def test_find_database_index_index(self):
"""
Test the find database index function with an index as input
"""
db_index = utilities.find_database_index(cfg.bowtie2_db_index,
"bowtie2")
self.assertEqual(db_index, cfg.bowtie2_db_index)
def update_configuration(args):
""" Update the run settings based on the arguments provided """
# get the full path for the output directory
args.output_dir = os.path.abspath(args.output_dir)
# set if temp output should be removed
args.remove_temp_output = not args.store_temp_output
# if intermediate output should be removed, then also remove temp output
if args.remove_intermediate_output:
args.remove_temp_output = True
# check the input files are non-empty and readable
args.input = []
if (args.input1 and args.input2):
args.input.append(os.path.abspath(args.input1))
args.input.append(os.path.abspath(args.input2))
if (args.unpaired):
args.input.append(os.path.abspath(args.unpaired))
utilities.is_file_readable(args.input[0], exit_on_error=True)
if len(args.input) == 2:
utilities.is_file_readable(args.input[1], exit_on_error=True)
elif len(args.input) > 2:
sys.exit("ERROR: Please provide at most 2 input files.")
elif len(args.input) == 0:
sys.exit(
"ERROR: Please provide --input1/--input2 or --unpaired (input) files."
)
#Store original file paths for FASTQC
for input in args.input:
original_input_files.append(input)
# create the output directory if needed
utilities.create_directory(args.output_dir)
# set bowtie2 options
if args.bowtie2_options:
# parse the options from the user into any array of options
args.bowtie2_options = utilities.format_options_to_list(
args.bowtie2_options)
else:
# if not set by user, then set to default options
args.bowtie2_options = config.bowtie2_options
# add the quality scores to the bowtie2 options
args.bowtie2_options += [
config.bowtie2_flag_start + args.trimmomatic_quality_scores
]
# set the mode for single end input file
if len(args.input) == 1:
args.decontaminate_pairs = "unpaired"
# set the bowtie2 mode based on the pairs input
args.discordant = False
if args.decontaminate_pairs != "lenient":
args.discordant = True
# update the quality score option into a flag for trimmomatic
args.trimmomatic_quality_scores = config.trimmomatic_flag_start + args.trimmomatic_quality_scores
# find the location of trimmomatic, trimmomatic does not need to be executable
if not args.bypass_trim:
args.trimmomatic_path = utilities.find_dependency(
args.trimmomatic_path,
config.trimmomatic_jar,
"trimmomatic",
"--trimmomatic",
bypass_permissions_check=True)
# find the location of bmtagger, if set to run
if args.reference_db:
if args.bmtagger:
args.bmtagger_path = utilities.find_dependency(
args.bmtagger_path,
config.bmtagger_exe,
"bmtagger",
"--bmtagger",
bypass_permissions_check=False)
# add this folder to path, so as to be able to find other dependencies like bmfilter
utilities.add_exe_to_path(os.path.dirname(args.bmtagger_path))
else:
# find the location of bowtie2, if not running with bmtagger
args.bowtie2_path = utilities.find_dependency(
args.bowtie2_path,
config.bowtie2_exe,
"bowtie2",
"--bowtie2",
bypass_permissions_check=False)
# find the location of trf, if set to run
if not args.bypass_trf:
args.trf_path = utilities.find_dependency(
args.trf_path,
config.trf_exe,
"trf",
"--trf",
bypass_permissions_check=False)
# if fastqc is set to be run, check if the executable can be found
if args.fastqc_start or args.fastqc_end:
args.fastqc_path = utilities.find_dependency(
args.fastqc_path,
config.fastqc_exe,
"fastqc",
"--fastqc",
bypass_permissions_check=False)
# set the default output prefix
if args.output_prefix == None:
if args.input[0].endswith(".gz") or args.input[0].endswith(".bz2"):
# remove compression extension if present
infile_base = os.path.splitext(
os.path.splitext(os.path.basename(args.input[0]))[0])[0]
else:
infile_base = os.path.splitext(os.path.basename(args.input[0]))[0]
args.output_prefix = infile_base + "_kneaddata"
# find the bowtie2 indexes for each of the reference databases
# reference database inputs can be directories, indexes, or index files
if args.reference_db:
reference_indexes = []
database_type = "bowtie2"
if args.bmtagger:
database_type = "bmtagger"
for directory in args.reference_db:
reference_indexes.append(
utilities.find_database_index(os.path.abspath(directory),
database_type))
args.reference_db = reference_indexes
return args