comparison tools/protein_analysis/psortb.py @ 18:eb6ac44d4b8e draft

Suite v0.2.8, record Promoter 2 verion + misc internal updates

17:e6cc27d182a8

with a # character as used elsewhere in Galaxy.
"""
import sys
import os
import tempfile
from seq_analysis_utils import stop_err, split_fasta, run_jobs, thread_count

FASTA_CHUNK = 500

if "-v" in sys.argv or "--version" in sys.argv:
    """Return underlying PSORTb's version"""
    sys.exit(os.system("psort --version"))

if len(sys.argv) != 8:
    stop_err("Require 7 arguments, number of threads (int), type (e.g. archaea), "
             "output (e.g. terse/normal/long), cutoff, divergent, input protein "
             "FASTA file & output tabular file")

num_threads = thread_count(sys.argv[1], default=4)
org_type = sys.argv[2]

tabular_file = sys.argv[7]

if out_type == "terse":
    header = ['SeqID', 'Localization', 'Score']
elif out_type == "normal":
    stop_err("Normal output not implemented yet, sorry.")
elif out_type == "long":
    if org_type == "-n":
        #Gram negative bacteria
        header = ['SeqID', 'CMSVM-_Localization', 'CMSVM-_Details', 'CytoSVM-_Localization', 'CytoSVM-_Details',
                  'ECSVM-_Localization', 'ECSVM-_Details', 'ModHMM-_Localization', 'ModHMM-_Details',

                  'Signal_a_Localization', 'Signal_a_Details',
                  'Cytoplasmic_Score', 'CytoplasmicMembrane_Score', 'Cellwall_Score',
                  'Extracellular_Score', 'Final_Localization', 'Final_Localization_Details', 'Final_Score',
                  'Secondary_Localization', 'PSortb_Version']
    else:
        stop_err("Expected -n, -p or -a for the organism type, not %r" % org_type)
else:
    stop_err("Expected terse, normal or long for the output type, not %r" % out_type)

tmp_dir = tempfile.mkdtemp()

def clean_tabular(raw_handle, out_handle):
    """Clean up tabular TMHMM output, returns output line count."""

        try:
            output = open(temp).readline()
        except IOError:
            output = ""
        clean_up(fasta_files + temp_files)
        stop_err("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")

    data_handle = open(temp)
    count += clean_tabular(data_handle, out_handle)
    data_handle.close()
    if not count:
        clean_up(fasta_files + temp_files)
        stop_err("No output from psortb")
out_handle.close()
print "%i records" % count

clean_up(fasta_files + temp_files)

18:eb6ac44d4b8e

with a # character as used elsewhere in Galaxy.
"""
import sys
import os
import tempfile
from seq_analysis_utils import sys_exit, split_fasta, run_jobs, thread_count

FASTA_CHUNK = 500

if "-v" in sys.argv or "--version" in sys.argv:
    """Return underlying PSORTb's version"""
    sys.exit(os.system("psort --version"))

if len(sys.argv) != 8:
    sys_exit("Require 7 arguments, number of threads (int), type (e.g. archaea), "
             "output (e.g. terse/normal/long), cutoff, divergent, input protein "
             "FASTA file & output tabular file")

num_threads = thread_count(sys.argv[1], default=4)
org_type = sys.argv[2]

tabular_file = sys.argv[7]

if out_type == "terse":
    header = ['SeqID', 'Localization', 'Score']
elif out_type == "normal":
    sys_exit("Normal output not implemented yet, sorry.")
elif out_type == "long":
    if org_type == "-n":
        #Gram negative bacteria
        header = ['SeqID', 'CMSVM-_Localization', 'CMSVM-_Details', 'CytoSVM-_Localization', 'CytoSVM-_Details',
                  'ECSVM-_Localization', 'ECSVM-_Details', 'ModHMM-_Localization', 'ModHMM-_Details',

                  'Signal_a_Localization', 'Signal_a_Details',
                  'Cytoplasmic_Score', 'CytoplasmicMembrane_Score', 'Cellwall_Score',
                  'Extracellular_Score', 'Final_Localization', 'Final_Localization_Details', 'Final_Score',
                  'Secondary_Localization', 'PSortb_Version']
    else:
        sys_exit("Expected -n, -p or -a for the organism type, not %r" % org_type)
else:
    sys_exit("Expected terse, normal or long for the output type, not %r" % out_type)

tmp_dir = tempfile.mkdtemp()

def clean_tabular(raw_handle, out_handle):
    """Clean up tabular TMHMM output, returns output line count."""

        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")

    data_handle = open(temp)
    count += clean_tabular(data_handle, out_handle)
    data_handle.close()
    if not count:
        clean_up(fasta_files + temp_files)
        sys_exit("No output from psortb")
out_handle.close()
print "%i records" % count

clean_up(fasta_files + temp_files)