comparison tools/protein_analysis/promoter2.py @ 26:20139cb4c844 draft

planemo upload for repository https://github.com/peterjc/pico_galaxy/tools/protein_analysis commit 221d4187992cbb993e02dc3ea0ef0150c7916a4a-dirty

25:41a42022f815

"""
import sys
import os
import commands
import tempfile
from seq_analysis_utils import stop_err, split_fasta, run_jobs, thread_count

FASTA_CHUNK = 500

if len(sys.argv) != 4:
    stop_err("Require three arguments, number of threads (int), input DNA FASTA file & output tabular file. "
             "Got %i arguments." % (len(sys.argv)-1))

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

def get_path_and_binary():
    platform = commands.getoutput("uname") #e.g. Linux
    shell_script = commands.getoutput("which promoter")
    if not os.path.isfile(shell_script):
        stop_err("ERROR: Missing promoter executable shell script")
    path = None
    for line in open(shell_script):
        if line.startswith("setenv"): #could then be tab or space!
            parts = line.rstrip().split(None, 2)
            if parts[0] == "setenv" and parts[1] == "PROM":
                path = parts[2]
    if not path:
        stop_err("ERROR: Could not find promoter path (PROM) in %r" % shell_script)
    if not os.path.isdir(path):
        stop_error("ERROR: %r is not a directory" % path)
    bin = "%s/bin/promoter_%s" % (path, platform)
    if not os.path.isfile(bin):
        stop_err("ERROR: Missing promoter binary %r" % bin)
    return path, bin

def make_tabular(raw_handle, out_handle):
    """Parse text output into tabular, return query count."""
    identifier = None

            try:
                position, score, likelihood = line.strip().split(None,2)
            except ValueError:
                print "WARNING: Problem with line: %r" % line
                continue
                #stop_err("ERROR: Problem with line: %r" % line)
            if likelihood not in ["ignored",
                                  "Marginal prediction",
                                  "Medium likely prediction",
                                  "Highly likely prediction"]:
                stop_err("ERROR: Problem with line: %r" % line)
            out_handle.write("%s\t%s\t%s\t%s\n" % (identifier, position, score, likelihood))
    return queries
    
working_dir, bin = get_path_and_binary()

if not os.path.isfile(fasta_file):
   stop_err("ERROR: Missing input FASTA file %r" % fasta_file)

#Note that if the input FASTA file contains no sequences,
#split_fasta returns an empty list (i.e. zero temp files).
#We deliberately omit the FASTA descriptions to avoid a
#bug in promoter2 with descriptions over 200 characters.

        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

    data_handle = open(temp)
    count = make_tabular(data_handle, out_handle)
    data_handle.close()
    if not count:
        clean_up(fasta_files + temp_files)
        stop_err("No output from promoter2")
    queries += count
out_handle.close()

clean_up(fasta_files + temp_files)
print "Results for %i queries" % queries

26:20139cb4c844

"""
import sys
import os
import commands
import tempfile
from seq_analysis_utils import sys_exit, split_fasta, run_jobs, thread_count

FASTA_CHUNK = 500

if len(sys.argv) != 4:
    sys_exit("Require three arguments, number of threads (int), input DNA FASTA file & output tabular file. "
             "Got %i arguments." % (len(sys.argv)-1))

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

def get_path_and_binary():
    platform = commands.getoutput("uname") #e.g. Linux
    shell_script = commands.getoutput("which promoter")
    if not os.path.isfile(shell_script):
        sys_exit("ERROR: Missing promoter executable shell script")
    path = None
    for line in open(shell_script):
        if line.startswith("setenv"): #could then be tab or space!
            parts = line.rstrip().split(None, 2)
            if parts[0] == "setenv" and parts[1] == "PROM":
                path = parts[2]
    if not path:
        sys_exit("ERROR: Could not find promoter path (PROM) in %r" % shell_script)
    if not os.path.isdir(path):
        sys_exit("ERROR: %r is not a directory" % path)
    bin = "%s/bin/promoter_%s" % (path, platform)
    if not os.path.isfile(bin):
        sys_exit("ERROR: Missing promoter binary %r" % bin)
    return path, bin

def make_tabular(raw_handle, out_handle):
    """Parse text output into tabular, return query count."""
    identifier = None

            try:
                position, score, likelihood = line.strip().split(None,2)
            except ValueError:
                print "WARNING: Problem with line: %r" % line
                continue
                #sys_exit("ERROR: Problem with line: %r" % line)
            if likelihood not in ["ignored",
                                  "Marginal prediction",
                                  "Medium likely prediction",
                                  "Highly likely prediction"]:
                sys_exit("ERROR: Problem with line: %r" % line)
            out_handle.write("%s\t%s\t%s\t%s\n" % (identifier, position, score, likelihood))
    return queries
    
working_dir, bin = get_path_and_binary()

if not os.path.isfile(fasta_file):
   sys_exit("ERROR: Missing input FASTA file %r" % fasta_file)

#Note that if the input FASTA file contains no sequences,
#split_fasta returns an empty list (i.e. zero temp files).
#We deliberately omit the FASTA descriptions to avoid a
#bug in promoter2 with descriptions over 200 characters.

        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

    data_handle = open(temp)
    count = make_tabular(data_handle, out_handle)
    data_handle.close()
    if not count:
        clean_up(fasta_files + temp_files)
        sys_exit("No output from promoter2")
    queries += count
out_handle.close()

clean_up(fasta_files + temp_files)
print "Results for %i queries" % queries