def main():
time0 = time.time()
sys.stdout.write(
"\nThis script would join the spades fastg contigs according to the reference."
"\nIt would add extra gap nodes (N) and/or overlap nodes (?) in between the connectible nodes and generate "
" a new fastg file."
"\n"
"\nThis is a BETA version:"
"\nAlthough it will not produce error connections, it usually replicates the same right connection."
"\nDon't be surprised if you find any other bugs.\n")
require_commands()
global options
if not options.which_blast:
try_this_bin = os.path.join(GO_DEP_PATH, "ncbi-blast", "blastn")
if os.path.isfile(try_this_bin) and executable(try_this_bin):
options.which_blast = os.path.split(try_this_bin)[0]
if not executable(os.path.join(options.which_blast, "blastn")):
sys.stdout.write(os.path.join(options.which_blast, "blastn") + " not accessible!")
exit()
if not executable(os.path.join(options.which_blast, "makeblastdb")):
sys.stdout.write(os.path.join(options.which_blast, "makeblastdb") + " not accessible!")
exit()
# fastg to fasta
fasta_file = options.in_fastg_file
del_complementary(fasta_file)
# make blast database if not made
include_index = check_db(which_blast=options.which_blast)
len_db = len(read_fasta(options.reference_fa_base)[1][0])
# make blast
new_fasta_matrix = blast_and_call_new_matrix(fasta_file=fasta_file, index_files=include_index, out_file=fasta_file + '.blast_in', len_db=len_db, which_blast=options.which_blast)
# write out fastg
write_fasta(out_file=fasta_file + '.Ncontigs_added.' + fasta_file.split('.')[-1], matrix=new_fasta_matrix, overwrite=False)
remove_temp_files(fasta_file)
sys.stdout.write('\n\nTotal cost: '+str(time.time()-time0)+'\n\n')
def require_options():
usage = "Usage: rm_low_coverage_duplicated_contigs.py *.fastg"
parser = OptionParser(usage=usage)
parser.add_option('--cov-t', dest='coverage_threshold', default=0.12,
help='With ratio (coverage of query/coverage of subject) below which, '
'the query would be exposed to discarded. Default: 0.12')
parser.add_option('--len-t', dest='length_threshold', default=0.9,
help='With overlap (length of hit of query/ length of query) above which, '
'the query would be exposed to discarded. Default: 0.9')
parser.add_option('--blur', dest='blur_bases', default=False, action='store_true',
help='Replace hit low-coverage bases with N.')
parser.add_option('--keep-temp', dest='keep_temp', default=False, action='store_true',
help='Keep temp blast files.')
parser.add_option("--which-blast", dest="which_blast", default="",
help="Assign the path to BLAST binary files if not added to the path.")
parser.add_option('-o', dest='output_dir',
help='Output directory. Default: along with the original file')
parser.add_option('-t', '--threads', dest="threads", default=4, type=int,
help="Threads of blastn.")
options, args = parser.parse_args()
if not args:
parser.print_help()
sys.stdout.write('\n######################################\nERROR: Insufficient REQUIRED arguments!\n\n')
exit()
if not options.which_blast:
try_this_bin = os.path.join(GO_DEP_PATH, "ncbi-blast", "blastn")
if os.path.isfile(try_this_bin) and executable(try_this_bin):
options.which_blast = os.path.split(try_this_bin)[0]
if not executable(os.path.join(options.which_blast, "blastn")):
sys.stdout.write(os.path.join(options.which_blast, "blastn") + " not accessible!")
exit()
if not executable(os.path.join(options.which_blast, "makeblastdb")):
sys.stdout.write(os.path.join(options.which_blast, "makeblastdb") + " not accessible!")
exit()
if options.treat_no_hits not in ["ex_no_con", "ex_no_hit", "keep_all"]:
sys.stdout.write('\n\nOption Error: you should choose assign one of "ex_no_con", "ex_no_hit"'
' and "keep_all" to variable treat_no_hits\n')
exit()
return options, args
def require_options():
usage = "Usage: rm_low_coverage_duplicated_contigs.py *.fastg"
parser = ArgumentParser(usage=usage)
parser.add_argument(
'assemblies',
metavar='assemblies',
type=str,
nargs='+',
help=
"Input FASTG format assembly graph files (split the files by spaces).")
parser.add_argument(
'--cov-t',
dest='coverage_threshold',
default=0.12,
help='With ratio (coverage of query/coverage of subject) below which, '
'the query would be exposed to discarded. Default: 0.12')
parser.add_argument(
'--len-t',
dest='length_threshold',
default=0.9,
help=
'With overlap (length of hit of query/ length of query) above which, '
'the query would be exposed to discarded. Default: 0.9')
parser.add_argument('--blur',
dest='blur_bases',
default=False,
action='store_true',
help='Replace hit low-coverage bases with N.')
parser.add_argument('--keep-temp',
dest='keep_temp',
default=False,
action='store_true',
help='Keep temp blast files.')
parser.add_argument(
"--which-blast",
dest="which_blast",
default="",
help="Assign the path to BLAST binary files if not added to the path.")
