format_mzdata_as_mgf.py

import xml.etree.ElementTree as et
import base64
import array


def parse_file(filename_or_obj):
    tree = et.parse(filename_or_obj)
    for spectrum in tree.findall('.//spectrum'):
        mz_elm = spectrum.find('mzArrayBinary').find('data')
        intensity_elm = spectrum.find('intenArrayBinary').find('data')

        precursor_mz = float(spectrum.find('.//precursorList/precursor/ionSelection/cvParam[@name="selected ion m/z"]').attrib['value'])
        precursor_charge = float(spectrum.find('.//precursorList/precursor/ionSelection/cvParam[@name="charge state"]').attrib['value'])

        mz = array.array('d', base64.b64decode(mz_elm.text))
        intensity = array.array('d', base64.b64decode(intensity_elm.text))
        yield precursor_mz, precursor_charge, mz, intensity

if __name__ == '__main__':
  import sys
  
  for precursor_mass, precursor_charge, mz, intensity in parse_file(sys.argv[1]):
      print "BEGIN IONS"
      print "PEPMASS={}".format(precursor_mass)
      print "CHARGE={}".format(precursor_charge)
      for x, y in zip(mz, intensity):
          print x, y
      print "END IONS"