parser.add_argument("-s", "--suffix_to_remove", action="store", dest="suffix_to_remove",
help="Suffix to remove from filename(not count extension)")
parser.add_argument("-g", "--gene_alias_file", action="store", dest="gene_alias_file",
help="File with aliases of genes (tab-separated, one alias per gene, no header)")
parser.add_argument("-w", "--write_dir_path", action="store_true", dest="write_dir_path",
help="write directory name(if directory is source of vcf files) in output file. Default: false")
parser.add_argument("-e", "--write_ext", action="store_true", dest="write_ext",
help="write extensions of vcf files in output file. Default: false")
parser.add_argument("-r", "--remove_nucleotide_substitutions", action="store_true", dest="rem_nuc_sub",
help="Remove nucleotide substitutions from output(preserve only AA substitutions)")
parser.add_argument("-c", "--convert_aa_to_single_letter", action="store_true", dest="convert_to_single_letter",
help="Convert aminoacids to single letters")
args = parser.parse_args()
args.input = make_list_of_path_to_files(args.input)
gene_alias_dict = SynDict()
if args.gene_alias_file:
gene_alias_dict.read(args.gene_alias_file, split_values=False)
out_fd = sys.stdout if args.output == "stdout" else open(args.output, "w")
summary_dict = TwoLvlDict()
for filename in args.input:
directory, prefix, extension = split_filename(filename)
if args.write_dir_path and args.write_ext:
name = filename
elif args.write_dir_path:
name = (directory + prefix) if directory else prefix
elif args.write_ext:
#!/usr/bin/env python
__author__ = 'Sergei F. Kliver'
import argparse
from RouToolPa.Tools.Abstract import Tool
from RouToolPa.Routines.File import make_list_of_path_to_files
parser = argparse.ArgumentParser()
parser.add_argument(
"-a",
"--group_a_list",
action="store",
dest="group_a_list",
required=True,
type=lambda s: make_list_of_path_to_files(s.split(",")),
help="Comma-separated list of files/directories with ids from group A")
parser.add_argument(
"-b",
"--group_b_list",
action="store",
dest="group_b_list",
required=True,
type=lambda s: make_list_of_path_to_files(s.split(",")),
help="Comma-separated list of files/directories with ids from group B")
parser.add_argument("-o",
"--output",
action="store",
dest="output",
help="Output file with allowed ids. Default - stdout")
parser.add_argument(
"-m",
"--write_dir_path",
action="store_true",
dest="write_dir_path",
help=
"Write directory name(if directory is source of files) in output file. Default: false"
)
parser.add_argument(
"-e",
"--write_ext",
action="store_true",
dest="write_ext",
help=
"Write extensions of files with sequences in output file. Default: false")
args = parser.parse_args()
files_list = sorted(make_list_of_path_to_files(args.input, file_filter))
out_fd = sys.stdout if args.output == "stdout" else open(args.output, "w")
if args.write_header:
out_fd.write("#file/sample\tnumber_of_sequences\n")
for filename in files_list:
if args.output != "stdout":
print("Counting variants in %s ..." % filename)
directory, prefix, extension = split_filename(filename)
filetype = detect_filetype_by_extension(filename)
number_of_sequences = 0
with open(filename, "r") as seq_fd:
try:
for record in SeqIO.parse(seq_fd, filetype):
number_of_sequences += 1
