[Question] Will `spec2vec`, `ms2deepscore` be included?

[Adafede]

Although requiring additional (eventually optional) dependencies, is the implementation of these similarities planned in the future?

This would bring something totally missing in the R-ecosystem, and not "only" flavors of dot products or cosine?

[jorainer]

That's a great question indeed. It was on the original plan to actually add R code wrapping these tools directly into SpectriPy - but then, also after discussions during the hackathon we opted to add example tutorials (quarto documents) that show how this can be done using R and Python code within a quarto file, instead of just R code that behind the scene calls Python code. At present we focused more on improving performance and efficiency of the translation of MS data between R and Python data structures. Adding the support directly to SpectriPy would also increase the dependencies of the package considerably - and the need to have code to manage and load the pre-trained models.

Summarizing: the short-term plan is to add a tutorial (maybe to Metabonaut) that shows how spec2vec and ms2deepscore can be used with SpectriPy. Then, we might also consider wrapping that code within R functions for a better R-user experience. Also, contributions are highly welcome - in case you want to give it a shot :) .

[jorainer]

Note: we mention these two algorithms now also in the perspective section in PR #71

[Adafede]

Thank you for mentioning it in #71, I'll let it open for now, will maybe motivate someone to give it a try! 😉