Allow searching for library compounds without mass spectrum

[niekdejonge]

@jdsmith145 Josh Smith (from Tomas Pluscal lab) mentioned that it would be nice to find good matches from a list of already known metabolites for the analyzed species. These compounds often won't have a mass spectrum available, but will be available in structural libraries.

What we could implement in MS2Query is a ranking system for compounds without spectra. I would implement this, by taking the MS2Query approach where we search for closely related library structures and calculating the MS2Deepscore with all of them. The only difference is that the similarity with spectra of the original structure cannot be included.

This would make it possible to do biologically informed searches, using public structural libraries. Might be interesting to implement for MS2Query 2.0.

Josh might be interested in providing us with some case studies for this approach.

[niekdejonge]

So this would be a bit of a targeted analogue search approach. I think quite some people might be interested in something like this.

[justinjjvanderhooft]

Interesting approach indeed! We would need some thresholds on structural similarity based on case studies to see what is workable, but definitely worth a try!

[florian-huber]

I am not sure I understand what the approach would be.
Do you mean, searching for analogues with the additional constraint that they should be close to a list of known structures?

[niekdejonge]

Close to that. Instead of predicting an analogue from the entire library, just a ranking/prediction is made for the provided list of structures. But this prediction will be based on the MS2Deepscore scores with closely related structures in the library.