def bwa_aln(bam_path,reference_fasta_path,logger):
uuid_dir=os.path.dirname(bam_path)
fastq_dir=os.path.join(uuid_dir,'fastq')
realn_dir=os.path.join(uuid_dir,'realn')
os.makedirs(realn_dir,exist_ok=True)
fastqlist=fastq_util.buildfastqlist(fastq_dir)
pefastqdict=fastq_util.buildpefastqdict(fastqlist)
logger.info('pefastqdict=%s' % pefastqdict)
sefastqlist=fastq_util.buildsefastqlist(fastqlist)
logger.info('sefastqlist=%s' % sefastqlist)
o1fastqlist=fastq_util.buildo1fastqlist(fastqlist)
logger.info('o1fastqlist=%s' % o1fastqlist)
o2fastqlist=fastq_util.buildo2fastqlist(fastqlist)
logger.info('o2fastqlist=%s' % o2fastqlist)
bam_path_list=list()
for read1 in sorted(pefastqdict.keys()):
bam_path=bwa_aln_paired(bam_path,fastq_dir,read1,pefastqdict[read1],realn_dir,reference_fasta_path,logger)
bam_path_list.append(bam_path)
for seread in sefastqlist:
bam_path=bwa_aln_single(bam_path,fastq_dir,seread,realn_dir,'s',reference_fasta_path,logger)
bam_path_list.append(bam_path)
for o1read in o1fastqlist:
bam_path=bwa_aln_single(bam_path,fastq_dir,o1read,realn_dir,'o1',reference_fasta_path,logger)
bam_path_list.append(bam_path)
for o2read in o2fastqlist:
bam_path=bwa_aln_single(bam_path,fastq_dir,o2read,realn_dir,'o2',reference_fasta_path,logger)
bam_path_list.append(bam_path)
return bam_path_list
def fastqc_validate(uuid,bam_path,thread_count,engine,logger):
uuid_dir=os.path.dirname(bam_path)
fastq_dir=os.path.join(uuid_dir,'fastq')
fastqlist=fastq_util.buildfastqlist(fastq_dir)
logging.info('fastqlist=%s' % fastqlist)
pefastqdict=fastq_util.buildpefastqdict(fastqlist)
logger.info('pefastqdict=%s' % pefastqdict)
sefastqlist=fastq_util.buildsefastqlist(fastqlist)
logger.info('sefastqlist=%s' % sefastqlist)
o1fastqlist=fastq_util.buildo1fastqlist(fastqlist)
logger.info('o1fastqlist=%s' % o1fastqlist)
o2fastqlist=fastq_util.buildo2fastqlist(fastqlist)
logger.info('o2fastqlist=%s' % o2fastqlist)
for read1 in sorted(pefastqdict.keys()):
#read1
fq_path=os.path.join(fastq_dir,read1)
do_fastqc(uuid,fq_path,thread_count,engine,logger)
fastqc_to_db(uuid,fq_path,engine,logger)
do_guess_encoding(uuid,fq_path,engine,logger)
guess_enc_db(uuid,fq_path,engine,logger)
fastq_length=fastq_util.get_fastq_length(uuid,fastq_dir,read1,engine,logger) # removable
#read2
fq_path=os.path.join(fastq_dir,pefastqdict[read1])
do_fastqc(uuid,fq_path,thread_count,engine,logger)
fastqc_to_db(uuid,fq_path,engine,logger)
do_guess_encoding(uuid,fq_path,engine,logger)
guess_enc_db(uuid,fq_path,engine,logger)
for seread in sefastqlist:
fq_path=os.path.join(fastq_dir,seread)
do_fastqc(uuid,fq_path,thread_count,engine,logger)
fastqc_to_db(uuid,fq_path,engine,logger)
do_guess_encoding(uuid,fq_path,engine,logger)
guess_enc_db(uuid,fq_path,engine,logger)
fastq_length=fastq_util.get_fastq_length(uuid,fastq_dir,seread,engine,logger) # removable
for o1read in o1fastqlist:
