Mercurial > repos > peterjc > mira4_assembler
diff tools/mira4_0/mira4_convert.py @ 4:1713289d9908 draft default tip
v0.0.11 tweak for use with bioconda dependencies
author | peterjc |
---|---|
date | Thu, 10 Aug 2017 11:09:10 -0400 |
parents | 4eb32a3d67d1 |
children |
line wrap: on
line diff
--- a/tools/mira4_0/mira4_convert.py Fri Oct 02 06:12:23 2015 -0400 +++ b/tools/mira4_0/mira4_convert.py Thu Aug 10 11:09:10 2017 -0400 @@ -3,44 +3,46 @@ This focuses on the miraconvert binary. """ + +from __future__ import print_function + import os -import sys -import subprocess import shutil -import time -import tempfile +import subprocess +import sys + from optparse import OptionParser + try: from io import BytesIO except ImportError: - #Should we worry about Python 2.5 or older? + # Should we worry about Python 2.5 or older? from StringIO import StringIO as BytesIO -#Do we need any PYTHONPATH magic? +# Do we need any PYTHONPATH magic? from mira4_make_bam import depad -WRAPPER_VER = "0.0.7" # Keep in sync with the XML file +WRAPPER_VER = "0.0.11" # Keep in sync with the XML file -def sys_exit(msg, err=1): - sys.stderr.write(msg+"\n") - sys.exit(err) def run(cmd): - #Avoid using shell=True when we call subprocess to ensure if the Python - #script is killed, so too is the child process. + # Avoid using shell=True when we call subprocess to ensure if the Python + # script is killed, so too is the child process. try: - child = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE) - except Exception, err: - sys_exit("Error invoking command:\n%s\n\n%s\n" % (" ".join(cmd), err)) - #Use .communicate as can get deadlocks with .wait(), + child = subprocess.Popen(cmd, universal_newlines=True, + stdout=subprocess.PIPE, stderr=subprocess.PIPE) + except Exception as err: + sys.exit("Error invoking command:\n%s\n\n%s\n" % (" ".join(cmd), err)) + # Use .communicate as can get deadlocks with .wait(), stdout, stderr = child.communicate() return_code = child.returncode if return_code: cmd_str = " ".join(cmd) # doesn't quote spaces etc if stderr and stdout: - sys_exit("Return code %i from command:\n%s\n\n%s\n\n%s" % (return_code, cmd_str, stdout, stderr)) + sys.exit("Return code %i from command:\n%s\n\n%s\n\n%s" % (return_code, cmd_str, stdout, stderr)) else: - sys_exit("Return code %i from command:\n%s\n%s" % (return_code, cmd_str, stderr)) + sys.exit("Return code %i from command:\n%s\n%s" % (return_code, cmd_str, stderr)) + def get_version(mira_binary): """Run MIRA to find its version number""" @@ -51,14 +53,15 @@ child = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.STDOUT) - except Exception, err: + 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 + +# Parse Command Line usage = """Galaxy MIRA4 wrapper script v%s - use as follows: $ python mira4_convert.py ... @@ -98,7 +101,7 @@ help="Show version and quit") options, args = parser.parse_args() if args: - sys_exit("Expected options (e.g. --input example.maf), not arguments") + sys.exit("Expected options (e.g. --input example.maf), not arguments") input_maf = options.input out_maf = options.maf @@ -107,47 +110,50 @@ out_ace = options.ace out_cstats = options.cstats -try: +if "MIRA4" in os.environ: mira_path = os.environ["MIRA4"] -except KeyError: - sys_exit("Environment variable $MIRA4 not set") -mira_convert = os.path.join(mira_path, "miraconvert") -if not os.path.isfile(mira_convert): - sys_exit("Missing miraconvert under $MIRA4, %r\nFolder contained: %s" - % (mira_convert, ", ".join(os.listdir(mira_path)))) + mira_convert = os.path.join(mira_path, "miraconvert") + if not os.path.isfile(mira_convert): + sys.exit("Missing miraconvert under $MIRA4, %r\nFolder contained: %s" + % (mira_convert, ", ".join(os.listdir(mira_path)))) +else: + sys.stderr.write("DEBUG: Since $MIRA4 is not set, assuming mira binaries are on $PATH.