parser.add_argument(
'-o',
dest='output_dir',
help='Output directory. Default: along with the original file')
parser.add_argument('-t',
'--threads',
dest="threads",
default=4,
type=int,
help="Threads of blastn.")
parser.add_argument(
"-v",
"--version",
action="version",
version="GetOrganelle v{version}".format(version=get_versions()))
options = parser.parse_args()
if not options.assemblies:
parser.print_help()
sys.stdout.write(
'\n######################################\nERROR: Insufficient REQUIRED arguments!\n\n'
)
exit()
if not options.which_blast:
try_this_bin = os.path.join(GO_DEP_PATH, "ncbi-blast", "blastn")
if os.path.isfile(try_this_bin) and executable(try_this_bin):
output, err = subprocess.Popen(try_this_bin + " -version",
stdout=subprocess.PIPE,
stderr=subprocess.STDOUT,
shell=True).communicate()
if "not found" in output.decode("utf8"):
sys.stdout.write(output.decode("utf8") + "\n")
else:
options.which_blast = os.path.split(try_this_bin)[0]
if not executable(os.path.join(options.which_blast, "blastn")):
sys.stdout.write(
os.path.join(options.which_blast, "blastn") + " not accessible!")
exit()
if not executable(os.path.join(options.which_blast, "makeblastdb")):
sys.stdout.write(
os.path.join(options.which_blast, "makeblastdb") +
" not accessible!")
exit()
if options.treat_no_hits not in ["ex_no_con", "ex_no_hit", "keep_all"]:
sys.stdout.write(
'\n\nOption Error: you should choose assign one of "ex_no_con", "ex_no_hit"'
' and "keep_all" to variable treat_no_hits\n')
exit()
return options, options.assemblies
dest="which_spades",
default="",
help="Assign the path to SPAdes binary files if not added to the path. "
"Default: try GetOrganelleDep/" + SYSTEM_NAME +
"/SPAdes first, then $PATH")
options, args = parser.parse_args()
if not (options.seed_dir and options.fastq_file_1 and options.fastq_file_2
and options.output_sh_file):
parser.print_help()
sys.stdout.write('\nERROR: Insufficient arguments!\n')
exit()
if options.fastq_file_1 == options.fastq_file_2:
raise IOError("1st fastq file should NOT be the same with 2nd fastq file!")
if not options.which_bowtie2:
try_this_bin = os.path.join(GO_DEP_PATH, "bowtie2", "bowtie2")
if os.path.isfile(try_this_bin) and executable(try_this_bin):
options.which_bowtie2 = os.path.split(try_this_bin)[0]
if not options.which_spades:
try_this_bin = os.path.join(GO_DEP_PATH, "SPAdes", "bin", "spades.py")
if os.path.isfile(try_this_bin) and executable(try_this_bin):
options.which_spades = os.path.split(try_this_bin)[0]
if not executable(os.path.join(options.which_bowtie2, "bowtie2")):
sys.stdout.write("Warning: " +
os.path.join(options.which_bowtie2, "bowtie2") +
" not accessible!")
if not executable(os.path.join(options.which_spades, "spades.py")):
sys.stdout.write("Warning: " +
os.path.join(options.which_spades, "spades.py") +
" not accessible!")
out_f_h = open(options.output_sh_file + '.sh', 'w')
else:
sys.stdout.write("skipped\n")
if changed:
os.remove(seed_file)
# check BLAST and make blast database
if os.path.exists(os.path.join(DEP_DIR, SYSTEM_NAME, "ncbi-blast")):
files_to_check = ["blastn", "makeblastdb"]
for check_f in files_to_check:
check_file_path = os.path.join(DEP_DIR, SYSTEM_NAME, "ncbi-blast", check_f)
if not os.path.exists(check_file_path):
raise EnvironmentError(check_file_path + " not exists!")
os.chmod(check_file_path, 0o755)
# TODO set overwrite=True
if executable(os.path.join(DEP_DIR, SYSTEM_NAME, "ncbi-blast", "makeblastdb")):
initialize_notation_database(which_blast=os.path.join(DEP_DIR, SYSTEM_NAME, "ncbi-blast"), overwrite=not RESUME)
elif executable("blastn"):
initialize_notation_database(which_blast="", overwrite=not RESUME)
else:
raise EnvironmentError("makeblastdb not found in the $PATH nor in " +
os.path.join(DEP_DIR, SYSTEM_NAME, "ncbi-blast") + "!\n"
"change directory to GetOrganelle and git clone git://github.com/Kinggerm/GetOrganelleDep\n"
"or get BLAST via http://ftp.ncbi.nlm.nih.gov/blast/executables/magicblast/LATEST")
# check Bowtie2 and build seed index
if os.path.exists(os.path.join(DEP_DIR, SYSTEM_NAME, "bowtie2")):
files_to_check = ["bowtie2", "bowtie2-align-l", "bowtie2-build", "bowtie2-build-l"]
for check_f in files_to_check:
check_file_path = os.path.join(DEP_DIR, SYSTEM_NAME, "bowtie2", check_f)
if not os.path.exists(check_file_path):