fq_path=os.path.join(fastq_dir,o1read)
do_fastqc(uuid,fq_path,thread_count,engine,logger)
fastqc_to_db(uuid,fq_path,engine,logger)
do_guess_encoding(uuid,fq_path,engine,logger)
guess_enc_db(uuid,fq_path,engine,logger)
fastq_length=fastq_util.get_fastq_length(uuid,fastq_dir,seread,engine,logger) # removable
for o2read in o2fastqlist:
fq_path=os.path.join(fastq_dir,o2read)
do_fastqc(uuid,fq_path,thread_count,engine,logger)
fastqc_to_db(uuid,fq_path,engine,logger)
do_guess_encoding(uuid,fq_path,engine,logger)
guess_enc_db(uuid,fq_path,engine,logger)
fastq_length=fastq_util.get_fastq_length(uuid,fastq_dir,seread,engine,logger) # removable
return fastq_length
def bwa_mem(uuid, bam_path, reference_fasta_path, readgroup_path_dict, thread_count, engine, logger):
uuid_dir = os.path.dirname(bam_path)
logger.info("uuid_dir=%s" % uuid_dir)
fastq_dir = os.path.join(uuid_dir, "fastq")
logger.info("fastq_dir=%s" % fastq_dir)
realn_dir = os.path.join(uuid_dir, "realn")
logger.info("realn_dir=%s" % realn_dir)
os.makedirs(realn_dir, exist_ok=True)
fastqlist = fastq_util.buildfastqlist(fastq_dir)
logging.info("fastqlist=%s" % fastqlist)
pefastqdict = fastq_util.buildpefastqdict(fastqlist)
logger.info("pefastqdict=%s" % pefastqdict)
sefastqlist = fastq_util.buildsefastqlist(fastqlist)
logger.info("sefastqlist=%s" % sefastqlist)
o1fastqlist = fastq_util.buildo1fastqlist(fastqlist)
logger.info("o1fastqlist=%s" % o1fastqlist)
o2fastqlist = fastq_util.buildo2fastqlist(fastqlist)
logger.info("o2fastqlist=%s" % o2fastqlist)
bam_path_list = list()
for read1 in sorted(pefastqdict.keys()):
rg_str = bam_util.get_readgroup_str(read1, readgroup_path_dict, logger)
bam_path = bwa_mem_paired(
uuid,
bam_path,
fastq_dir,
read1,
pefastqdict[read1],
realn_dir,
reference_fasta_path,
rg_str,
thread_count,
engine,
logger,
)
bam_path_list.append(bam_path)
for seread in sefastqlist:
rg_str = bam_util.get_readgroup_str(seread, readgroup_path_dict, logger)
bam_path = bwa_mem_single(
uuid,
bam_path,
fastq_dir,
seread,
realn_dir,
"s",
reference_fasta_path,
rg_str,
thread_count,
engine,
logger,
)
bam_path_list.append(bam_path)
for o1read in o1fastqlist:
rg_str = bam_util.get_readgroup_str(o1read, readgroup_path_dict, logger)
bam_path = bwa_mem_single(
uuid,
bam_path,
fastq_dir,
o1read,
realn_dir,
"o1",
reference_fasta_path,
rg_str,
thread_count,
engine,
logger,
)
bam_path_list.append(bam_path)
for o2read in o2fastqlist:
rg_str = bam_util.get_readgroup_str(o2read, readgroup_path_dict, logger)
bam_path = bwa_mem_single(
uuid,
bam_path,
fastq_dir,
o2read,
realn_dir,
"o2",
reference_fasta_path,
rg_str,
thread_count,
engine,
logger,
)
bam_path_list.append(bam_path)
return bam_path_list
def bwa(uuid,bam_path,reference_fasta_path,readgroup_path_dict,thread_count,engine,logger):
uuid_dir=os.path.dirname(bam_path)
logger.info('uuid_dir=%s' % uuid_dir)
