logger.info("Existing binning output file: "+contig_bins_file)
logger.info("Final binning output file: "+output_path)
logger.info("Depth: "+str(depth))
logger.info("Threshold: "+str(threshold))
logger.info("Number of threads: "+str(nthreads))
logger.info("GraphBin2 started")
start_time = time.time()
# Get length and coverage of contigs
#--------------------------------------------------------
contig_lengths = {}
contig_names = BidirectionalMap()
contig_num = 0
with open(contig_paths, "r") as file:
for line in file.readlines():
if not line.startswith("#"):
strings = line.strip().split()
contig_names[contig_num] = strings[0]
contig_lengths[contig_num] = int(strings[1])
contig_num += 1
contig_names_rev = contig_names.inverse
str(n_bins))
except:
logger.error(
"Please make sure that the correct path to the binning result file is provided and it is having the correct format."
)
logger.info("Exiting GraphBin... Bye...!")
sys.exit(1)
logger.info("Constructing the assembly graph")
# Get the links from the .asqg file
#-----------------------------------
links = []
my_map = BidirectionalMap()
node_count = 0
try:
# Get contig connections from .asqg file
with open(assembly_graph_file) as file:
line = file.readline()
while line != "":
# Count the number of contigs
if "VT" in line:
start = 'contig-'
end = ''
contig_num = int(
while line != "":
strings = line.split("\t")
start_n = 'edge_'
end_n = ''
contig_num = int(
re.search('%s(.*)%s' % (start_n, end_n), strings[0]).group(1)) - 1
coverages[contig_num] = int(strings[1])
line = my_file.readline()
# Get the links from the .gfa file
#-----------------------------------
my_map = BidirectionalMap()
node_count = 0
my_names_map = BidirectionalMap()
links = []
# try:
# Get contig connections from .gfa file
with open(assembly_graph_file) as file:
line = file.readline()
while line != "":
# Count the number of contigs