-N', adaID + ' p' + str(j + 1), '-l',
'h_rt=' + cluster_time[threads >= 30], '-l', 'h_vmem=' + vmem,
stampy_bin, '--overwrite', '-g',
get_reference_premap_index_filename(
data_folder, adaID, ext=False), '-h',
get_reference_premap_hash_filename(
data_folder, adaID, ext=False), '-o',
get_premapped_filename(
data_folder, adaID, type='sam', part=(j + 1)),
'--processpart=' + str(j + 1) + '/' + str(threads),
'--insertsize=450', '--insertsd=100', '--substitutionrate=' +
str(subsrate), '--gapopen=' + str(gapopen),
'--gapextend=' + str(gapextend), '-M'
] + input_filenames
call_list = map(str, call_list)
if VERBOSE >= 2:
print ' '.join(call_list)
job_ID = sp.check_output(call_list)
job_ID = job_ID.split()[2]
job_IDs[j] = job_ID
# Monitor output
time_wait = 10 # secs
while not jobs_done.all():
# Sleep some time
time.sleep(time_wait)
# Get the output of qstat to check the status of jobs
qstat_output = sp.check_output(['qstat'])
qstat_output = qstat_output.split(
'\n')[:-1] # The last is an empty line
if VERBOSE >= 3:
print qstat_output
if len(qstat_output) < 3:
jobs_done[:] = True
break
else:
qstat_output = [line.split()[0] for line in qstat_output[2:]]
time_wait = 10 # secs
for j in xrange(threads):
if jobs_done[j]:
continue
if job_IDs[j] not in qstat_output:
# Convert to BAM for merging
if VERBOSE >= 1:
print 'Convert premapped reads to BAM for merging: adaID '+\
adaID+', part '+str(j+1)+ ' of '+ \
str(threads)
convert_sam_to_bam(output_file_parts[j])
# We do not need to wait if we did the conversion (it takes
# longer than some secs)
time_wait = 0
jobs_done[j] = True
if summary:
with open(summary_filename, 'a') as f:
f.write('Stampy premapped (' + str(threads) + ' threads).\n')
# Concatenate output files
if VERBOSE >= 1:
print 'Concatenate premapped reads: adaID ' + adaID + '...',
output_filename = get_premapped_filename(data_folder,
adaID,
type='bam',
unsorted=True)
pysam.cat('-o', output_filename, *output_file_parts)
if VERBOSE >= 1:
print 'done.'
if summary:
with open(summary_filename, 'a') as f:
f.write('BAM files concatenated (unsorted).\n')
# Sort the file by read names (to ensure the pair_generator)
# NOTE: we exclude the extension and the option -f because of a bug in samtools
if VERBOSE >= 1:
print 'Sort premapped reads: adaID ' + adaID
output_filename_sorted = get_premapped_filename(data_folder,
adaID,
type='bam',
unsorted=False)
pysam.sort('-n', output_filename, output_filename_sorted[:-4])
if summary:
with open(summary_filename, 'a') as f:
f.write('Joint BAM file sorted.\n')
# Reheader the file without BAM -> SAM -> BAM
if VERBOSE >= 1:
print 'Reheader premapped reads: adaID ' + adaID
header_filename = get_premapped_filename(data_folder,
adaID,
type='sam',
part=1)
pysam.reheader(header_filename, output_filename_sorted)
if summary:
with open(summary_filename, 'a') as f:
f.write('Joint BAM file reheaded.\n')
if VERBOSE >= 1:
print 'Remove temporary files: adaID ' + adaID
remove_premapped_tempfiles(data_folder, adaID, VERBOSE=VERBOSE)
if summary:
with open(summary_filename, 'a') as f:
f.write('Temp premapping files removed.\n')
f.write('\n')
def map_stampy(data_folder,
adaID,
fragment,
VERBOSE=0,
threads=1,
cluster_time='23:59:59',
maxreads=-1,
summary=True,
rescue=False,
dry=False):
'''Map using stampy'''
frag_gen = fragment[:2]
if summary:
summary_filename = get_map_summary_filename(data_folder,
adaID,
frag_gen,
rescue=rescue)
# Set mapping penalty scores: softer for rescues and F3 and F5
global subsrate
if rescue:
subsrate = '0.2'
stampy_gapopen = 5 # Default: 40
stampy_gapextend = 1 # Default: 3
elif frag_gen not in ('F3', 'F5'):
stampy_gapopen = 60 # Default: 40
stampy_gapextend = 5 # Default: 3
else:
stampy_gapopen = 30 # Default: 40
stampy_gapextend = 2 # Default: 3
if VERBOSE:
print 'Map via stampy: ' + adaID + ' ' + frag_gen
if not rescue:
input_filename = get_divided_filename(data_folder,
adaID,
fragment,
type='bam')
# NOTE: we introduced fragment nomenclature late, e.g. F3a. Check for that
if not os.path.isfile(input_filename):
if frag_gen == 'F3':
input_filename = input_filename.replace('F3a', 'F3')
else:
input_filename = get_divided_filename(data_folder,
adaID,
'unmapped',
type='bam')
# Check existance of input file, because stampy creates output anyway
if not os.path.isfile(input_filename):
if summary:
with open(summary_filename, 'a') as f:
f.write('Failed (input file for mapping not found).\n')
raise ValueError(samplename + ', fragment ' + fragment +
': input file not found.')
