def initialize_seed_database(which_bowtie2,
fasta_f,
overwrite=False,
verbose=False):
# bowtie index
new_seed_file = fasta_f + ".modified"
changed = check_fasta_seq_names(fasta_f, new_seed_file)
if changed:
seed_file = new_seed_file
else:
seed_file = fasta_f
output_base = remove_db_postfix(fasta_f) + ".index"
sys.stdout.write("bowtie2-build " + os.path.basename(fasta_f) + " ... ")
sys.stdout.flush()
if overwrite or sum([
os.path.exists(output_base + postfix)
for postfix in (".1.bt2l", ".2.bt2l", ".3.bt2l", ".4.bt2l",
".rev.1.bt2l", ".rev.2.bt2l")
]) != 6:
build_bowtie2_db(seed_file=seed_file,
seed_index_base=output_base,
which_bowtie2=which_bowtie2,
overwrite=overwrite,
random_seed=12345,
silent=verbose,
verbose_log=verbose)
sys.stdout.write("finished\n")
else:
sys.stdout.write("skipped\n")
if changed:
os.remove(seed_file)
def initialize_seed_database(which_bowtie2, overwrite=False):
for fasta_f in os.listdir(SEQ_DIR):
if fasta_f.endswith(".fasta") and fasta_f[:-6] in ("embplant_pt", "other_pt", "embplant_mt", "embplant_nr",
"animal_mt", "fungus_mt"):
fasta_f = os.path.join(SEQ_DIR, fasta_f)
new_seed_file = fasta_f + ".modified"
changed = check_fasta_seq_names(fasta_f, new_seed_file)
if changed:
seed_file = new_seed_file
else:
seed_file = fasta_f
output_base = remove_db_postfix(fasta_f) + ".index"
sys.stdout.write("bowtie2-build " + output_base + " ... ")
sys.stdout.flush()
if overwrite or sum([os.path.exists(output_base + postfix)
for postfix in
(".1.bt2l", ".2.bt2l", ".3.bt2l", ".4.bt2l", ".rev.1.bt2l", ".rev.2.bt2l")]) != 6:
build_bowtie2_db(seed_file=seed_file, seed_index_base=output_base, which_bowtie2=which_bowtie2,
overwrite=overwrite, random_seed=12345, silent=True)
sys.stdout.write("finished\n")
else:
sys.stdout.write("skipped\n")
if changed:
os.remove(seed_file)