def get_output_genome_directory_list(option_map, config_map, genome_directory_list):
# output genome directory list
output_genome_directory_list = []
# get working directory
working_directory = option_map[working_directory_str]
# get reference_genome_directory
reference_genome_directory = utils.set_directory_path(config_map[ConfigString.reference_genome_directory_str])
parent_genome_directory_name = "sigma_alignments_output"
# get parent genome directory
parent_genome_directory = working_directory + parent_genome_directory_name
# mkdir parent_genome_directory
if not os.path.isdir(parent_genome_directory):
check_call(["mkdir", parent_genome_directory], stdout = PIPE, stderr = sys.stderr)
# for loop genome_index_base_list
for genome_directory in genome_directory_list:
# retrieve genome_directory_child_path
genome_name = os.path.basename(genome_directory)
child_genome_directory = utils.set_directory_path(parent_genome_directory) + genome_name
output_genome_directory_list.append(child_genome_directory)
# mkdir child_genome_directory
if not os.path.isdir(child_genome_directory):
check_call(["mkdir", child_genome_directory], stdout = PIPE, stderr = sys.stderr)
return output_genome_directory_list
def get_genome_index_base(genome_directory_list):
# genome_index_base_list
genome_index_base_list = [] # initialize dictionary
# for loop genome_index_base_list
for genome_directory in genome_directory_list:
# retrieve genome_directory_child_path
genome_name = os.path.basename(genome_directory)
# set bowtie_index_count as 0
bowtie_index_count = 0
# get genome_index_base
genome_index_base = utils.set_directory_path(genome_directory) + genome_name
# loop genome_index_path
for filename in os.listdir(genome_directory):
if utils.check_bowtie_index_format(filename):
bowtie_index_count += 1
# if bowtie index exist, then save to list
if bowtie_index_count > 0:
genome_index_base_list.append(genome_index_base)
# check genome index == 0
if len(genome_index_base_list) == 0:
sys.stderr.write("\n** Cannot open bowtie2 index files. Check the Reference_Genome_Directory in config file.\n")
utils.die("** Program exit!")
return genome_index_base_list
def get_bamlog_path(config_path, genome_output_directory):
# get sam log filename
config_file = os.path.basename(config_path)
(config_base, config_ext) = os.path.splitext(config_file)
genome_name = os.path.basename(genome_output_directory)
bamlog_path = utils.set_directory_path(genome_output_directory) + genome_name + ".align.log"
return bamlog_path
def get_filtered_bamout_path(config_path, genome_output_directory):
# get sam output filename
config_file = os.path.basename(config_path)
(config_base, config_ext) = os.path.splitext(config_file)
genome_name = os.path.basename(genome_output_directory)
filtered_bamout_path = utils.set_directory_path(genome_output_directory) + genome_name + ".filtered.bam"
return filtered_bamout_path
def search_genome_fasta_path(config_map):
# retrieve options
reference_genome_directory = utils.set_directory_path(config_map[ConfigString.reference_genome_directory_str])
# genome data structure
genome_directory_list = []
genome_fasta_path_list = []
fasta_count = 0
# loop the genome directory
dirs = [d for d in os.listdir(reference_genome_directory) if os.path.isdir(os.path.join(reference_genome_directory, d))]
for genome_enum, genome_name in enumerate(dirs):
# get genome directory
genome_directory = reference_genome_directory + genome_name
genome_directory_list.append(genome_directory)
genome_fasta_path_list.append([])
# loop fasta files
for filename in os.listdir(genome_directory):
# check fasta file
if utils.check_fasta_format(filename):
fasta_path = utils.set_directory_path(genome_directory) + filename
genome_fasta_path_list[genome_enum].append(fasta_path)
fasta_count += 1
# check no orgaism list
if len(genome_directory_list) == 0:
sys.stderr.write("\n** Cannot search genome directory. Check the reference genome directory.\n")
utils.die("** Program exit!")
# check no fasta file
if fasta_count == 0:
sys.stderr.write("\n** Cannot find genome fasta file. Check the reference genome directory.\n")
utils.die("** Program exit!")
return (genome_directory_list, genome_fasta_path_list)
def get_output_genome_directory_list_wo(option_map, config_map):
# output genome directory list
output_genome_directory_list = []
# get working directory
working_directory = option_map[working_directory_str]
# get reference_genome_directory
reference_genome_directory = utils.set_directory_path(config_map[ConfigString.reference_genome_directory_str])
parent_genome_directory_name = os.path.basename(os.path.normpath(reference_genome_directory))
# get parent genome directory
parent_genome_directory = working_directory + parent_genome_directory_name
# mkdir parent_genome_directory
if not os.path.isdir(parent_genome_directory):
sys.stderr.write("\n** Cannot open output directory. Check the Reference_Genome_Directory in config file exists in the working directory.\n")
utils.die("** Program exit!")
# loop the genome directory
dirs = [d for d in os.listdir(parent_genome_directory) if os.path.isdir(os.path.join(parent_genome_directory, d))]
for genome_enum, genome_name in enumerate(dirs):
# get genome directory
child_genome_directory = utils.set_directory_path(parent_genome_directory) + genome_name
# loop fasta files
for filename in os.listdir(child_genome_directory):
# check bam file exists
if utils.check_bam_format(filename):
# append the genome directory into the list only if bam file exists
output_genome_directory_list.append(child_genome_directory)
return output_genome_directory_list
def get_output_diff_sequence_path_sublist(output_genome_directory, genome_fasta_path_sublist):
# initialize list
output_diff_sequence_path_sublist = []
# for loop
for genome_fasta_path in genome_fasta_path_sublist:
genome_fasta_name = os.path.basename(genome_fasta_path)
(genome_fasta_name_base, genome_fasta_name_ext) = os.path.splitext(genome_fasta_name)
output_diff_sequence_name = genome_fasta_name_base + ".SNPs.txt"
output_diff_sequence_path = utils.set_directory_path(output_genome_directory) + output_diff_sequence_name
# append to list
output_diff_sequence_path_sublist.append(output_diff_sequence_path)
return output_diff_sequence_path_sublist