comparison tools/protein_analysis/tmhmm2.py @ 21:238eae32483c draft

"Check this is up to date with all 2020 changes (black etc)"

20:a19b3ded8f33

itself (see the SignalP XML file for settings).

Also tmhmm2 can fail without returning an error code, for example if run on a
64 bit machine with only the 32 bit binaries installed. This script will spot
when there is no output from tmhmm2, and raise an error.
"""

from __future__ import print_function

import os
import sys

if "-v" in sys.argv or "--version" in sys.argv:
    sys.exit("TMHMM wrapper version 0.0.16")

if len(sys.argv) != 4:
    sys.exit("Require three arguments, number of threads (int), input protein FASTA file & output tabular file")

num_threads = thread_count(sys.argv[1], default=4)
fasta_file = sys.argv[2]
tabular_file = sys.argv[3]

        first60 = first60[8:]
        assert predhel.startswith("PredHel="), line
        predhel = predhel[8:]
        assert topology.startswith("Topology="), line
        topology = topology[9:]
        out_handle.write("%s\t%s\t%s\t%s\t%s\t%s\n"
                         % (identifier, length, exp_aa, first60, predhel, topology))
        count += 1
    return count


# Note that if the input FASTA file contains no sequences,
# split_fasta returns an empty list (i.e. zero temp files).
fasta_files = split_fasta(fasta_file, os.path.join(tmp_dir, "tmhmm"), FASTA_CHUNK)
temp_files = [f + ".out" for f in fasta_files]
jobs = ["tmhmm -short %s > %s" % (fasta, temp)
        for fasta, temp in zip(fasta_files, temp_files)]


def clean_up(file_list):
    """Remove temp files, and if possible the temp directory."""
    for f in file_list:

        try:
            output = open(temp).readline()
        except IOError:
            output = ""
        clean_up(fasta_files + temp_files)
        sys.exit("One or more tasks failed, e.g. %i from %r gave:\n%s" % (error_level, cmd, output),
                 error_level)
del results
del jobs

out_handle = open(tabular_file, "w")
out_handle.write("#ID\tlen\tExpAA\tFirst60\tPredHel\tTopology\n")

21:238eae32483c

itself (see the SignalP XML file for settings).

Also tmhmm2 can fail without returning an error code, for example if run on a
64 bit machine with only the 32 bit binaries installed. This script will spot
when there is no output from tmhmm2, and raise an error.
"""  # noqa: E501

from __future__ import print_function

import os
import sys

if "-v" in sys.argv or "--version" in sys.argv:
    sys.exit("TMHMM wrapper version 0.0.16")

if len(sys.argv) != 4:
    sys.exit(
        "Require three arguments, number of threads (int), input protein "
        "FASTA file & output tabular file"
    )

num_threads = thread_count(sys.argv[1], default=4)
fasta_file = sys.argv[2]
tabular_file = sys.argv[3]

        first60 = first60[8:]
        assert predhel.startswith("PredHel="), line
        predhel = predhel[8:]
        assert topology.startswith("Topology="), line
        topology = topology[9:]
        out_handle.write(
            "%s\t%s\t%s\t%s\t%s\t%s\n"
            % (identifier, length, exp_aa, first60, predhel, topology)
        )
        count += 1
    return count


# Note that if the input FASTA file contains no sequences,
# split_fasta returns an empty list (i.e. zero temp files).
fasta_files = split_fasta(fasta_file, os.path.join(tmp_dir, "tmhmm"), FASTA_CHUNK)
temp_files = [f + ".out" for f in fasta_files]
jobs = [
    "tmhmm -short %s > %s" % (fasta, temp)
    for fasta, temp in zip(fasta_files, temp_files)
]


def clean_up(file_list):
    """Remove temp files, and if possible the temp directory."""
    for f in file_list:

        try:
            output = open(temp).readline()
        except IOError:
            output = ""
        clean_up(fasta_files + temp_files)
        sys.exit(
            "One or more tasks failed, e.g. %i from %r gave:\n%s"
            % (error_level, cmd, output),
            error_level,
        )
del results
del jobs

out_handle = open(tabular_file, "w")
out_handle.write("#ID\tlen\tExpAA\tFirst60\tPredHel\tTopology\n")