Strips all positive polarity scans from an mzXML file.

mz_strip.py

#!/usr/bin/env python3
import sys, os, argparse, re, glob
import xml.etree.ElementTree as ET

# interpret CLI arguments
parser = argparse.ArgumentParser(description='Strips all positive polarity scans from an mzXML file.')
parser.add_argument('INPUT', nargs='+', help='all files to be processed (wildcards allowed, e.g. *.mzXML)')
parser.add_argument('-n', '--negative', help='strip negative polarity scans instead', action='store_true')
parser.add_argument('-v', '--verbose', help='display more information', action='store_true')

args = parser.parse_args()

for f in args.INPUT:
    if args.verbose: print('Reading {}...'.format(f))
    # parse input file
    try:
        xml = ET.parse(f)
    except OSError:
        print('Error: could not open {}'.format(f))
        raise
    except ET.ParseError as msg:
        print('Error: could not parse {} as XML.'.format(f))
        if args.verbose: print('Details: {}'.format(msg))
        continue
    # extract data:
    root = xml.getroot()
    # find the namespace (changes for different versions):
    m = re.search('\{(.*)\}', root.tag)
    ns = {'x': m.group(1) if m else ''}
    # get started on the run:
    msRun = root.find('x:msRun',ns)
    if not args.negative:
        for scan in msRun.findall("x:scan[@polarity='+']",ns):
            msRun.remove(scan)
    else:
        for scan in msRun.findall("x:scan[@polarity='-']",ns):
            msRun.remove(scan)
    # write back out:
    try:
        path =  '{}_stripped.mzXML'.format(os.path.basename(f)[:-6])
        ET.register_namespace('',ns['x'])
        ET.ElementTree(root).write(path)
    except OSError:
        print('Error: could not create output file {}'.format(path))
        raise