Bug with building offset_dict

[michaelmarty]

Describe the bug

With a particular mzML file that someone shared with me, the offset_dict is not built correctly the first time. When I run:

msrun = pymzml.run.Reader(test)
msrun[0]

it gives "KeyError: 0". But, if I run:

msrun = pymzml.run.Reader(test)
msrun[1]
msrun[0]

all is good. It seems to only happen when calling msrun[0]. Other data that is not zero indexed is fine. The file was converted from MGF and is uploaded here.

exportMGF_10spectra.zip

Any idea how to fix this?

To Reproduce
See above.

Desktop (please complete the following information):

• OS: Windows
• Version: 2.5.11

[fu]

Hi Michael, this is an odd behaviour. What works straight away is run["index-0"] - hope I can find more time to figure out how to fix that in the best way.

[michaelmarty]

Hi Christian,
Thanks for the tip! Unfortunately, not every mzML encoder includes "index=n" in the ID column. For example, I'm running into an interesting issue with a Sciex converted file (not mine otherwise I'd share it). It's giving correct index values, but all the ID values are 1. I think the issue is that the ID line looks like: "sample=1 period=1 cycle=1963 experiment=1" for each, and it's parsing that to 1. It works if I use "cycle=1963" to retrieve it, but is there a way I can fix the ID? I tried to write over the spec.ID, and it won't let me. Thanks!
MTM

[StSchulze]

Hi Michael,
I've had similar issues with Sciex converted files in the past. You should be able to use spec.id_dict to return a dictionary that contains "sample", "period", "cylce" and "experiment" as keys. That way, you should be able to iterate over the whole run and extract the ID the way you need it.

[michaelmarty]

Thanks! That's really helpful, and I'm starting to understand how things are coded. Where is spec.ID set? I'm not familiar with the pymzML layout, but I might try to add in some parsing to handle issues where there are duplicate ID values if I get the time.

[MKoesters]

Hi Michael,

since your file has an index:

<indexList count="2">
    <index name="spectrum">
      <offset idRef="index=0">2635</offset>
      <offset idRef="index=1">33750</offset>
      <offset idRef="index=2">67212</offset>
      <offset idRef="index=3">102299</offset>
      <offset idRef="index=4">136047</offset>
      <offset idRef="index=5">167541</offset>
      <offset idRef="index=6">202969</offset>
      <offset idRef="index=7">239570</offset>
      <offset idRef="index=8">276334</offset>
      <offset idRef="index=9">315555</offset>
    </index>

you could use something like

msrun = pymzml.run.Reader(test, index_regex=re.compile(b'"index=(?P<ID>\d+)">(?P<offset>\d+)'))
msrun[0]
msrun[1]

You can use whatever regex you want, but it needs the capturing groups "ID" and "offset"

While testing this, I realized there is a bug, so in order for this to work we need to merge #390

If you are looking where spec.ID is set, that is done in the spectrum class:

pymzML/pymzml/spec.py

Lines 815 to 835 in 7128af1

	@property
	def ID(self):
	"""
	Access the native id (last number in the id attribute) of the spectrum.
	Returns:
	ID (str): native ID of the spectrum
	"""
	if self._ID is None:
	if self.element:
	match = regex_patterns.SPECTRUM_ID_PATTERN.search(
	self.element.get("id", None)
	)
	if match:
	try:
	self._ID = int(match.group(1))
	except ValueError:
	self._ID = match.group(1)
	else:
	self._ID = self.element.get("id")
	return self._ID