> the search for small molecule drugs building on this observation seems to have been fruitless

I imagine this is such a perfect use case for large-scale data analytics. Maybe with the next (or n+2 or n+10) generation of wearables we may be able to track some of the macro effects of diet, exercise, air quality, sleep, etc. at such a big scale that we can determine how individual factors can affect mood and behaviour.

Ignoring the privacy and political concerns aside, it seems like such a cool problem space.