|
0
|
1 from xml.sax import make_parser, ContentHandler
|
|
|
2 from optparse import OptionParser
|
|
|
3
|
|
|
4
|
|
|
5 def main():
|
|
|
6 (options, _) = _parse_args()
|
|
|
7 with open(options.output, "w") as out:
|
|
|
8 parser = make_parser()
|
|
|
9 handler = _get_handler(options, out)
|
|
|
10 parser.setContentHandler(handler)
|
|
|
11 parser.parse(open(options.input, "r"))
|
|
|
12
|
|
|
13
|
|
|
14 def _get_handler(option, out):
|
|
|
15 return PeptideHandler(out)
|
|
|
16
|
|
|
17
|
|
|
18 class PeptideHandler(ContentHandler):
|
|
|
19 record_values = {
|
|
|
20 "IdentificationRun": ["search_engine"],
|
|
|
21 "PeptideIdentification": ["score_type", "significance_threshold", "MZ", "RT"],
|
|
|
22 "PeptideHit": ["score", "sequence", "charge"],
|
|
|
23 }
|
|
|
24
|
|
|
25 def __init__(self, output):
|
|
|
26 self.output = output
|
|
|
27
|
|
|
28 def __record_values(self, keys, attrs):
|
|
|
29 for key in keys:
|
|
|
30 setattr(self, key, attrs.get(key, None))
|
|
|
31
|
|
|
32 def startElement(self, name, attrs):
|
|
|
33 self._set_attributes(name, attrs)
|
|
|
34
|
|
|
35 def endElement(self, name):
|
|
|
36 if name == "PeptideHit":
|
|
|
37 self._write_peptide()
|
|
|
38 # reset values for element
|
|
|
39 self._set_attributes(name, {})
|
|
|
40
|
|
|
41 def _write_peptide(self):
|
|
|
42 col_keys = ["score", "peptide", "score_type", "charge", "MZ", "RT"]
|
|
|
43 row_values = self._get_values(col_keys)
|
|
|
44 row = "\t".join(row_values)
|
|
|
45 self._write_line(row)
|
|
|
46
|
|
|
47 def _write_line(self, line):
|
|
|
48 self.output.write(line)
|
|
|
49 self.output.write("\n")
|
|
|
50
|
|
|
51 def _get_values(self, keys):
|
|
|
52 return [getattr(self, key, "") for key in keys]
|
|
|
53
|
|
|
54 def _set_attributes(self, name, attrs):
|
|
|
55 for element_name, element_attributes in self.record_values.iteritems():
|
|
|
56 if name == element_name:
|
|
|
57 self.__record_values(element_attributes, attrs)
|
|
|
58
|
|
|
59
|
|
|
60 def _parse_args():
|
|
|
61 parser = OptionParser()
|
|
|
62 parser.add_option("--input", dest="input")
|
|
|
63 parser.add_option("--output", dest="output")
|
|
|
64 parser.add_option("--type", dest="type", choices=["peptide"])
|
|
|
65 return parser.parse_args()
|
|
|
66
|
|
|
67 if __name__ == "__main__":
|
|
|
68 main()
|
