def get_QL_trash_fpath(fpath, outdir_path, quality, length):
# Function configures path to trash (quality and length) file according to filter options.
#
# :param fpath: path to input file;
# :type fpath: str;
# :param outdir_path: path to output directory;
# :type outdir_path: str;
# :param quality: threshold for quality filter;
# :type quality: float;
# :param length: threshold for length filter;
# :type length: int or None, if filter is disabled;
trash_fpath = os.path.join(outdir_path, "trash")
# Add quality to name name of trash file
if not is_fasta(fpath):
trash_fpath += "-q{}".format(quality)
# end if
# Add length to name name of trash file
if not length is None:
trash_fpath += "-m{}".format(length)
# end if
# Add appropriate extention
trash_fpath += re.search(ext_pattern, fpath).group(1)
return trash_fpath
def get_QL_filter(fpath, quality, length):
# Function returns a filter function.
# This filter in turn returns True if annotation line
# passed to it passes all filters and False otherwise.
# Returns quality and length filter.
#
# :param fpath: path to input file;
# :type fpath: str;
# :param quality: threshold for quality filter;
# :type quality: float;
# :param length: threshold for length filter;
# :type length: int or None, if filter is disabled;
filters = list()
# Ad quality filter
# There will be minus instead of quality for fasta files
if not is_fasta(fpath):
# Someone can try to classify fasta files and then bin fast5 accoeding to the classification.
# In this case TypeError will be raised. Then just return True.
def qual_filter(x):
try:
return x[0] >= quality
except TypeError:
return True
# end try
# end def qual_filter
filters.append(qual_filter)
# end if
# Add length filter
if not length is None:
filters.append(lambda x: x[1] >= length)
# end if
# Return "integral" filter
return lambda line: all((f(line) for f in filters))
from glob import glob
try:
opts, args = getopt.gnu_getopt(sys.argv[1:], "hvr:d:o:s:q:m:i:c:ut:n", [
"help", "version", "taxannot-resdir=", "indir=", "outdir=",
"binning-sensitivity=", "min-qual=", "min-seq-len=", "min-pident=",
"min-coverage=", "untwist-fast5", "threads=", "no-trash"
])
except getopt.GetoptError as gerr:
print(str(gerr))
platf_depend_exit(2)
# end try
from src.filesystem import is_fasta, is_fastq, is_fast5
is_fastqa = lambda f: is_fasta(f) or is_fastq(f)
from datetime import datetime
now = datetime.now().strftime("%Y-%m-%d %H.%M.%S")
# |== Default parameters: ==|
fq_fa_list = list() # list with input FASTQ and FASTA files paths
fast5_list = list() # list with input FAST5 files paths
tax_annot_res_dir = "barapost_result" # path to directory with classification results
indir_path = None # path to input directory
outdir_path = "binning_result_{}".format(now.replace(
' ', '_')) # path to output directory
sens = ("genus", 5) # binning sensitivity
min_qual = 10 # minimum mean read quality to keep
min_qlen = None # minimum seqeunce length to keep
def test_is_fasta_txt(self, fpath_txt: str):
# Should not recognize file with `.txt` extention
assert fls.is_fasta(fpath_txt) == False
def test_is_fasta_fa_bz2(self, fpath_fa_bz2: str):
# Should not recognize file with `.fa.bz2` extention
assert fls.is_fasta(fpath_fa_bz2) == False
def test_is_fasta_fa_gz(self, fpath_fa_gz: str):
# Should recognize file with `.fa.gz` extention
assert fls.is_fasta(fpath_fa_gz) == True
def test_is_fasta_fa(self, fpath_fa: str):
# Should recognize file with `.fa` extention
assert fls.is_fasta(fpath_fa) == True
