mzML support?

[michabirklbauer]

• Currently the streamlit implementation only supports MGF files as spectra input
• Theoretically fragannot and fraggraph support input of mzML
  • theoretically because not everything works depending on which file format is used, e.g. I had to fix some of the code in fraggraph to support MGF
• Another point to consider: file extensions might not always be available on the server, since files are cached in-memory and not necessarily on disk with a name - I did my best to also cache filenames, however this is not guaranteed to be available
• Question is, is it worth implementing mzML support, considering all the overhead it's gonna create?

[michabirklbauer]

Could implement a common spectrum datastructure with things that we need

[caetera]

Doesn't pyteomics provide a relatively unified interface to both MGF and mzML (and other common formats)? At least to the spectral data. I mean the overhead should not be very large.

Regarding the extension problem, one can relatively easily detect major formats by reading, say, the first 100 bytes of the file itself - i.e. mzML file should start with <?xml version="1.0" encoding="utf-8"?><(indexed)mzML, MGF should have BEGIN IONS somewhere close to the top, etc. Alternatively, it is possible to chain-feed the file to all parsers, and the first one that won't complain is the right one. Finally, the format can be submitted by the user, although it adds an extra "burden" for him/her.