Run MUMmer on every pair of files from a list and plot the result. Fixes a bug in their gnuplot script.

mummer_all_vs_all.py

import argparse
import itertools
import os
import subprocess

"""
Run MUMmer on every pair of files from the list of input files.
"""

# Default path to the MUMmer binary
# Get MUMmer at https://sourceforge.net/projects/mummer/
MUMMER_PATH = '/usr/local/bin/MUMmer3.23/'


def _run_in_subprocess(args):
    res = subprocess.run(args, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
    if res.stdout:
        print(res.stdout.decode('ascii'))
    if res.stderr:
        print(res.stderr.decode('ascii'))


def run_mummer(mummer_path, file1, file2):
    """Runs MUMmer (actually the dnadiff script) on the specified pair of files."""
    prefix = '_'.join((os.path.splitext(os.path.basename(filename))[0] for filename in (file1, file2)))
    dnadiff_binary = mummer_path + 'nucmer'
    args = [dnadiff_binary, '--maxmatch', '-p', prefix, file1, file2]
    _run_in_subprocess(args)
    return prefix


def fix_gp_file(filename):
    """
    Due to a bug in MUMmerplot, there are invalid mouse directives inside the output gp file -
    the file used to create the plot with gnuplot.
    The simplest solution is to remove those lines from the file.
    """
    with open(filename) as gpfile:
        lines = gpfile.readlines()

    with open(filename, 'w') as gpfile:
        for line in itertools.filterfalse(lambda line: line.startswith('set mouse'), lines):
            gpfile.write(line)


def run_mummer_plot(mummer_path, output_prefix):
    """Runs MUMmerplot on the delta file with the given output prefix."""
    plot_binary = mummer_path + 'mummerplot'
    delta_file = output_prefix + '.delta'
    plot_args = [plot_binary, '--postscript', '-p', output_prefix, delta_file]
    _run_in_subprocess(plot_args)
    # A bug in MUMmerplot (it's old software) adds mouse directives that gnuplot can't process
    gp_file = output_prefix + '.gp'
    fix_gp_file(gp_file)
    # Now run gnuplot
    _run_in_subprocess(['gnuplot', gp_file])
    # Remove redundant files
    for suffix in ('.fplot', '.rplot'):
        if os.path.isfile(output_prefix + suffix):
            os.remove(output_prefix + suffix)


def parse_arguments():
    parser = argparse.ArgumentParser()
    parser.add_argument('input_files', nargs='+', help='A list of input file names')
    parser.add_argument('--mummer-path', default=MUMMER_PATH,
                        help='Path to the MUMmer binary executable. (default: %(default)s)')
    parser.add_argument('--plot', action='store_true', help='Plot the results.')
    args = parser.parse_args()
    return args


if __name__ == '__main__':
    args = parse_arguments()
    print('{} input files'.format(len(args.input_files)))
    for pair in itertools.combinations(args.input_files, 2):
        print('MUM''ing {}, {}...'.format(*pair))
        prefix = run_mummer(args.mummer_path, *pair)
        if args.plot:
            print('Plotting...')
            run_mummer_plot(args.mummer_path, prefix)