# parallelize if requested
if threads == 1:
output_filename = get_mapped_filename(data_folder,
adaID,
frag_gen,
type='sam',
rescue=rescue)
# Map
call_list = [
stampy_bin, '-g',
get_index_file(data_folder, adaID, frag_gen, ext=False), '-h',
get_hash_file(data_folder, adaID, frag_gen, ext=False), '-o',
output_filename, '--overwrite', '--substitutionrate=' + subsrate,
'--gapopen', stampy_gapopen, '--gapextend', stampy_gapextend
]
if stampy_sensitive:
call_list.append('--sensitive')
# Take only a (random) subsample: stampy uses the fraction of reads
# intead of the number
if maxreads > 0:
# FIXME: figure out the -s option and the --numrecords option
call_list.extend(['--numrecords', maxreads])
#n_pairs_tot = get_number_reads(input_filename, 'bam') / 2
#frac_pairs = 1.0 * maxreads / n_pairs_tot
#random_seed = np.random.randint(1e5)
#call_list.extend(['-s', frac_pairs + random_seed])
call_list = call_list + ['-M', input_filename]
call_list = map(str, call_list)
if VERBOSE >= 2:
print ' '.join(call_list)
if not dry:
sp.call(call_list)
if summary:
with open(summary_filename, 'a') as f:
f.write('Stampy mapped (single thread).\n')
output_filename = get_mapped_filename(data_folder,
adaID,
frag_gen,
type='bam',
rescue=rescue)
convert_sam_to_bam(output_filename)
else:
if summary:
with open(summary_filename, 'a') as f:
f.write('Dry run works (single thread).\n')
if VERBOSE >= 1:
print 'Dry run works (single thread)'
return
else:
# Submit map script
jobs_done = np.zeros(threads, bool)
job_IDs = np.zeros(threads, 'S30')
for j in xrange(threads):
# Get output filename
output_filename = get_mapped_filename(data_folder,
adaID,
frag_gen,
type='sam',
part=(j + 1),
rescue=rescue)
# Map
call_list = [
'qsub', '-cwd', '-b', 'y', '-S', '/bin/bash', '-o', JOBLOGOUT,
'-e', JOBLOGERR, '-N',
'm' + adaID.replace('-', '') + frag_gen + str(j + 1), '-l',
'h_rt=' + cluster_time, '-l', 'h_vmem=' + vmem, stampy_bin,
'-g',
get_index_file(data_folder, adaID, frag_gen, ext=False), '-h',
get_hash_file(data_folder, adaID, frag_gen,
ext=False), '-o', output_filename, '--overwrite',
'--processpart=' + str(j + 1) + '/' + str(threads),
'--substitutionrate=' + subsrate, '--gapopen', stampy_gapopen,
'--gapextend', stampy_gapextend
]
if stampy_sensitive:
call_list.append('--sensitive')
call_list = call_list + ['-M', input_filename]
call_list = map(str, call_list)
if VERBOSE >= 2:
print ' '.join(call_list)
if not dry:
job_ID = sp.check_output(call_list)
job_ID = job_ID.split()[2]
job_IDs[j] = job_ID
if dry:
if summary:
with open(summary_filename, 'a') as f:
f.write('Dry run works (multi thread).\n')
if VERBOSE >= 1:
print 'Dry run works (multi thread)'
return
# Monitor output
output_file_parts = [
get_mapped_filename(data_folder,
adaID,
frag_gen,
type='bam',
part=(j + 1),
rescue=rescue) for j in xrange(threads)
]
time_wait = 10 # secs
while not jobs_done.all():
# Sleep some time
time.sleep(time_wait)
# Get the output of qstat to check the status of jobs
qstat_output = sp.check_output(['qstat'])
qstat_output = qstat_output.split(
'\n')[:-1] # The last is an empty line
if len(qstat_output) < 3:
jobs_done[:] = True
break
else:
qstat_output = [line.split()[0] for line in qstat_output[2:]]
time_wait = 10 # secs
for j in xrange(threads):
if jobs_done[j]:
continue
if job_IDs[j] not in qstat_output:
# Convert to BAM for merging
if VERBOSE >= 1:
print 'Convert mapped reads to BAM for merging: adaID '+\
adaID+', fragment '+frag_gen+', part '+str(j+1)+ ' of '+ \
str(threads)
convert_sam_to_bam(output_file_parts[j])
# We do not need to wait if we did the conversion (it takes
# longer than some secs)
time_wait = 0