fastq_dir=os.path.join(uuid_dir,'fastq')
logger.info('fastq_dir=%s' % fastq_dir)
realn_dir=os.path.join(uuid_dir,'realn')
logger.info('realn_dir=%s' % realn_dir)
os.makedirs(realn_dir,exist_ok=True)
fastqlist=fastq_util.buildfastqlist(fastq_dir)
logging.info('fastqlist=%s' % fastqlist)
pefastqdict=fastq_util.buildpefastqdict(fastqlist)
logger.info('pefastqdict=%s' % pefastqdict)
sefastqlist=fastq_util.buildsefastqlist(fastqlist)
logger.info('sefastqlist=%s' % sefastqlist)
o1fastqlist=fastq_util.buildo1fastqlist(fastqlist)
logger.info('o1fastqlist=%s' % o1fastqlist)
o2fastqlist=fastq_util.buildo2fastqlist(fastqlist)
logger.info('o2fastqlist=%s' % o2fastqlist)
bam_path_list=list()
for read1 in sorted(pefastqdict.keys()):
rg_str=bam_util.get_readgroup_str(read1,readgroup_path_dict,logger)
fastq_length=fastq_util.get_fastq_length(uuid,fastq_dir,read1,engine,logger)
if fastq_length<MEM_ALN_CUTOFF:
bam_path=bwa_aln_paired(uuid,bam_path,fastq_dir,read1,pefastqdict[read1],realn_dir,reference_fasta_path,
rg_str,thread_count,engine,logger)
bam_path_list.append(bam_path)
else:
bam_path=bwa_mem_paired(uuid,bam_path,fastq_dir,read1,pefastqdict[read1],realn_dir,reference_fasta_path,
rg_str,thread_count,engine,logger)
bam_path_list.append(bam_path)
for seread in sefastqlist:
rg_str=bam_util.get_readgroup_str(seread,readgroup_path_dict,logger)
fastq_length=fastq_util.get_fastq_length(uuid,fastq_dir,seread,engine,logger)
if fastq_length<MEM_ALN_CUTOFF:
bam_path=bwa_aln_single(uuid,bam_path,fastq_dir,seread,realn_dir,'s',reference_fasta_path,
rg_str,thread_count,engine,logger)
bam_path_list.append(bam_path)
else:
bam_path=bwa_mem_single(uuid,bam_path,fastq_dir,seread,realn_dir,'s',reference_fasta_path,
rg_str,thread_count,engine,logger)
bam_path_list.append(bam_path)
for o1read in o1fastqlist:
rg_str=bam_util.get_readgroup_str(o1read,readgroup_path_dict,logger)
fastq_length=fastq_util.get_fastq_length(uuid,fastq_dir,o1read,engine,logger)
if fastq_length<MEM_ALN_CUTOFF:
bam_path=bwa_aln_single(uuid,bam_path,fastq_dir,o1read,realn_dir,'o1',reference_fasta_path,
rg_str,thread_count,engine,logger)
bam_path_list.append(bam_path)
else:
bam_path=bwa_mem_single(uuid,bam_path,fastq_dir,o1read,realn_dir,'o1',reference_fasta_path,
rg_str,thread_count,engine,logger)
bam_path_list.append(bam_path)
for o2read in o2fastqlist:
rg_str=bam_util.get_readgroup_str(o2read,readgroup_path_dict,logger)
fastq_length=fastq_util.get_fastq_length(uuid,fastq_dir,o2read,engine,logger)
if fastq_length<MEM_ALN_CUTOFF:
bam_path=bwa_aln_single(uuid,bam_path,fastq_dir,o2read,realn_dir,'o2',reference_fasta_path,
rg_str,thread_count,engine,logger)
bam_path_list.append(bam_path)
else:
bam_path=bwa_mem_single(uuid,bam_path,fastq_dir,o2read,realn_dir,'o2',reference_fasta_path,
rg_str,thread_count,engine,logger)
bam_path_list.append(bam_path)
return bam_path_list