def extract_proteins_from_alignments(dir_with_alignments, output_dir):
out_dir = FileRoutines.check_path(output_dir)
print type(FileRoutines)
input_files = make_list_of_path_to_files(
[dir_with_alignments] if isinstance(dir_with_alignments, str
) else dir_with_alignments)
FileRoutines.safe_mkdir(out_dir)
from Routines import MultipleAlignmentRoutines
for filename in input_files:
filename_list = FileRoutines.split_filename(filename)
output_file = "%s%s%s" % (out_dir, filename_list[1],
filename_list[2])
MultipleAlignmentRoutines.extract_sequences_from_alignment(
filename, output_file)
__author__ = 'Sergei F. Kliver'
import os
import argparse
from Bio import AlignIO
from Routines import MultipleAlignmentRoutines
from Routines.File import check_path, make_list_of_path_to_files, save_mkdir, split_filename
parser = argparse.ArgumentParser()
parser.add_argument(
"-i",
"--input",
action="store",
dest="input",
type=lambda x: make_list_of_path_to_files(x.split(",")),
help="Comma-separated list of files or directory with files "
"containing alignments(one alignment per file)")
parser.add_argument("-n",
"--max_gap_number",
action="store",
dest="max_gap_number",
default=0,
type=int,
help="Maximum number of gaps to retain column")
parser.add_argument("-o",
"--output_directory",
action="store",
dest="output",
type=check_path,
help="Output directory")
dest="draw_distribution",
help="Draw distribution of kmers")
parser.add_argument("-j",
"--jellyfish_path",
action="store",
dest="jellyfish_path",
help="Path to jellyfish")
parser.add_argument("-n",
"--dont_extract_kmer_list",
action="store_true",
dest="dont_extract_kmer_list",
help="Don't extract kmer list")
args = parser.parse_args()
args.input = make_list_of_path_to_files(args.input)
if args.count_both_strands and args.add_rev_com:
raise ValueError(
"Options -b/--count_both_strands and -r/--add_reverse_complement are not compatible"
)
if args.add_rev_com:
file_with_rev_com = args.base_prefix + "_with_rev_com.fasta"
record_dict = SeqIO.index_db("temp_index.idx", args.input, format="fasta")
SeqIO.write(rev_com_generator(record_dict, yield_original_record=True),
file_with_rev_com, "fasta")
args.base_prefix += "_with_rev_com"
base_file = "%s_%i_mer.jf" % (args.base_prefix, args.kmer_length)
kmer_table_file = "%s_%i_mer.counts" % (args.base_prefix, args.kmer_length)
kmer_file = "%s_%i_mer.kmer" % (args.base_prefix, args.kmer_length)
#!/usr/bin/env python
__author__ = 'Sergei F. Kliver'
import argparse
from Tools.Annotation import Exonerate
from Routines.File import make_list_of_path_to_files
parser = argparse.ArgumentParser()
parser.add_argument(
"-i",
"--input",
action="store",
dest="input",
required=True,
type=lambda s: make_list_of_path_to_files(s.split(",")),
help="Input comma-separated list of files/directories with exonerate output"
)
parser.add_argument("-o",
"--output_prefix",
action="store",
dest="output_prefix",
help="Prefix of output files")
args = parser.parse_args()
Exonerate.split_output(args.input, args.output_prefix)
def make_list_of_path_to_files_from_comma_sep_string(string):
return make_list_of_path_to_files(string.split(","))
type=float,
help="Maximum number of iterations")
parser.add_argument("-f",
"--offset",
action="store",
dest="offset",
type=float,
help="Offset (works like gap extension penalty)")
parser.add_argument("-g",
"--gap_open_penalty",
action="store",
dest="gap_open_penalty",
type=float,
help="Gap open penalty")
args = parser.parse_args()
save_mkdir(args.output)
MAFFT.threads = args.threads
MAFFT.parallel_align(make_list_of_path_to_files(args.input),
args.output,
output_suffix="alignment",
gap_open_penalty=args.gap_open_penalty,
offset=args.offset,
maxiterate=args.maxiterate,
quiet=args.quiet,
mode=args.mode,
number_of_processes=args.processes,
anysymbol=True)
"--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