\n") + mira_path = None + mira_convert = "miraconvert" 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_convert_ver, mira_convert)) + sys.exit("This wrapper is for MIRA V4.0, not:\n%s\n%s" % (mira_convert_ver, mira_convert)) if options.version: print("%s, MIRA wrapper version %s" % (mira_convert_ver, WRAPPER_VER)) sys.exit(0) if not input_maf: - sys_exit("Input MIRA file is required") + sys.exit("Input MIRA file is required") elif not os.path.isfile(input_maf): - sys_exit("Missing input MIRA file: %r" % input_maf) + sys.exit("Missing input MIRA file: %r" % input_maf) if not (out_maf or out_bam or out_fasta or out_ace or out_cstats): - sys_exit("No output requested") + sys.exit("No output requested") def check_min_int(value, name): try: i = int(value) - except: - sys_exit("Bad %s setting, %r" % (name, value)) + except ValueError: + sys.exit("Bad %s setting, %r" % (name, value)) if i < 0: - sys_exit("Negative %s setting, %r" % (name, value)) + sys.exit("Negative %s setting, %r" % (name, value)) return i + min_length = check_min_int(options.min_length, "minimum length") min_cover = check_min_int(options.min_cover, "minimum cover") min_reads = check_min_int(options.min_reads, "minimum reads") -#TODO - Run MIRA in /tmp or a configurable directory? -#Currently Galaxy puts us somewhere safe like: -#/opt/galaxy-dist/database/job_working_directory/846/ +# 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 = "." @@ -164,7 +170,7 @@ if out_bam: cmd_list.append("samnbb") if not out_fasta: - #Need this for samtools depad + # Need this for samtools depad out_fasta = os.path.join(temp, "depadded.fasta") if out_fasta: cmd_list.append("fasta") @@ -174,28 +180,30 @@ cmd_list.append("cstats") run(cmd_list) + def collect(old, new): if not os.path.isfile(old): - sys_exit("Missing expected output file %s" % old) + sys.exit("Missing expected output file %s" % old) shutil.move(old, new) + if out_maf: collect(os.path.join(temp, "converted.maf"), out_maf) if out_fasta: - #Can we look at the MAF file to see if there are multiple strains? + # Can we look at the MAF file to see if there are multiple strains? old = os.path.join(temp, "converted_AllStrains.unpadded.fasta") if os.path.isfile(old): collect(old, out_fasta) else: - #Might the output be filtered down to zero contigs? + # Might the output be filtered down to zero contigs? old = os.path.join(temp, "converted.fasta") if not os.path.isfile(old): - sys_exit("Missing expected output FASTA file") + sys.exit("Missing expected output FASTA file") elif os.path.getsize(old) == 0: print("Warning - no contigs (harsh filters?)") collect(old, out_fasta) else: - sys_exit("Missing expected output FASTA file (only generic file present)") + sys.exit("Missing expected output FASTA file (only generic file present)") if out_ace: collect(os.path.join(temp, "converted.maf"), out_ace) if out_cstats: @@ -207,7 +215,7 @@ if not os.path.isfile(old): old = os.path.join(temp, "converted.sam") if not os.path.isfile(old): - sys_exit("Missing expected intermediate file %s" % old) + sys.exit("Missing expected intermediate file %s" % old) h = BytesIO() msg = depad(out_fasta, old, out_bam, h) if msg: @@ -217,7 +225,7 @@ sys.exit(1) h.close() if out_fasta == os.path.join(temp, "depadded.fasta"): - #Not asked for by Galaxy, no longer needed + # Not asked for by Galaxy, no longer needed os.remove(out_fasta) if min_length or min_cover or min_reads: