Integrate with matchms

[niekdejonge]

The following things could be used by matchms:

• Loaders for different file formats
• Spectrum objects

Optional:

• Create a mass2motif object in the style of matchms Spectrum objects. E.g. reusing the fragments objects.

[joewandy]

Loader seems a bit tricky to replace. If you look at ms2lda loaders, you can see that there are some bits of processing there, e.g. normalisation, removing duplicates etc. I'm not sure if matchms loaders do the same? Also the metadata returned are not the same, and there are codes downstream that rely on the metadata.

Spectrum objects is a good idea. Should be relatively easy to do. We can still use ms2lda loaders but return matchms' Spectrum. Right now there isn't any class to store MS2 spectra, it looks like it's just a list of tuples? Then if the resulting mass2motif are also stored as matchms Spectrum, it increases compatibility with various tools. What do you think @niekdejonge ?

[niekdejonge]

Might be tricky indeed, since everywhere the spectra data is used, we have to change the code to understand the new format.

Matchms has a lot of preprocessing options already implemented. Including normalisation and filtering out spectra based on intensities. However, I am not sure about the duplicate removal.
Matchms Spectrum objects also store the metadata, so I expect that everything is stored there, but in a different format which might give some issues. But this would indeed require some reformatting later.
I think that if we go for the matchms Spectrum objects, we should also use the loaders and add the preprocessing steps of matchms as well. Otherwise we have to write custom code to convert our loaded information into Spectrum objects, which might result in compatibility issues with future versions of matchms.

But I think it is a good plan to try to use matchms, if I have the time I will look into this at a later stage.