jobs_done[j] = True
if summary:
with open(summary_filename, 'a') as f:
f.write('Stampy mapped (' + str(threads) + ' threads).\n')
# Concatenate output files
output_filename = get_mapped_filename(data_folder,
adaID,
frag_gen,
type='bam',
unsorted=True,
rescue=rescue)
if VERBOSE >= 1:
print 'Concatenate mapped reads: adaID ' + adaID + ', fragment ' + frag_gen
pysam.cat('-o', output_filename, *output_file_parts)
if summary:
with open(summary_filename, 'a') as f:
f.write('BAM files concatenated (unsorted).\n')
# Sort the file by read names (to ensure the pair_generator)
output_filename_sorted = get_mapped_filename(data_folder,
adaID,
frag_gen,
type='bam',
unsorted=False,
rescue=rescue)
# NOTE: we exclude the extension and the option -f because of a bug in samtools
if VERBOSE >= 1:
print 'Sort mapped reads: adaID ' + adaID + ', fragment ' + frag_gen
pysam.sort('-n', output_filename, output_filename_sorted[:-4])
if summary:
with open(summary_filename, 'a') as f:
f.write('Joint BAM file sorted.\n')
# Reheader the file without BAM -> SAM -> BAM
if VERBOSE >= 1:
print 'Reheader mapped reads: adaID ' + adaID + ', fragment ' + frag_gen
header_filename = get_mapped_filename(data_folder,
adaID,
frag_gen,
type='sam',
part=1,
rescue=rescue)
pysam.reheader(header_filename, output_filename_sorted)
if summary:
with open(summary_filename, 'a') as f:
f.write('Joint BAM file reheaded.\n')
# FIXME: check whether temp files are all deleted
if VERBOSE >= 1:
print 'Remove temporary files: adaID ' + adaID + ', fragment ' + frag_gen
remove_mapped_tempfiles(data_folder,
adaID,
frag_gen,
VERBOSE=VERBOSE,
rescue=rescue)
if summary:
with open(summary_filename, 'a') as f:
f.write('Temp mapping files removed.\n')
f.write('\n')
if cell_barcode in filtered_barcodes:
read_groups.add(cell_barcode)
# replace RG tag with the cell barcode
read.set_tag('RG', cell_barcode, 'Z', True)
nkeep += 1
output_bam.write(read)
else:
nfiltered += 1
else:
nmiss += 1
input_bam.close()
output_bam.close()
print('\n')
print('No. of alignments missing CB tag: ' + str(nmiss))
print('No. of alignments contains filtered out CB tag: ' + str(nmiss))
print('No. of Reads with passing-filter CB tag: ' + str(nfiltered))
print('No. of read groups: ' + str(len(read_groups)))
print('\n')
print("Writing header.sam with updated RG tags.")
# write out RG tags to header
rg_header = [{'ID': id} for id in read_groups]
header['RG'] = rg_header
with pysam.AlignmentFile(args.header_file, "wh", header=header) as outf:
a = pysam.AlignedSegment()
outf.write(a)
# re-header the output_bam file in place
pysam.reheader("-P", "-i", args.header_file, args.output_bam)
def map_stampy_multithread(sample, fragment, VERBOSE=0, threads=2, summary=True,
filtered=True):
'''Map using stampy, multithread (via cluster requests, queueing race conditions possible)'''
import hivwholeseq
JOBDIR = hivwholeseq.__path__[0].rstrip('/')+'/'
JOBLOGOUT = JOBDIR+'logout/'
JOBLOGERR = JOBDIR+'logerr/'
cluster_time = ['23:59:59', '0:59:59']
vmem = '8G'
pname = patient.id
sample = patient.sample_table.loc[samplename]
seq_run = sample['run']
data_folder = MiSeq_runs[seq_run]['folder']
adaID = sample['adaID']
if VERBOSE:
print 'Map via stampy: '+pname+' '+samplename+' '+fragment
if summary:
summary_filename = get_map_initial_summary_filename(pname, samplename, fragment)
# Specific fragment (e.g. F5 --> F5bi)
frag_spec = filter(lambda x: fragment in x, sample['fragments'])
if not len(frag_spec):
raise ValueError(str(patient)+', '+samplename+': fragment '+fragment+' not found.')
