def index_reference(fasta_file, logger):
gf_utils.info_header(logger, "index reference file using bowtie2")
process = subprocess.Popen(['bowtie2-build', '-f', fasta_file, fasta_file],
stderr=subprocess.PIPE,
stdout=subprocess.PIPE)
process.wait()
gf_utils.log_process(
logger,
process,
log_error_to="bowtie2: reference indexing successfully completed")
gf_utils.info_header(logger,
'bowtie2: reference indexing successfully completed')
def run_samtools_bam(path_to_tmp_file, fasta_file, prefix, logger):
sam_output = path_to_tmp_file + "/" + prefix + ".sam"
bam_output = path_to_tmp_file + "/" + prefix + ".bam"
bam_sorted_output = path_to_tmp_file + "/" + prefix + ".sorted"
bam_sorted_index_output = bam_sorted_output + ".bam"
pileup_output = path_to_tmp_file + "/" + prefix + '.pileup'
gf_utils.info_header(logger, 'converting .sam to .bam')
process = subprocess.Popen([
'samtools', 'view', '-b', '-o', bam_output, '-q', '1', '-S',
sam_output + '.mod'
],
stderr=subprocess.PIPE,
stdout=subprocess.PIPE)
process.wait()
gf_utils.log_process(logger, process, log_error_to="info")
gf_utils.info_header(logger, 'samtools sorting .bam file')
process = subprocess.Popen(
['samtools', 'sort', bam_output, bam_sorted_output],
stderr=subprocess.PIPE,
stdout=subprocess.PIPE)
process.wait()
gf_utils.log_process(logger, process, log_error_to="info")
gf_utils.info_header(logger, 'samtools indexing .bam file')
process = subprocess.Popen(['samtools', 'index', bam_sorted_index_output],
stderr=subprocess.PIPE,
stdout=subprocess.PIPE)
process.wait()
gf_utils.log_process(logger, process, log_error_to="info")
gf_utils.info_header(logger, 'generating mpileup file')
gf_utils.info_header(
logger, ' '.join(
map(str, [
'samtools', 'mpileup', '-A', '-B', '-f', fasta_file,
bam_sorted_index_output
])))
with open(pileup_output, 'w') as sam_pileup:
process = subprocess.Popen([
'samtools', 'mpileup', '-A', '-B', '-f', fasta_file,
bam_sorted_index_output
],
stderr=subprocess.PIPE,
stdout=subprocess.PIPE)
for x in process.stdout:
sam_pileup.write(x)
process.wait()
gf_utils.log_process(logger,
process,
log_error_to="couldn't generate mpileup file")
def samtools_faidx(fasta_file, logger):
process = subprocess.Popen(['samtools', 'faidx', fasta_file],
stderr=subprocess.PIPE,
stdout=subprocess.PIPE)
process.wait()
gf_utils.log_process(logger, process, log_error_to="info")
fai_index = fasta_file + '.fai'
if os.path.getsize(fai_index) > 0:
gf_utils.info_header(logger, 'samtootls faidx sucsessfully created')
else:
gf_utils.info_header(
logger,
'samtootls faidx output is empty, check reference fasta format')
sys.exit(1)
def modify_bowtie_sam(path_sam_folder, prefix, logger):
path_sam_file = path_sam_folder + "/" + prefix + ".sam" # prefix = sample_id (eg. molis id)
with open(path_sam_file) as sam, open(path_sam_file + '.mod',
'w') as sam_mod:
for line in sam:
if not line.startswith('@'):
fields = line.split('\t')
flag = int(fields[1])
flag = (flag - 256) if (flag > 256) else flag
sam_mod.write('\t'.join([fields[0], str(flag)] + fields[2:]))
else:
sam_mod.write(line)
gf_utils.info_header(
logger, 'SAM file successfully mofified to unset secondary alignment')
def run_bowtie_on_indices(fasta_file, forward_fastq, reverse_fastq, outdir,
workflow_name, version, prefix, bowtie_options,
logger):
sam_output = outdir + "/tmp/" + prefix + '.sam'
gf_utils.info_header(logger, 'running bowtie command')
bowtie_options_combined = ' '.join(
bowtie_options
) # bowtie_option is a list need to be joined by space before assigned to the command line
process = subprocess.Popen([
'bowtie2', '-1', forward_fastq, '-2', reverse_fastq, '-x', fasta_file,
'-S', sam_output, bowtie_options_combined
],
stderr=subprocess.PIPE,
stdout=subprocess.PIPE)
process.wait()
gf_utils.info_header(logger, 'SAM file successfully created')