comparison tools/mira4_0/mira4.py @ 32:56b421d59805 draft

planemo upload for repository https://github.com/peterjc/galaxy_mira/tree/master/tools/mira4_0 commit fd979d17340cde155de176604744831d9597c6b6

31:fd95aaef8818

#!/usr/bin/env python
"""A simple wrapper script to call MIRA and collect its output.
"""
import os
import sys
import subprocess
import shutil
import time
import tempfile
from optparse import OptionParser

#Do we need any PYTHONPATH magic?
from mira4_make_bam import make_bam

WRAPPER_VER = "0.0.4" #Keep in sync with the XML file


def get_version(mira_binary):
    """Run MIRA to find its version number"""
    # At the commend line I would use: mira -v | head -n 1
    # however there is some pipe error when doing that here.
    cmd = [mira_binary, "-v"]
    try:
        child = subprocess.Popen(cmd,
                                 stdout=subprocess.PIPE,
                                 stderr=subprocess.STDOUT)
    except Exception, err:
        sys.stderr.write("Error invoking command:\n%s\n\n%s\n" % (" ".join(cmd), err))
        sys.exit(1)
    ver, tmp = child.communicate()
    del child
    return ver.split("\n", 1)[0].strip()

#Parse Command Line
usage = """Galaxy MIRA4 wrapper script v%s - use as follows:

$ python mira4.py ...

This will run the MIRA binary and collect its output files as directed.

    sys.exit("This wrapper is for MIRA V4.0, not:\n%s\n%s" % (mira_ver, mira_binary))
mira_convert_ver = get_version(mira_convert)
if not mira_convert_ver.strip().startswith("4.0"):
    sys.exit("This wrapper is for MIRA V4.0, not:\n%s\n%s" % (mira_ver, mira_convert))
if options.version:
    print "%s, MIRA wrapper version %s" % (mira_ver, WRAPPER_VER)
    if mira_ver != mira_convert_ver:
        print "WARNING: miraconvert %s" % mira_convert_ver
    sys.exit(0)

if not manifest:
    sys.exit("Manifest is required")
elif not os.path.isfile(manifest):

    """
    handle = open(manifest, "r")
    text = handle.read()
    handle.close()

    #At time of writing, this is at the end of a file,
    #but could be followed by a space in future...
    text = text.replace("-DI:trt=/tmp", "-DI:trt=" + tempfile.gettempdir())

    #Want to try to ensure this gets written to disk before MIRA attempts
    #to open it - any networked file system may impose a delay...
    handle = open(manifest, "w")
    handle.write(text)
    handle.flush()
    os.fsync(handle.fileno())
    handle.close()


def log_manifest(manifest):
    """Write the manifest file to stderr."""
    sys.stderr.write("\n%s\nManifest file\n%s\n" % ("="*60, "="*60))
    with open(manifest) as h:
        for line in h:
            sys.stderr.write(line)
    sys.stderr.write("\n%s\nEnd of manifest\n%s\n" % ("="*60, "="*60))


def collect_output(temp, name, handle):
    """Moves files to the output filenames (global variables)."""
    f = "%s/%s_assembly/%s_d_results" % (temp, name, name)

        sys.exit("Empty output folder")
    missing = []

    old_maf = "%s/%s_out.maf" % (f, name)
    if not os.path.isfile(old_maf):
        #Triggered extractLargeContigs.sh?
        old_maf = "%s/%s_LargeContigs_out.maf" % (f, name)

    #De novo or single strain mapping,
    old_fasta = "%s/%s_out.unpadded.fasta" % (f, name)
    ref_fasta = "%s/%s_out.padded.fasta" % (f, name)
    if not os.path.isfile(old_fasta):
        #Mapping (StrainX versus reference) or de novo
        old_fasta = "%s/%s_out_StrainX.unpadded.fasta" % (f, name)
        ref_fasta = "%s/%s_out_StrainX.padded.fasta" % (f, name)
    if not os.path.isfile(old_fasta):
        old_fasta = "%s/%s_out_ReferenceStrain.unpadded.fasta" % (f, name)
        ref_fasta = "%s/%s_out_ReferenceStrain.padded.fasta" % (f, name)
        

    missing = False
    for old, new in [(old_maf, out_maf),
                     (old_fasta, out_fasta)]:
        if not os.path.isfile(old):

        log_manifest(manifest)
        sys.stderr.write("Contents of %r:\n" % f)
        for filename in sorted(os.listdir(f)):
            sys.stderr.write("%s\n" % filename)

