#!/usr/bin/env python
__author__ = 'Sergei F. Kliver'
import argparse
from RouToolPa.Routines import EggNOGRoutines
parser = argparse.ArgumentParser()
parser.add_argument(
"-i",
"--input_pep_dir",
action="store",
dest="input_pep_dir",
required=True,
type=EggNOGRoutines.check_path,
help="Directory with input proteins(splited by families or something like)"
)
parser.add_argument("-o",
"--output_pep_dir",
action="store",
dest="output_pep_dir",
required=True,
type=EggNOGRoutines.check_path,
help="Directory to write proteins splited by species")
args = parser.parse_args()
EggNOGRoutines.split_proteins_per_species(args.input_pep_dir,
args.output_pep_dir,
input_format="fasta",
output_format="fasta")
#!/usr/bin/env python
__author__ = 'Sergei F. Kliver'
import argparse
from RouToolPa.Routines import EggNOGRoutines
parser = argparse.ArgumentParser()
parser.add_argument("-d",
"--dir_with_alignments",
action="store",
dest="dir_with_alignments",
required=True,
help="Directory with alignments")
parser.add_argument("-o",
"--output_dir",
action="store",
dest="output_dir",
required=True,
help="Directory to write extracted proteins")
args = parser.parse_args()
EggNOGRoutines.extract_proteins_from_alignments(args.dir_with_alignments,
args.output_dir)
#!/usr/bin/env python
__author__ = 'Sergei F. Kliver'
import sys
import argparse
from RouToolPa.Routines import EggNOGRoutines
parser = argparse.ArgumentParser()
parser.add_argument("-i", "--fam_file", action="store", dest="fam_file", required=True,
help="File with families obtained from hits to hmm profiles of EggNOG")
parser.add_argument("-o", "--output", action="store", dest="output", default="stdout",
help="File to write output")
args = parser.parse_args()
out_fd = sys.stdout if args.output == "stdout" else open(args.output, "w")
EggNOGRoutines.edit_profile_names_in_fam_file(args.fam_file, args.output)
args = parser.parse_args()
hits_file = "%s/%s.hits" % (args.output_dir, args.output_prefix)
top_hits_file = "%s/%s.top_hits" % (args.output_dir, args.output_prefix)
fam_file = "%s/%s.fam" % (args.output_dir, args.output_prefix)
ortholog_file = "%s/%s.orthologs" % (args.output_dir, args.output_prefix)
HMMER3.threads = 1
HMMER3.path = args.path
HMMER3.timelog = "%s/%s.timelog" % (args.output_dir, args.output_prefix)
HMMER3.parallel_hmmscan(args.input,
args.input_seq,
args.output_prefix,
args.output_dir,
num_of_seqs_per_scan=None,
threads=args.threads,
combine_output_to_single_file=True,
dont_output_alignments=args.no_alignment,
biopython_165_compartibility=True)
HMMER3.extract_top_hits(hits_file,
args.output_prefix,
parsing_mode=args.parsing_mode)
HMMER3.get_families_from_top_hits(top_hits_file, fam_file)
EggNOGRoutines.edit_profile_names_in_fam_file(fam_file, ortholog_file)
"""
for tmp_dir in args.hmmscan_output_dir, args.tblout_dir, args.domtblout_dir, args.pfamtblout_dir, "splited_fasta":
shutil.rmtree(tmp_dir)
"""
help="File with common names and associated ensembl ids. "
"Format of lines: common name\t<list of ensembl ids>")
parser.add_argument(
"-s",
"--species_id",
action="store",
dest="species_id",
help="NCBI taxonomy id of the species. "
"If not set ensembl ids will be treated as already labeled by species id")
parser.add_argument("-o",
"--output_prefix",
action="store",
dest="output_prefix",
required=True,
help="Output_prefix")
parser.add_argument("-r",
"--allow_repeating_of_keys",
action="store_true",
dest="allow_repeating_of_keys",
help="Allow repeating of keys in synonym file")
args = parser.parse_args()
EggNOGRoutines.extract_eggnog_fam_by_protein_syn_dict(
SynDict(filename=args.input, split_values=True),
SynDict(filename=args.protein_common_names_synonyms,
split_values=True,
allow_repeats_of_key=args.allow_repeating_of_keys),
output_prefix=args.output_prefix,
species_id=args.species_id)
#!/usr/bin/env python
__author__ = 'Sergei F. Kliver'
import sys
import argparse
from RouToolPa.Routines import EggNOGRoutines
parser = argparse.ArgumentParser()
parser.add_argument("-i", "--members_file", action="store", dest="members_file", required=True,
help="Members .tsv file from EggNOG")
parser.add_argument("-o", "--output", action="store", dest="output", default="stdout",
help="File to write output")
args = parser.parse_args()
out_fd = sys.stdout if args.output == "stdout" else open(args.output, "w")
EggNOGRoutines.convert_members_tsv_to_fam(args.members_file, args.output)
from RouToolPa.Routines import EggNOGRoutines
parser = argparse.ArgumentParser()
parser.add_argument("-i",
"--input",
action="store",
dest="input",
required=True,
help="Eggnog tsv file")
parser.add_argument("-o",
"--output_prefix",
action="store",
dest="output_prefix",
required=True,
help="Prefix of output files")
parser.add_argument(
"-e",
"--email",
action="store",
dest="email",
help=
"E-mail for request ot NCBI Taxonomy. If not set species ids will not be converted to species"
"names by queries to NCBI Taxonomy database(requires Internet connection)")
args = parser.parse_args()
EggNOGRoutines.get_species_from_eggnog_tsv(args.input,
args.output_prefix,
email=args.email)