frag_spec = frag_spec[0]
input_filename = get_input_filename(data_folder, adaID, frag_spec, type='bam')
# Submit map scripts in parallel to the cluster
jobs_done = np.zeros(threads, bool)
job_IDs = np.zeros(threads, 'S30')
for j in xrange(threads):
output_filename = get_mapped_to_initial_filename(pname, samplename,
fragment,
type='sam', part=(j+1))
# Map
call_list = ['qsub','-cwd',
'-b', 'y',
'-S', '/bin/bash',
'-o', JOBLOGOUT,
'-e', JOBLOGERR,
'-N', 'm '+samplename+fragment+' p'+str(j+1),
'-l', 'h_rt='+cluster_time[threads >= 10],
'-l', 'h_vmem='+vmem,
stampy_bin,
'--overwrite',
'-g', get_initial_index_filename(pname, fragment, ext=False),
'-h', get_initial_hash_filename(pname, fragment, ext=False),
'-o', output_filename,
'--processpart='+str(j+1)+'/'+str(threads),
'--substitutionrate='+subsrate,
'--gapopen', stampy_gapopen,
'--gapextend', stampy_gapextend]
if stampy_sensitive:
call_list.append('--sensitive')
call_list = call_list + ['-M', input_filename]
call_list = map(str, call_list)
if VERBOSE >= 2:
print ' '.join(call_list)
job_ID = sp.check_output(call_list)
job_ID = job_ID.split()[2]
job_IDs[j] = job_ID
# Monitor output
output_file_parts = [get_mapped_to_initial_filename(pname, samplename,
fragment,
type='bam', part=(j+1))
for j in xrange(threads)]
time_wait = 10 # secs
while not jobs_done.all():
# Sleep some time
time.sleep(time_wait)
# Get the output of qstat to check the status of jobs
qstat_output = sp.check_output(['qstat'])
qstat_output = qstat_output.split('\n')[:-1] # The last is an empty line
if len(qstat_output) < 3:
jobs_done[:] = True
break
else:
qstat_output = [line.split()[0] for line in qstat_output[2:]]
time_wait = 10 # secs
for j in xrange(threads):
if jobs_done[j]:
continue
if job_IDs[j] not in qstat_output:
# Convert to BAM for merging
if VERBOSE >= 1:
print 'Convert mapped reads to BAM for merging: sample '+\
samplename+', part '+str(j+1)+ ' of '+ \
str(threads)
convert_sam_to_bam(output_file_parts[j])
# We do not need to wait if we did the conversion (it takes
# longer than some secs)
time_wait = 0
jobs_done[j] = True
if summary:
with open(summary_filename, 'a') as f:
f.write('Stampy mapped ('+str(threads)+' threads).\n')
# Concatenate output files
output_filename = get_mapped_to_initial_filename(pname, samplename,
fragment,
type='bam', unsorted=True)
if VERBOSE >= 1:
print 'Concatenate premapped reads: sample '+samplename
pysam.cat('-o', output_filename, *output_file_parts)
if summary:
with open(summary_filename, 'a') as f:
f.write('BAM files concatenated (unsorted).\n')
# Sort the file by read names (to ensure the pair_generator)
output_filename_sorted = get_mapped_to_initial_filename(pname, samplename,
fragment,
type='bam')
# NOTE: we exclude the extension and the option -f because of a bug in samtools
if VERBOSE >= 1:
print 'Sort mapped reads: sample '+samplename
pysam.sort('-n', output_filename, output_filename_sorted[:-4])
if summary:
with open(summary_filename, 'a') as f:
f.write('Joint BAM file sorted.\n')
# Reheader the file without BAM -> SAM -> BAM
if VERBOSE >= 1:
print 'Reheader mapped reads: sample '+samplename
header_filename = get_mapped_to_initial_filename(pname, samplename,
fragment,
type='sam', part=1)
pysam.reheader(header_filename, output_filename_sorted)
if summary:
with open(summary_filename, 'a') as f:
f.write('Joint BAM file reheaded.\n')
if VERBOSE >= 1:
print 'Remove temporary files: sample '+samplename
remove_mapped_init_tempfiles(pname, samplename, fragment, VERBOSE=VERBOSE)
if summary:
with open(summary_filename, 'a') as f:
f.write('Temp mapping files removed.\n')
f.write('\n')
