def makeblastdb(cls):
fobj_list = utl.check_for_files(
"{}*.nsq".format(glv.conf.blastdb))
# exit blastdb
if len(fobj_list) != 0:
return
root_ext_pair = os.path.splitext(glv.conf.ref_fasta)
cmd1 = ''
if root_ext_pair[1] == '.gz':
bgzip = "bgzip -cd -@ {} {}"
mkdb = "makeblastdb -in - -title {} -dbtype nucl -out {}"
cmd1 = "{} | {}".format(
bgzip.format(
glv.conf.parallele_full_thread,
glv.conf.ref_fasta),
mkdb.format(
glv.conf.blastdb_title,
glv.conf.blastdb))
else:
mkdb = "makeblastdb -in {} -title {} -dbtype nucl -out {}"
cmd1 = "{}".format(
mkdb.format(
glv.conf.ref_fasta,
glv.conf.blastdb_title,
glv.conf.blastdb))
utl.try_exec(cmd1)
def _copy_ini_file(self):
# ini file
self.ini_file_path
# out_dir
self.out_dir
# back up
ini_base = os.path.basename(self.ini_file_path)
out_dir_ini_file = "{}/{}".format(self.out_dir, ini_base)
utl.save_to_tmpfile(out_dir_ini_file)
cmd = "cp {} {}".format(self.ini_file_path, out_dir_ini_file)
utl.try_exec(cmd)
def prepare_ref(self):
# ref_fasta_user: existence confirmation
if os.path.isfile(glv.conf.ref_fasta_user):
log.info("{} found.".format(glv.conf.ref_fasta_user))
else:
log.info("{} not found. exit.".format(glv.conf.ref_fasta_user))
sys.exit(1)
# ext, basename, without_ext
# https://note.nkmk.me/python-os-basename-dirname-split-splitext/
basename_user = os.path.basename(glv.conf.ref_fasta_user)
root_ext_pair = os.path.splitext(glv.conf.ref_fasta_user)
without_ext = root_ext_pair[0]
basename_without_ext = os.path.basename(without_ext)
ext = root_ext_pair[1]
# ref_fasta_slink_system
# symbolic link user's fasta to sys_ref_dir as .org(.gz)
if ext == '.gz':
glv.conf.ref_fasta_slink_system = "{}/{}{}".format(
glv.conf.ref_dir, basename_user, '.org_slink.gz')
# for blast
glv.conf.blastdb_title = basename_without_ext
else:
glv.conf.ref_fasta_slink_system = "{}/{}{}".format(
glv.conf.ref_dir, basename_user, '.org_slink')
# for blast
glv.conf.blastdb_title = basename_user
glv.conf.blastdb = "{}/{}{}".format(
glv.conf.ref_dir, glv.conf.blastdb_title, '.blastdb')
log.info("glv.conf.blastdb={}".format(glv.conf.blastdb))
if os.path.isfile(glv.conf.ref_fasta_slink_system):
log.info("{} exist.".format(glv.conf.ref_fasta_slink_system))
else:
utl.ln_s(
glv.conf.ref_fasta_user, glv.conf.ref_fasta_slink_system)
log.info("ext ({}).".format(ext))
# convert to bgz if ext is .gz and set to ref_fasta
if ext == '.gz':
# it should be convert to bgz in ref_dir
glv.conf.ref_fasta = "{}/{}".format(
glv.conf.ref_dir, basename_user)
log.info("ext {}, glv.conf.ref_fasta={}.".format(
ext, glv.conf.ref_fasta))
# half of thread?
cmd1 = 'bgzip -cd -@ {} {} | bgzip -@ {} > {}'.format(
glv.conf.parallele_full_thread,
glv.conf.ref_fasta_slink_system,
glv.conf.parallele_full_thread,
glv.conf.ref_fasta)
else:
# it should be convert to bgz in ref_dir
glv.conf.ref_fasta = "{}/{}{}".format(
glv.conf.ref_dir,
basename_user,
'.gz')
cmd1 = 'bgzip -c -@ {} {} > {}'.format(
glv.conf.parallele_full_thread,
glv.conf.ref_fasta_slink_system,
glv.conf.ref_fasta)
# execute
if os.path.isfile(glv.conf.ref_fasta):
log.info("{} exist.".format(glv.conf.ref_fasta))
else:
log.info("{} not exist. do cmd={}".format(
glv.conf.ref_fasta, cmd1))
utl.try_exec(cmd1)
# make fai file
cmd2 = 'samtools faidx {}'.format(
glv.conf.ref_fasta, glv.conf.log_dir)
glv.conf.ref_fasta_fai = "{}{}".format(glv.conf.ref_fasta, '.fai')
if os.path.isfile(glv.conf.ref_fasta_fai):
log.info("{} exist.".format(glv.conf.ref_fasta_fai))
else:
utl.try_exec(cmd2)
# read fasta to dict vprimer.cnf.refseq
glv.conf.ref_fasta_pickle = "{}{}".format(
glv.conf.ref_fasta, '.pickle')
self._read_fasta()
# ref to makeblastdb
Blast.makeblastdb()
return self