    #For mapping mode, probably most people would expect a BAM file
    #using the reference FASTA file...
    if out_bam and out_bam != "-":
        if out_maf and out_maf != "-":
            msg = make_bam(mira_convert, out_maf, ref_fasta, out_bam, handle)
        else:
            #Not collecting the MAF file, use original location        
            msg = make_bam(mira_convert, old_maf, ref_fasta, out_bam, handle)
        if msg:
            sys.exit(msg)

def clean_up(temp, name):
    folder = "%s/%s_assembly" % (temp, name)
    if os.path.isdir(folder):
        shutil.rmtree(folder)

#TODO - Run MIRA in /tmp or a configurable directory?
#Currently Galaxy puts us somewhere safe like:
#/opt/galaxy-dist/database/job_working_directory/846/
temp = "."

name = "MIRA"

override_temp(manifest)

cmd_list = [mira_binary, "-t", str(threads), manifest]
cmd = " ".join(cmd_list)

assert os.path.isdir(temp)
d = "%s_assembly" % name
#This can fail on my development machine if stale folders exist
#under Galaxy's .../database/job_working_directory/ tree:
assert not os.path.isdir(d), "Path %r already exists:\n%s" % (d, os.path.abspath(d))
try:
    #Check path access
    os.mkdir(d)
except Exception, err:
    log_manifest(manifest)
    sys.stderr.write("Error making directory %s\n%s" % (d, err))
    sys.exit(1)

#print os.path.abspath(".")
#print cmd

if out_log and out_log != "-":
    handle = open(out_log, "w")
else:
    handle = open(os.devnull, "w")

del m
handle.write("\n")
handle.write("============================ Starting MIRA now ===============================\n")
handle.flush()
try:
    #Run MIRA
    child = subprocess.Popen(cmd_list,
                             stdout=handle,
                             stderr=subprocess.STDOUT)
except Exception, err:
    log_manifest(manifest)
    sys.stderr.write("Error invoking command:\n%s\n\n%s\n" % (cmd, err))
    #TODO - call clean up?
    handle.write("Error invoking command:\n%s\n\n%s\n" % (cmd, err))
    handle.close()
    sys.exit(1)
#Use .communicate as can get deadlocks with .wait(),
stdout, stderr = child.communicate()
assert not stdout and not stderr #Should be empty as sent to handle
run_time = time.time() - start_time
return_code = child.returncode
handle.write("\n")
handle.write("============================ MIRA has finished ===============================\n")
handle.write("MIRA took %0.2f hours\n" % (run_time / 3600.0))
if return_code:
    print "MIRA took %0.2f hours" % (run_time / 3600.0)
    handle.write("Return error code %i from command:\n" % return_code)
    handle.write(cmd + "\n")
    handle.close()
    clean_up(temp, name)
    log_manifest(manifest)
    sys.stderr.write("Return error code %i from command:\n" % return_code)
    sys.stderr.write(cmd + "\n")
    sys.exit(eturn_code)
handle.flush()

if os.path.isfile("MIRA_assembly/MIRA_d_results/ec.log"):
    handle.write("\n")
    handle.write("====================== Extract Large Contigs failed ==========================\n")

        handle.write(line)
    e.close()
    handle.write("============================ (end of ec.log) =================================\n")
    handle.flush()

#print "Collecting output..."
start_time = time.time()
collect_output(temp, name, handle)
collect_time = time.time() - start_time
handle.write("MIRA took %0.2f hours; collecting output %0.2f minutes\n" % (run_time / 3600.0, collect_time / 60.0))
print("MIRA took %0.2f hours; collecting output %0.2f minutes\n" % (run_time / 3600.0, collect_time / 60.0))

if os.path.isfile("MIRA_assembly/MIRA_d_results/ec.log"):
    #Treat as an error, but doing this AFTER collect_output
    sys.stderr.write("Extract Large Contigs failed\n")
    handle.write("Extract Large Contigs failed\n")
    handle.close()
    sys.exit(1)

