We have been using MelonnPan on microbial sequence features collected and sequenced from mouse skin. In these samples, we have already identified some bacteria in high frequencies that are capable of producing S-equol (which is, actually, our metabolite of interest). However, after running the MelonnPan model, it did not return S-equol as a possible metabolite, which we found odd.
Is this possibly due to a limited (mouse-wise) trained model? We used the default/provided trained model of your package. If so, do you know of any open-source paired datasets which does include this metabolite and which we could use to train our model? Or, is there any already trained model for mice specifically?
Any guidance would be much appreciated! Thank you so much for your time and help!