def make_index_and_hash(data_folder, adaID, fragment, VERBOSE=0, summary=True):
'''Make index and hash files for consensus'''
frag_gen = fragment[:2]
# NOTE: we can use --overwrite here, because there is no concurrency (every
# job has its own hash)
# 1. Make genome index file
sp.call([stampy_bin,
'--species="HIV fragment '+frag_gen+'"',
'--overwrite',
'-G', get_index_file(data_folder, adaID, frag_gen, ext=False),
get_consensus_filename(data_folder, adaID, frag_gen, trim_primers=True),
])
if VERBOSE:
print 'Built index: '+adaID+' '+frag_gen
# 2. Build a hash file
sp.call([stampy_bin,
'--overwrite',
'-g', get_index_file(data_folder, adaID, frag_gen, ext=False),
'-H', get_hash_file(data_folder, adaID, frag_gen, ext=False),
])
if VERBOSE:
print 'Built hash: '+adaID+' '+frag_gen
if summary:
with open(get_map_summary_filename(data_folder, adaID, frag_gen), 'a') as f:
f.write('\n')
f.write('Stampy index and hash written.')
f.write('\n')
def make_index_and_hash(data_folder, adaID, fragment, VERBOSE=0, summary=True):
'''Make index and hash files for consensus'''
frag_gen = fragment[:2]
# NOTE: we can use --overwrite here, because there is no concurrency (every
# job has its own hash)
# 1. Make genome index file
sp.call([
stampy_bin,
'--species="HIV fragment ' + frag_gen + '"',
'--overwrite',
'-G',
get_index_file(data_folder, adaID, frag_gen, ext=False),
get_consensus_filename(data_folder, adaID, frag_gen,
trim_primers=True),
])
if VERBOSE:
print 'Built index: ' + adaID + ' ' + frag_gen
# 2. Build a hash file
sp.call([
stampy_bin,
'--overwrite',
'-g',
get_index_file(data_folder, adaID, frag_gen, ext=False),
'-H',
get_hash_file(data_folder, adaID, frag_gen, ext=False),
])
if VERBOSE:
print 'Built hash: ' + adaID + ' ' + frag_gen
if summary:
with open(get_map_summary_filename(data_folder, adaID, frag_gen),
'a') as f:
f.write('\n')
f.write('Stampy index and hash written.')
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')
for fragment in fragments_sample:
frag_gen = fragment[:2]
# Submit to the cluster self if requested
if submit:
fork_self(seq_run, adaID, frag_gen,
VERBOSE=VERBOSE,
threads=threads, maxreads=maxreads,
filter_reads=filter_reads,
summary=summary,
rescue=use_rescue)
continue
if summary:
sfn = get_map_summary_filename(data_folder, adaID, frag_gen,
rescue=use_rescue)
with open(sfn, 'w') as f:
f.write('Call: python map_to_consensus.py'+\
' --run '+seq_run+\
' --adaIDs '+adaID+\
' --fragments '+frag_gen+\
' --threads '+str(threads)+\
' --verbose '+str(VERBOSE))
if maxreads != -1:
f.write(' --maxreads '+str(maxreads))
if filter_reads:
f.write(' --filter')
f.write('\n')
if not check_consensus_length(data_folder, adaID, fragment, VERBOSE=VERBOSE):
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')
# Submit to the cluster self if requested
if submit:
fork_self(seq_run,
adaID,
frag_gen,
VERBOSE=VERBOSE,
threads=threads,
maxreads=maxreads,
filter_reads=filter_reads,
summary=summary,
rescue=use_rescue)
continue
if summary:
sfn = get_map_summary_filename(data_folder,
adaID,
frag_gen,
rescue=use_rescue)
with open(sfn, 'w') as f:
f.write('Call: python map_to_consensus.py'+\
' --run '+seq_run+\
' --adaIDs '+adaID+\
' --fragments '+frag_gen+\
' --threads '+str(threads)+\
' --verbose '+str(VERBOSE))
if maxreads != -1:
f.write(' --maxreads ' + str(maxreads))
if filter_reads:
f.write(' --filter')
f.write('\n')
if not check_consensus_length(