#print "Cleaning up..."
clean_up(temp, name)

handle.write("\nDone\n")
handle.close()
print("Done")

32:56b421d59805

#!/usr/bin/env python
"""A simple wrapper script to call MIRA and collect its output.
"""

from __future__ import print_function

import os
import shutil
import subprocess
import sys
import tempfile
import time

from optparse import OptionParser

# Do we need any PYTHONPATH magic?
from mira4_make_bam import make_bam

WRAPPER_VER = "0.0.10"  # Keep in sync with the XML file


def get_version(mira_binary):
    """Run MIRA to find its version number."""
    # At the commend line I would use: mira -v | head -n 1
    # however there is some pipe error when doing that here.
    cmd = [mira_binary, "-v"]
    try:
        child = subprocess.Popen(cmd,
                                 stdout=subprocess.PIPE,
                                 stderr=subprocess.STDOUT)
    except Exception as err:
        sys.stderr.write("Error invoking command:\n%s\n\n%s\n" % (" ".join(cmd), err))
        sys.exit(1)
    ver, tmp = child.communicate()
    del child
    return ver.split("\n", 1)[0].strip()


# Parse Command Line
usage = """Galaxy MIRA4 wrapper script v%s - use as follows:

$ python mira4.py ...

This will run the MIRA binary and collect its output files as directed.

    sys.exit("This wrapper is for MIRA V4.0, not:\n%s\n%s" % (mira_ver, mira_binary))
mira_convert_ver = get_version(mira_convert)
if not mira_convert_ver.strip().startswith("4.0"):
    sys.exit("This wrapper is for MIRA V4.0, not:\n%s\n%s" % (mira_ver, mira_convert))
if options.version:
    print("%s, MIRA wrapper version %s" % (mira_ver, WRAPPER_VER))
    if mira_ver != mira_convert_ver:
        print("WARNING: miraconvert %s" % mira_convert_ver)
    sys.exit(0)

if not manifest:
    sys.exit("Manifest is required")
elif not os.path.isfile(manifest):

    """
    handle = open(manifest, "r")
    text = handle.read()
    handle.close()

    # At time of writing, this is at the end of a file,
    # but could be followed by a space in future...
    text = text.replace("-DI:trt=/tmp", "-DI:trt=" + tempfile.gettempdir())

    # Want to try to ensure this gets written to disk before MIRA attempts
    # to open it - any networked file system may impose a delay...
    handle = open(manifest, "w")
    handle.write(text)
    handle.flush()
    os.fsync(handle.fileno())
    handle.close()


def log_manifest(manifest):
    """Write the manifest file to stderr."""
    sys.stderr.write("\n%s\nManifest file\n%s\n" % ("=" * 60, "=" * 60))
    with open(manifest) as h:
        for line in h:
            sys.stderr.write(line)
    sys.stderr.write("\n%s\nEnd of manifest\n%s\n" % ("=" * 60, "=" * 60))


def collect_output(temp, name, handle):
    """Moves files to the output filenames (global variables)."""
    f = "%s/%s_assembly/%s_d_results" % (temp, name, name)

        sys.exit("Empty output folder")
    missing = []

    old_maf = "%s/%s_out.maf" % (f, name)
    if not os.path.isfile(old_maf):
        # Triggered extractLargeContigs.sh?
        old_maf = "%s/%s_LargeContigs_out.maf" % (f, name)

    # De novo or single strain mapping,
    old_fasta = "%s/%s_out.unpadded.fasta" % (f, name)
    ref_fasta = "%s/%s_out.padded.fasta" % (f, name)
    if not os.path.isfile(old_fasta):
        # Mapping (StrainX versus reference) or de novo
        old_fasta = "%s/%s_out_StrainX.unpadded.fasta" % (f, name)
        ref_fasta = "%s/%s_out_StrainX.padded.fasta" % (f, name)
    if not os.path.isfile(old_fasta):
        old_fasta = "%s/%s_out_ReferenceStrain.unpadded.fasta" % (f, name)
        ref_fasta = "%s/%s_out_ReferenceStrain.padded.fasta" % (f, name)

    missing = False
    for old, new in [(old_maf, out_maf),
                     (old_fasta, out_fasta)]:
        if not os.path.isfile(old):

        log_manifest(manifest)
        sys.stderr.write("Contents of %r:\n" % f)
        for filename in sorted(os.listdir(f)):
            sys.stderr.write("%s\n" % filename)

    # For mapping mode, probably most people would expect a BAM file
    # using the reference FASTA file...
    if out_bam and out_bam != "-":
        if out_maf and out_maf != "-":
            msg = make_bam(mira_convert, out_maf, ref_fasta, out_bam, handle)
        else:
            # Not collecting the MAF file, use original location
            msg = make_bam(mira_convert, old_maf, ref_fasta, out_bam, handle)
        if msg:
            sys.exit(msg)


def clean_up(temp, name):
    folder = "%s/%s_assembly" % (temp, name)
    if os.path.isdir(folder):
        shutil.rmtree(folder)


# TODO - Run MIRA in /tmp or a configurable directory?
# Currently Galaxy puts us somewhere safe like:
# /opt/galaxy-dist/database/job_working_directory/846/
temp = "."

name = "MIRA"

override_temp(manifest)

cmd_list = [mira_binary, "-t", str(threads), manifest]
cmd = " ".join(cmd_list)

assert os.path.isdir(temp)
d = "%s_assembly" % name
# This can fail on my development machine if stale folders exist
# under Galaxy's .../database/job_working_directory/ tree:
assert not os.path.isdir(d), "Path %r already exists:\n%s" % (d, os.path.abspath(d))
try:
    # Check path access
    os.mkdir(d)
except Exception as err:
    log_manifest(manifest)
    sys.stderr.write("Error making directory %s\n%s" % (d, err))
    sys.exit(1)

# print(os.path.abspath("."))
# print(cmd)

if out_log and out_log != "-":
    handle = open(out_log, "w")
else:
    handle = open(os.devnull, "w")

del m
handle.write("\n")
handle.write("============================ Starting MIRA now ===============================\n")
handle.flush()
try:
    # Run MIRA
    child = subprocess.Popen(cmd_list,
                             stdout=handle,
                             stderr=subprocess.STDOUT)
except Exception as err:
    log_manifest(manifest)
    sys.stderr.write("Error invoking command:\n%s\n\n%s\n" % (cmd, err))
    # TODO - call clean up?
    handle.write("Error invoking command:\n%s\n\n%s\n" % (cmd, err))
    handle.close()
    sys.exit(1)
# Use .communicate as can get deadlocks with .wait(),
stdout, stderr = child.communicate()
assert not stdout and not stderr  # Should be empty as sent to handle
run_time = time.time() - start_time
return_code = child.returncode
handle.write("\n")
handle.write("============================ MIRA has finished ===============================\n")
handle.write("MIRA took %0.2f hours\n" % (run_time / 3600.0))
if return_code:
    print("MIRA took %0.2f hours" % (run_time / 3600.0))
    handle.write("Return error code %i from command:\n" % return_code)
    handle.write(cmd + "\n")
    handle.close()
    clean_up(temp, name)
    log_manifest(manifest)
    sys.stderr.write("Return error code %i from command:\n" % return_code)
    sys.stderr.write(cmd + "\n")
    sys.exit(return_code)
handle.flush()

if os.path.isfile("MIRA_assembly/MIRA_d_results/ec.log"):
    handle.write("\n")
    handle.write("====================== Extract Large Contigs failed ==========================\n")

        handle.write(line)
    e.close()
    handle.write("============================ (end of ec.log) =================================\n")
    handle.flush()

# print("Collecting output...")
start_time = time.time()
collect_output(temp, name, handle)
collect_time = time.time() - start_time
handle.write("MIRA took %0.2f hours; collecting output %0.2f minutes\n"
             % (run_time / 3600.0, collect_time / 60.0))
print("MIRA took %0.2f hours; collecting output %0.2f minutes\n"
      % (run_time / 3600.0, collect_time / 60.0))

if os.path.isfile("MIRA_assembly/MIRA_d_results/ec.log"):
    # Treat as an error, but doing this AFTER collect_output
    sys.stderr.write("Extract Large Contigs failed\n")
    handle.write("Extract Large Contigs failed\n")
    handle.close()
    sys.exit(1)

# print "Cleaning up..."
clean_up(temp, name)

handle.write("\nDone\n")
handle.close()
print("Done")